https://heym.run/ 2026-08-18T00:00:00.000Z weekly 1 https://heym.run/features 2026-08-18T00:00:00.000Z weekly 0.9 https://heym.run/playground 2026-07-28T00:00:00.000Z weekly 0.9 https://heym.run/platform 2026-08-18T00:00:00.000Z weekly 0.88 https://heym.run/nodes 2026-07-16T00:00:00.000Z weekly 0.88 https://heym.run/solutions 2026-08-02T00:00:00.000Z weekly 0.88 https://heym.run/solutions/knowledge-systems Persistent Memory and Second Brain for AI Agents https://i.ytimg.com/vi/bvUqTVu5gUc/maxresdefault.jpg In this video, we look at Heym’s Agent Memory Graph, a second brain for AI agents that need persistent memory over time. We use an orchestrator and sub-agent example to show how agents can keep, inspect, and share memory across workflows. With persistent memory enabled, each agent can store long-term context. Heym visualizes this memory as a graph, where teams can search entities, inspect relationships, and understand how different pieces of information are connected. This is useful for long-running agents, multi-agent systems, and workflows that need context over time. Instead of starting from zero in every run, agents can learn, remember, and reuse knowledge. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIAgents #AgentMemory #SecondBrain #AgenticAI #WorkflowAutomation #GraphMemory https://www.youtube-nocookie.com/embed/bvUqTVu5gUc 2026-08-02T00:00:00.000Z monthly 0.8 https://heym.run/solutions/human-reviewed-operations AI Drafts Your Support Replies. You Decide What Gets Sent. Email Agent | Heym https://i.ytimg.com/vi/OyrorEWCM5g/maxresdefault.jpg An AI agent can draft every reply in a busy support inbox. What matters is who decides what actually reaches the customer. In this 90-second demo, a Heym workflow reads a new customer email, writes a reply and then pauses. Nothing is sent until a person reviews it. You can approve the draft as it is, fix a sentence before approving, or reject it so the customer never sees it. What you will see: • A new email arrives and the HITL Support Reply workflow picks it up • The agent drafts a reply and requests a human review, so the run waits • The reviewer gets a notification with a link to a review page showing a summary and the full draft • Accept sends the draft as written • Edit & Continue sends it with your changes • Refuse stops it, and the agent records that no reply was sent Chapters 0:00 A full inbox and a new email 0:16 Heym reads the email and drafts a reply 0:28 The workflow waits for you 0:46 Fix one sentence, then approve 0:58 Reject it and nothing is sent 1:12 Nothing goes out until you approve Heym is a source-available, self-hostable platform for building AI workflows on a visual canvas, with agents, RAG, MCP and human-in-the-loop review built in. Try the template: https://heym.run/templates/hitl-support-reply-agent Human-in-the-loop docs: https://heym.run/docs/reference/human-in-the-loop Website: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #HumanInTheLoop #CustomerSupport https://www.youtube-nocookie.com/embed/OyrorEWCM5g 2026-08-02T00:00:00.000Z monthly 0.8 https://heym.run/solutions/multi-agent-research Persistent Memory and Second Brain for AI Agents https://i.ytimg.com/vi/bvUqTVu5gUc/maxresdefault.jpg In this video, we look at Heym’s Agent Memory Graph, a second brain for AI agents that need persistent memory over time. We use an orchestrator and sub-agent example to show how agents can keep, inspect, and share memory across workflows. With persistent memory enabled, each agent can store long-term context. Heym visualizes this memory as a graph, where teams can search entities, inspect relationships, and understand how different pieces of information are connected. This is useful for long-running agents, multi-agent systems, and workflows that need context over time. Instead of starting from zero in every run, agents can learn, remember, and reuse knowledge. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIAgents #AgentMemory #SecondBrain #AgenticAI #WorkflowAutomation #GraphMemory https://www.youtube-nocookie.com/embed/bvUqTVu5gUc 2026-08-02T00:00:00.000Z monthly 0.8 https://heym.run/solutions/browser-automation Browser Automation That Heals Itself | Heym https://i.ytimg.com/vi/z-KB_UZCfvM/maxresdefault.jpg Your browser automation breaks the moment a website ships a redesign. Heym fixes it while the run is still going: auto heal finds a sturdier locator, retries the step and saves the fix for every run after. In 90 seconds, an MCP client calls a Heym workflow that drives a real browser with Playwright, hands the page data to an AI agent, survives a portal redesign mid run, and restores the login session from saved cookies so the next run finishes in seconds. What you will see • Triggering a workflow from any MCP client and watching it run live on the canvas • Response control with the Output node, plus execution logs for every node • Screenshots captured at every step of the browser run • The Playwright code generated from visual steps, and the hardened sandbox for custom code • Execution history, and bringing a past run back to the canvas • Agent runs with every tool call and decision on record • Auto heal repairing a broken selector at run time • Cookie restore that skips the login entirely How auto heal works When a selector fails twice, Heym sends the page HTML, a screenshot and the failed step to your model, gets back a text or role based locator such as role=button[name='Export CSV'], and retries. The working step is stored with your saved steps, so the next run needs no LLM call at all. How cookie restore works Keep a Playwright storageState or a cookie list in a global variable. The Playwright node restores the session before the first step, checks an authenticated selector, and only runs your login fallback steps when the session has actually expired. Chapters 0:00 Run a workflow from an MCP client 0:15 Response control and execution logs 0:25 Screenshots at every step 0:35 The Playwright code behind your steps 0:45 Execution history and replaying a run 0:55 Agent runs, tool calls and decisions 1:05 Auto heal fixes a broken selector 1:16 Cookie restore for faster runs Try Heym: https://heym.run Browser automation with Heym: https://heym.run/solutions/browser-automation Playwright node docs: https://heym.run/docs/nodes/playwright-node GitHub: https://github.com/heymrun/heym The supplier portal in this video is a demo site built for the walkthrough. #BrowserAutomation #Playwright #AIAgents #MCP #Heym https://www.youtube-nocookie.com/embed/z-KB_UZCfvM 2026-08-04T00:00:00.000Z monthly 0.8 https://heym.run/solutions/team-skills Build and Run Skills in Heym https://i.ytimg.com/vi/eIStbSewKcA/maxresdefault.jpg In this video, we look at Heym Skills. Heym can run skills natively inside your workflows. Skills can include Python code, assets, files, execution permissions, timeout settings, and Drive access. In this demo, we use a translation skill that takes a text input, translates it into German, and creates PDF and DOCX output files with the Heym logo. The video covers: • Running skills inside Heym workflows • Using a template with ready-to-use skills • Translating text and generating PDF and DOCX files • Managing skill files, Python code, SVG assets, and fonts • Setting timeout and Drive permissions • Downloading generated files from the Heym interface • Iterating on a skill with AI • Editing skill configuration • Viewing skill history • Seeing what changed, who changed it, and when • Reverting to previous skill versions With Heym Skills, teams can create reusable capabilities, run them inside workflows, manage files, iterate with AI, and keep a full change history. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AgenticAI #AIWorkflowAutomation #Skills #Python #DevTools https://www.youtube-nocookie.com/embed/eIStbSewKcA 2026-08-04T00:00:00.000Z monthly 0.8 https://heym.run/solutions/operational-dashboards Heym Dashboards: Build Interactive Dashboards with AI https://i.ytimg.com/vi/UAtFnCXmIQw/maxresdefault.jpg In this video, we walk through the Dashboard feature in Heym. Heym Dashboards let you create different dashboard experiences for workflows, teams, and business use cases. You can add visual elements such as logos, hyperlinks, images, and text, then organize them using the Tidy Up option. Each dashboard can also be explored within a selected time range, helping users focus on a specific period and inspect the related data in more detail. You can create new widgets using the Heym AI Assistant by simply describing what you need. After the widget is generated, you can fine-tune it to improve the layout, content, and data presentation. For more control, you can open the related canvas and edit the underlying workflow or widget logic directly. Heym Dashboards are not static reporting pages. They are connected to the workflows behind them. Create dashboards visually. Generate widgets with AI. Fine-tune the result. Edit the underlying logic. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We welcome contributions, feature ideas, workflow templates, documentation improvements, and pull requests. #AI #Automation #Workflow #Agents #Heym #Dashboards https://www.youtube-nocookie.com/embed/UAtFnCXmIQw 2026-08-04T00:00:00.000Z monthly 0.8 https://heym.run/solutions/ai-observability Tracing AI Workflows in Heym https://i.ytimg.com/vi/RyMdJDihzbo/maxresdefault.jpg In this video, we walk through the tracing experience in Heym using a simple travel workflow. One agent finds activities to do in the city. The other agent finds local food ideas and restaurant recommendations. From the canvas, you can inspect each agent node, open and collapse execution details, and use the timing view to understand how long each node took. When you select a node from the timing view, Heym opens the related trace. In the Traces tab, you can inspect the full LLM call behind that node, including the request, response, timing details, and pricing information. For teams building production AI workflows, this visibility matters. You need to know what ran, how long it took, what was sent to the model, what came back, and how much the model call cost. Heym Traces help you debug, monitor, and improve AI workflows from one visual workspace. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, documentation improvements, templates, and pull requests. #Heym #AIWorkflowAutomation #AIAgents #LLMObservability #WorkflowAutomation #AITracing #LLM #Automation #SelfHosted #DeveloperTools https://www.youtube-nocookie.com/embed/RyMdJDihzbo 2026-08-04T00:00:00.000Z monthly 0.8 https://heym.run/solutions/embedded-ai-backends Heym MCP Server: Expose Workflows as Tools for Claude, Cursor and OpenAI https://i.ytimg.com/vi/rsLKKzMXsAI/maxresdefault.jpg In this video, we walk through Heym’s built-in MCP server feature. Heym can expose existing workflows as MCP tools, so they can be called directly from MCP-compatible clients such as Claude, Cursor, and other AI tools. In the demo, we first connect a Heym workflow to Claude using a custom connector. Claude can then call the workflow through MCP, Heym executes it in the background, and the result comes back directly inside Claude. Then we show how to connect multiple Heym workflows to Cursor. The selected workflows appear as tools in Cursor, and each tool can call the related workflow inside Heym. This makes Heym more than a workflow builder. It becomes an execution layer for your AI toolchain. You can build workflows visually in Heym, expose them as MCP tools, and run them from the clients your team already uses. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, workflow templates, documentation improvements, and pull requests. #ai #automation #workflow #agents #heym #mcpserver #mcpclients #mcpworkflows #mcptools https://www.youtube-nocookie.com/embed/rsLKKzMXsAI 2026-08-04T00:00:00.000Z monthly 0.8 https://heym.run/solutions/ai-evaluation Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ 2026-08-04T00:00:00.000Z monthly 0.8 https://heym.run/solutions/workflow-monitoring Turn Workflow Failures into Automated Actions with Heym Alerts https://i.ytimg.com/vi/YjL1t3mog1Y/maxresdefault.jpg In this video, we look at Alerts in Heym. Heym Alerts let teams define alert conditions, backtest them before enabling, review alert history, and connect alerts directly to workflows. In this demo, we create an alert that checks all workflows with five errors in the last three minutes. Heym uses AI to help generate the alert configuration, so you do not need to write long queries manually. The video covers: • Creating a new alert in Heym • Using AI to generate alert configuration • Selecting model and credentials • Monitoring all workflows or a single workflow • Setting error thresholds and time windows • Backtesting alerts before enabling them • Understanding how often an alert would have fired • Choosing alert actions • Creating and running a workflow from an alert • Passing alert data into an existing workflow • Creating notifications • Reducing noisy repeated alerts • Reviewing alert history • Pausing, sharing, renaming, and deleting alerts With Heym Alerts, teams can monitor workflow failures, reduce alert noise, and turn important events into automated, reviewable actions. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #Alerts #AIWorkflowAutomation #AgenticAI #WorkflowAutomation #DevTools https://www.youtube-nocookie.com/embed/YjL1t3mog1Y 2026-08-09T00:00:00.000Z monthly 0.8 https://heym.run/docs 2026-09-23T00:00:00.000Z weekly 0.85 https://heym.run/enterprise 2026-08-02T00:00:00.000Z monthly 0.8 https://heym.run/convert 2026-06-13T00:00:00.000Z monthly 0.8 https://heym.run/blog 2026-09-27T00:00:00.000Z weekly 0.9 https://heym.run/changelog 2026-09-27T00:00:00.000Z weekly 0.8 https://heym.run/templates 2026-07-16T00:00:00.000Z weekly 0.85 https://heym.run/beginner-templates Build Your First AI Agent in One Sentence | Heym https://i.ytimg.com/vi/wp9TY_OQUKw/maxresdefault.jpg Describe the job in one sentence and Heym's AI Assistant lays out the workflow for you. In this video, a single sentence becomes a working AI agent that turns a messy note into a clean bullet summary. What happens in the video: 1. Start with an empty canvas. 2. Type one sentence: "Summarize this note into bullets." 3. The assistant adds three steps: YourMessage, ChatAgent and AgentReply. 4. Paste a messy note and press Run. 5. A clean summary comes back. The workflow is the Build Your First AI Agent template. Open it, connect an OpenAI-compatible credential to ChatAgent, and change the system instruction to fit your own job. Already have Brave Search API or Website Loader workflows? Both are Heym templates. Describe the workflow you want in Chat, and it can run them to search the web and read pages before it builds. Chat uses the workflows you already have, so add them first. Chapters 0:00 An empty canvas and one sentence 0:18 The assistant lays out the steps 0:38 Paste a messy note and run it 0:56 A clean summary comes back 1:08 Web research from Chat Links Template: https://heym.run/templates/first-ai-agent Brave Search API template: https://heym.run/templates/brave-search-api Website Loader template: https://heym.run/templates/jina-web-fetcher Heym: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #WorkflowAutomation #Heym https://www.youtube-nocookie.com/embed/wp9TY_OQUKw 2026-07-24T00:00:00.000Z weekly 0.82 https://heym.run/about 2026-04-08T00:00:00.000Z monthly 0.7 https://heym.run/agents 2026-05-31T00:00:00.000Z monthly 0.7 https://heym.run/editorial-policy 2026-09-02T00:00:00.000Z monthly 0.6 https://heym.run/privacy-policy 2026-03-29T00:00:00.000Z monthly 0.6 https://heym.run/terms-of-service 2026-03-29T00:00:00.000Z monthly 0.6 https://heym.run/impressum 2026-03-29T00:00:00.000Z monthly 0.5 https://heym.run/ai-agent-runtime 2026-09-13T00:00:00.000Z monthly 0.9 https://heym.run/self-hosted-ai-agents 2026-09-13T00:00:00.000Z monthly 0.9 https://heym.run/ai-agent-governance 2026-09-13T00:00:00.000Z monthly 0.9 https://heym.run/mcp-agent-runtime 2026-09-13T00:00:00.000Z monthly 0.9 https://heym.run/migrate 2026-09-13T00:00:00.000Z monthly 0.8 https://heym.run/migrate/n8n-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/make-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/zapier-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/dify-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/flowise-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/langflow-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/activepieces-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/pipedream-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/node-red-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/windmill-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/sim-ai-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/migrate/huginn-to-heym 2026-09-13T00:00:00.000Z monthly 0.7 https://heym.run/docs/getting-started/core-concepts 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/getting-started/introduction 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/getting-started/migrate-to-heym 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/getting-started/quick-start 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/getting-started/running-and-deployment 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/getting-started/why-heym 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/agent-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/amazon-s3-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/bigquery-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/chart-output-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/clickhouse-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/code-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/codex-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/condition-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/console-log-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/converter-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/crawler-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/cron-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/datatable-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/decision-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/disable-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/discord-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/discord-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/drive-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/error-handler-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/execute-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/file-upload-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/github-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/google-drive-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/google-sheets-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/grist-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/heym-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/heym-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/html-output-mapper-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/http-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/imap-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/input-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/jira-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/json-output-mapper-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/linear-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/llm-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/loop-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/mcp-call-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/merge-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/notion-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/opencode-go-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/output-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/playwright-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/plugin-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/plugin-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/rabbitmq-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/rag-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/redis-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/send-email-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/sentry-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/set-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/slack-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/slack-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/sticky-note-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/supabase-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/switch-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/telegram-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/telegram-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/throw-error-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/variable-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/wait-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/websocket-send-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/nodes/websocket-trigger-node 2026-09-23T00:00:00.000Z monthly 0.72 https://heym.run/docs/tabs/alerts-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/analytics-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/board-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/chat-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/credentials-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/dashboard-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/datatable-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/drive-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/evals-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/logs-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/mcp-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/scheduled-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/teams-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/templates-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/traces-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/global-variables-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/vectorstores-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/tabs/workflows-tab 2026-09-23T00:00:00.000Z monthly 0.62 https://heym.run/docs/reference/agent-architecture 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/agent-persistent-memory 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/ai-assistant 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/audit 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/canvas-features 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/chat-voice 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/chat-with-heym 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/contextual-showcase 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/credentials 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/credentials-sharing 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/download-import 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/drive 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/edit-history 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/enterprise 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/environment-variables 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/execution-history 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/execution-tokens 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/expression-dsl 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/expression-evaluation-dialog 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/file-generation 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/global-variables 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/guardrails 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/features 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/human-in-the-loop 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/keyboard-shortcuts 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/cluster 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/node-types 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/opentelemetry 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/parallel-execution 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/plugin-authoring 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/portal 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/quick-drawer 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/security 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/user-settings 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/sso 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/sse-streaming 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/teams 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/integrations 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/triggers 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/webhooks 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/workflow-analysis 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/workflow-organization 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/docs/reference/workflow-structure 2026-09-23T00:00:00.000Z monthly 0.66 https://heym.run/blog/ai-agent-templates Build Your First AI Agent in One Sentence | Heym https://i.ytimg.com/vi/wp9TY_OQUKw/maxresdefault.jpg Describe the job in one sentence and Heym's AI Assistant lays out the workflow for you. In this video, a single sentence becomes a working AI agent that turns a messy note into a clean bullet summary. What happens in the video: 1. Start with an empty canvas. 2. Type one sentence: "Summarize this note into bullets." 3. The assistant adds three steps: YourMessage, ChatAgent and AgentReply. 4. Paste a messy note and press Run. 5. A clean summary comes back. The workflow is the Build Your First AI Agent template. Open it, connect an OpenAI-compatible credential to ChatAgent, and change the system instruction to fit your own job. Already have Brave Search API or Website Loader workflows? Both are Heym templates. Describe the workflow you want in Chat, and it can run them to search the web and read pages before it builds. Chat uses the workflows you already have, so add them first. Chapters 0:00 An empty canvas and one sentence 0:18 The assistant lays out the steps 0:38 Paste a messy note and run it 0:56 A clean summary comes back 1:08 Web research from Chat Links Template: https://heym.run/templates/first-ai-agent Brave Search API template: https://heym.run/templates/brave-search-api Website Loader template: https://heym.run/templates/jina-web-fetcher Heym: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #WorkflowAutomation #Heym https://www.youtube-nocookie.com/embed/wp9TY_OQUKw Ask Your Company Files a Question, Get an Answer You Can Check - Self Hosted | Heym https://i.ytimg.com/vi/8Et2k-z56kc/maxresdefault.jpg Your team asks the same questions every week. How many remote days do we get? What is the travel allowance? The answers are already in your policy documents, but nobody wants to dig through a PDF to find them. This 90-second video shows the Chat with Drive Docs workflow in Heym. You upload your company PDFs to Drive and ask questions in a chat. The agent finds the right file, reads it, and answers with the exact sentence it used, so anyone can check where the answer came from. When the files do not cover a question, it says so. Ask whether you can bring your dog to the office and it searches all four policies, finds nothing, and tells you it does not know instead of making something up. How the workflow works: • userQuery takes the question from the Chat Portal • docAgent lists your Drive files with the listDriveFiles tool • readDriveFile fetches the PDF, and the pdf_base64_to_text skill extracts the text • answer sends the reply back to the chat The template is free to use. Open it, choose a model credential, upload your PDFs to Drive, and start asking: https://heym.run/templates/drive-chat-with-docs Chapters 0:00 Put the company files in Drive 0:12 Ask a question 0:32 The answer, with the line from the file 0:48 A question the files do not answer 1:00 It says it does not know 1:12 From the file, not from the chat Heym is an AI workflow automation platform with a visual editor: https://heym.run The company, policies and files in this video are fictional. #aiagents #documentai #workflowautomation https://www.youtube-nocookie.com/embed/8Et2k-z56kc Extract Invoice Data from Any PDF Layout, Automatically | Heym https://i.ytimg.com/vi/MeBoKY9HgMA/maxresdefault.jpg Invoices arrive as PDFs, and someone usually copies the numbers into a spreadsheet by hand. This 90-second video shows a Heym workflow doing that job instead. Drop in a PDF invoice and an AI agent reads it, pulls out the vendor, invoice number, date, total and line items, and writes them as a clean row in a table. When the next invoice uses a completely different layout, the same columns fill in, with dates and amounts normalized to one format. How the workflow works: • InvoiceURL takes the link to the PDF • InvoiceExtractor, an AI agent, sends the PDF to LlamaParse and reads the result • It returns structured JSON: vendor_name, invoice_number, invoice_date, due_date, subtotal, tax, total, currency and line_items • InvoiceData passes that JSON on to a DataTable, QuickBooks, Xero or your next step The template is free. Open it, add your LlamaParse key and a model credential, and run it on your own invoices: https://heym.run/templates/invoice-pdf-extractor Chapters 0:00 An invoice arrives as a PDF 0:10 Drop it in, the fields land in a table 0:34 Another invoice, a different layout 0:46 The same fields fill in 0:58 Nobody copies the numbers 1:10 The invoice arrives, the row is ready Heym is an AI workflow automation platform with a visual editor: https://heym.run The invoices, vendors and amounts in this video are fictional. #invoiceautomation #aiagents #workflowautomation https://www.youtube-nocookie.com/embed/MeBoKY9HgMA AI Drafts Your Support Replies. You Decide What Gets Sent. Email Agent | Heym https://i.ytimg.com/vi/OyrorEWCM5g/maxresdefault.jpg An AI agent can draft every reply in a busy support inbox. What matters is who decides what actually reaches the customer. In this 90-second demo, a Heym workflow reads a new customer email, writes a reply and then pauses. Nothing is sent until a person reviews it. You can approve the draft as it is, fix a sentence before approving, or reject it so the customer never sees it. What you will see: • A new email arrives and the HITL Support Reply workflow picks it up • The agent drafts a reply and requests a human review, so the run waits • The reviewer gets a notification with a link to a review page showing a summary and the full draft • Accept sends the draft as written • Edit & Continue sends it with your changes • Refuse stops it, and the agent records that no reply was sent Chapters 0:00 A full inbox and a new email 0:16 Heym reads the email and drafts a reply 0:28 The workflow waits for you 0:46 Fix one sentence, then approve 0:58 Reject it and nothing is sent 1:12 Nothing goes out until you approve Heym is a source-available, self-hostable platform for building AI workflows on a visual canvas, with agents, RAG, MCP and human-in-the-loop review built in. Try the template: https://heym.run/templates/hitl-support-reply-agent Human-in-the-loop docs: https://heym.run/docs/reference/human-in-the-loop Website: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #HumanInTheLoop #CustomerSupport https://www.youtube-nocookie.com/embed/OyrorEWCM5g Browser Automation That Heals Itself | Heym https://i.ytimg.com/vi/z-KB_UZCfvM/maxresdefault.jpg Your browser automation breaks the moment a website ships a redesign. Heym fixes it while the run is still going: auto heal finds a sturdier locator, retries the step and saves the fix for every run after. In 90 seconds, an MCP client calls a Heym workflow that drives a real browser with Playwright, hands the page data to an AI agent, survives a portal redesign mid run, and restores the login session from saved cookies so the next run finishes in seconds. What you will see • Triggering a workflow from any MCP client and watching it run live on the canvas • Response control with the Output node, plus execution logs for every node • Screenshots captured at every step of the browser run • The Playwright code generated from visual steps, and the hardened sandbox for custom code • Execution history, and bringing a past run back to the canvas • Agent runs with every tool call and decision on record • Auto heal repairing a broken selector at run time • Cookie restore that skips the login entirely How auto heal works When a selector fails twice, Heym sends the page HTML, a screenshot and the failed step to your model, gets back a text or role based locator such as role=button[name='Export CSV'], and retries. The working step is stored with your saved steps, so the next run needs no LLM call at all. How cookie restore works Keep a Playwright storageState or a cookie list in a global variable. The Playwright node restores the session before the first step, checks an authenticated selector, and only runs your login fallback steps when the session has actually expired. Chapters 0:00 Run a workflow from an MCP client 0:15 Response control and execution logs 0:25 Screenshots at every step 0:35 The Playwright code behind your steps 0:45 Execution history and replaying a run 0:55 Agent runs, tool calls and decisions 1:05 Auto heal fixes a broken selector 1:16 Cookie restore for faster runs Try Heym: https://heym.run Browser automation with Heym: https://heym.run/solutions/browser-automation Playwright node docs: https://heym.run/docs/nodes/playwright-node GitHub: https://github.com/heymrun/heym The supplier portal in this video is a demo site built for the walkthrough. #BrowserAutomation #Playwright #AIAgents #MCP #Heym https://www.youtube-nocookie.com/embed/z-KB_UZCfvM Watch an AI Agent Workflow Run, Step by Step, in Heym https://i.ytimg.com/vi/f22KcSSgAyA/maxresdefault.jpg A draft scores 0.96 on an AI-writing check. We hand it to a Heym workflow and follow the run from the first node to the final response. It comes back at 0.27. The workflow is the Decision Model AI Text Humanizer. An agent rewrites the text five sentences at a time, a decision model scores each batch, and a loop keeps the best of five rounds. Everything it does along the way stays visible. In this video: • The run on the canvas, node by node • The Execution Log with status, timing and output for each node, and the final response • Inside the agent: every checkAi tool call, its score, and the rewrite it triggered • The decisions the loop made across five rounds • The execution timeline and the full LLM trace, with prompt, tool calls, tokens, cost and latency • A failing step, automatic retries, and Ask AI explaining the error • Pinned data, so the next run skips a slow step • Run history with Bring to Canvas, and edit history with diffs, preview and revert Chapters 0:00 A draft scored 0.96, run on the canvas 0:20 Execution Log and the response 0:30 Inside the agent: tool calls and decisions 0:42 Traces: timing, tokens and cost 0:52 Debugging: retries, errors and Ask AI 1:02 Pinned data and run history 1:18 Edit history Try the workflow: https://heym.run/templates/decision-model-ai-text-humanizer Heym: https://heym.run Docs Execution Log and pinned data: https://heym.run/docs/reference/canvas-features Traces: https://heym.run/docs/tabs/traces-tab Run history: https://heym.run/docs/reference/execution-history Edit history: https://heym.run/docs/reference/edit-history Source code: https://github.com/heymrun/heym #Heym #AIAgents #WorkflowAutomation https://www.youtube-nocookie.com/embed/f22KcSSgAyA Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ 2026-09-27T00:00:00.000Z weekly 0.8 https://heym.run/blog/llm-router JEV, Laya and Decision Models in Heym https://i.ytimg.com/vi/P8oEA-K3jPw/maxresdefault.jpg In this video, we look at Decision Models in Heym and how they can be used inside agentic workflows. We start with JEV and other models such as Laya as examples of the emerging decision model category. Unlike generative LLMs that primarily produce text, decision models are designed for fast, structured decisions such as choices, scores, routing, classification, and confidence checks. In Heym, these models can be used for model routing, LLM evaluation, guardrails, tool gating, inbox triage, re-ranking, large-scale labeling, real-time control, and confidence gating. The video covers: • JEV and Laya as examples of decision models • Native Decision nodes in Heym • LLM-as-a-judge workflows • Model routing across different LLMs • Choice, score, and null decision outputs • AI-generated decision logic • Default fallback models • Base URL and API key configuration • Testing decision model connections • Language-based decision workflows • Automatic model selection inside orchestrator agents • Decision model visibility in Traces • Routing simple tool calls to smaller models • Routing creative tasks to more suitable models • Reducing token usage and improving response time The main idea is simple: You do not need to use the same model for every step. Heym can use decision models to choose the right model, the right action, and the right path dynamically while the workflow is running. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #JEV #Laya #DecisionModels #ModelRouting #LLMEvaluation #AgenticAI https://www.youtube-nocookie.com/embed/P8oEA-K3jPw Importing Heym Workflow Templates in One Minute https://i.ytimg.com/vi/pfW3DQMEk-g/maxresdefault.jpg This step by step tutorial shows how to import free AI workflow templates from the Heym website into your own self hosted workspace. You will see how to open the templates page, pick a template, use Copy JSON, and paste it into the canvas so all nodes and connections appear instantly. We finish by editing node settings, adding API keys, and running the workflow so you have a complete template to production path you can repeat for any use case. #AI #AIAutomation #AIWorkflows #Workflow #Automation #NoCode #SelfHosted #OpenSource #Tutorial #HowTo #Developer #Heym #AITools #JSON #AIAgent https://www.youtube-nocookie.com/embed/pfW3DQMEk-g 2026-09-24T00:00:00.000Z weekly 0.8 https://heym.run/blog/system-one-models JEV, Laya and Decision Models in Heym https://i.ytimg.com/vi/P8oEA-K3jPw/maxresdefault.jpg In this video, we look at Decision Models in Heym and how they can be used inside agentic workflows. We start with JEV and other models such as Laya as examples of the emerging decision model category. Unlike generative LLMs that primarily produce text, decision models are designed for fast, structured decisions such as choices, scores, routing, classification, and confidence checks. In Heym, these models can be used for model routing, LLM evaluation, guardrails, tool gating, inbox triage, re-ranking, large-scale labeling, real-time control, and confidence gating. The video covers: • JEV and Laya as examples of decision models • Native Decision nodes in Heym • LLM-as-a-judge workflows • Model routing across different LLMs • Choice, score, and null decision outputs • AI-generated decision logic • Default fallback models • Base URL and API key configuration • Testing decision model connections • Language-based decision workflows • Automatic model selection inside orchestrator agents • Decision model visibility in Traces • Routing simple tool calls to smaller models • Routing creative tasks to more suitable models • Reducing token usage and improving response time The main idea is simple: You do not need to use the same model for every step. Heym can use decision models to choose the right model, the right action, and the right path dynamically while the workflow is running. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #JEV #Laya #DecisionModels #ModelRouting #LLMEvaluation #AgenticAI https://www.youtube-nocookie.com/embed/P8oEA-K3jPw 2026-09-21T00:00:00.000Z weekly 0.8 https://heym.run/blog/self-hosted-ai-agents 2026-09-20T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-spending-limits 2026-09-12T00:00:00.000Z weekly 0.8 https://heym.run/blog/a2a-vs-mcp Heym MCP Server: Expose Workflows as Tools for Claude, Cursor and OpenAI https://i.ytimg.com/vi/rsLKKzMXsAI/maxresdefault.jpg In this video, we walk through Heym’s built-in MCP server feature. Heym can expose existing workflows as MCP tools, so they can be called directly from MCP-compatible clients such as Claude, Cursor, and other AI tools. In the demo, we first connect a Heym workflow to Claude using a custom connector. Claude can then call the workflow through MCP, Heym executes it in the background, and the result comes back directly inside Claude. Then we show how to connect multiple Heym workflows to Cursor. The selected workflows appear as tools in Cursor, and each tool can call the related workflow inside Heym. This makes Heym more than a workflow builder. It becomes an execution layer for your AI toolchain. You can build workflows visually in Heym, expose them as MCP tools, and run them from the clients your team already uses. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, workflow templates, documentation improvements, and pull requests. #ai #automation #workflow #agents #heym #mcpserver #mcpclients #mcpworkflows #mcptools https://www.youtube-nocookie.com/embed/rsLKKzMXsAI 2026-09-09T00:00:00.000Z weekly 0.8 https://heym.run/blog/llm-as-a-judge Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ 2026-09-02T00:00:00.000Z weekly 0.8 https://heym.run/blog/enterprise-ai-agents Set Up Heym Locally in Under 2 Minutes https://i.ytimg.com/vi/P6YvlupUboU/maxresdefault.jpg In this video, we show how to get Heym running locally in about two minutes. Heym is an agentic AI workflow automation platform that can be self-hosted on infrastructure you control. After cloning the GitHub repository, you can use run.sh for local development or deploy.sh for running Heym as a managed single-instance Docker environment. The video covers: • Cloning the Heym repository • Starting a PostgreSQL database • Reviewing the minimal setup • Configuring the required environment variables • Enabling registration for local testing • Running Heym locally • Creating an account on your local instance • Starting to build agentic workflows on your own machine Heym only needs one main dependency to start: PostgreSQL. Watch the video to see how to run Heym locally and start building agentic workflows on infrastructure you control. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #SelfHosted #AgenticAI #AIWorkflowAutomation #Docker #PostgreSQL #DevTools https://www.youtube-nocookie.com/embed/P6YvlupUboU 2026-09-25T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-governance 2026-08-25T00:00:00.000Z weekly 0.8 https://heym.run/blog/spec-driven-development 2026-08-21T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-sandbox Build and Run Skills in Heym https://i.ytimg.com/vi/eIStbSewKcA/maxresdefault.jpg In this video, we look at Heym Skills. Heym can run skills natively inside your workflows. Skills can include Python code, assets, files, execution permissions, timeout settings, and Drive access. In this demo, we use a translation skill that takes a text input, translates it into German, and creates PDF and DOCX output files with the Heym logo. The video covers: • Running skills inside Heym workflows • Using a template with ready-to-use skills • Translating text and generating PDF and DOCX files • Managing skill files, Python code, SVG assets, and fonts • Setting timeout and Drive permissions • Downloading generated files from the Heym interface • Iterating on a skill with AI • Editing skill configuration • Viewing skill history • Seeing what changed, who changed it, and when • Reverting to previous skill versions With Heym Skills, teams can create reusable capabilities, run them inside workflows, manage files, iterate with AI, and keep a full change history. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AgenticAI #AIWorkflowAutomation #Skills #Python #DevTools https://www.youtube-nocookie.com/embed/eIStbSewKcA 2026-08-16T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-web-scraping Browser Automation in Heym with Playwright https://i.ytimg.com/vi/4wAKE_x4LTo/maxresdefault.jpg In this video, we look at browser automation in Heym. Heym makes it easy to build Playwright automations directly inside your agentic workflows. In this demo, we show how to open a website, scroll the page, take a screenshot, read website text, and extract data from the Heym Templates page. The video covers: • Running browser automation inside a Heym workflow • Using URL and browser steps as workflow inputs • Taking screenshots with Playwright • Viewing browser automation output in execution logs • Letting AI decide browser actions from user-provided steps • Logging and reusing steps to avoid repeated AI requests • Auto-healing when selectors or inputs change • Running workflows from Heym Chat • Building a Playwright automation from an empty canvas • Reading selected page elements from the UI • Extracting the current number of Heym templates With Heym browser automation, you can open pages, perform actions, extract data, take screenshots, reuse steps, and combine Playwright with AI-powered workflows. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #BrowserAutomation #Playwright #AIWorkflowAutomation #AgenticAI #DevTools https://www.youtube-nocookie.com/embed/4wAKE_x4LTo 2026-08-14T00:00:00.000Z weekly 0.8 https://heym.run/blog/llm-cost-alerts Turn Workflow Failures into Automated Actions with Heym Alerts https://i.ytimg.com/vi/YjL1t3mog1Y/maxresdefault.jpg In this video, we look at Alerts in Heym. Heym Alerts let teams define alert conditions, backtest them before enabling, review alert history, and connect alerts directly to workflows. In this demo, we create an alert that checks all workflows with five errors in the last three minutes. Heym uses AI to help generate the alert configuration, so you do not need to write long queries manually. The video covers: • Creating a new alert in Heym • Using AI to generate alert configuration • Selecting model and credentials • Monitoring all workflows or a single workflow • Setting error thresholds and time windows • Backtesting alerts before enabling them • Understanding how often an alert would have fired • Choosing alert actions • Creating and running a workflow from an alert • Passing alert data into an existing workflow • Creating notifications • Reducing noisy repeated alerts • Reviewing alert history • Pausing, sharing, renaming, and deleting alerts With Heym Alerts, teams can monitor workflow failures, reduce alert noise, and turn important events into automated, reviewable actions. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #Alerts #AIWorkflowAutomation #AgenticAI #WorkflowAutomation #DevTools https://www.youtube-nocookie.com/embed/YjL1t3mog1Y 2026-08-09T00:00:00.000Z weekly 0.8 https://heym.run/blog/agent-skills Build and Run Skills in Heym https://i.ytimg.com/vi/eIStbSewKcA/maxresdefault.jpg In this video, we look at Heym Skills. Heym can run skills natively inside your workflows. Skills can include Python code, assets, files, execution permissions, timeout settings, and Drive access. In this demo, we use a translation skill that takes a text input, translates it into German, and creates PDF and DOCX output files with the Heym logo. The video covers: • Running skills inside Heym workflows • Using a template with ready-to-use skills • Translating text and generating PDF and DOCX files • Managing skill files, Python code, SVG assets, and fonts • Setting timeout and Drive permissions • Downloading generated files from the Heym interface • Iterating on a skill with AI • Editing skill configuration • Viewing skill history • Seeing what changed, who changed it, and when • Reverting to previous skill versions With Heym Skills, teams can create reusable capabilities, run them inside workflows, manage files, iterate with AI, and keep a full change history. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AgenticAI #AIWorkflowAutomation #Skills #Python #DevTools https://www.youtube-nocookie.com/embed/eIStbSewKcA 2026-08-08T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-document-processing 2026-08-05T00:00:00.000Z weekly 0.8 https://heym.run/blog/best-n8n-alternatives-2026 Set Up Heym Locally in Under 2 Minutes https://i.ytimg.com/vi/P6YvlupUboU/maxresdefault.jpg In this video, we show how to get Heym running locally in about two minutes. Heym is an agentic AI workflow automation platform that can be self-hosted on infrastructure you control. After cloning the GitHub repository, you can use run.sh for local development or deploy.sh for running Heym as a managed single-instance Docker environment. The video covers: • Cloning the Heym repository • Starting a PostgreSQL database • Reviewing the minimal setup • Configuring the required environment variables • Enabling registration for local testing • Running Heym locally • Creating an account on your local instance • Starting to build agentic workflows on your own machine Heym only needs one main dependency to start: PostgreSQL. Watch the video to see how to run Heym locally and start building agentic workflows on infrastructure you control. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #SelfHosted #AgenticAI #AIWorkflowAutomation #Docker #PostgreSQL #DevTools https://www.youtube-nocookie.com/embed/P6YvlupUboU 2026-07-30T00:00:00.000Z weekly 0.8 https://heym.run/blog/google-drive-automation Ask Your Company Files a Question, Get an Answer You Can Check - Self Hosted | Heym https://i.ytimg.com/vi/8Et2k-z56kc/maxresdefault.jpg Your team asks the same questions every week. How many remote days do we get? What is the travel allowance? The answers are already in your policy documents, but nobody wants to dig through a PDF to find them. This 90-second video shows the Chat with Drive Docs workflow in Heym. You upload your company PDFs to Drive and ask questions in a chat. The agent finds the right file, reads it, and answers with the exact sentence it used, so anyone can check where the answer came from. When the files do not cover a question, it says so. Ask whether you can bring your dog to the office and it searches all four policies, finds nothing, and tells you it does not know instead of making something up. How the workflow works: • userQuery takes the question from the Chat Portal • docAgent lists your Drive files with the listDriveFiles tool • readDriveFile fetches the PDF, and the pdf_base64_to_text skill extracts the text • answer sends the reply back to the chat The template is free to use. Open it, choose a model credential, upload your PDFs to Drive, and start asking: https://heym.run/templates/drive-chat-with-docs Chapters 0:00 Put the company files in Drive 0:12 Ask a question 0:32 The answer, with the line from the file 0:48 A question the files do not answer 1:00 It says it does not know 1:12 From the file, not from the chat Heym is an AI workflow automation platform with a visual editor: https://heym.run The company, policies and files in this video are fictional. #aiagents #documentai #workflowautomation https://www.youtube-nocookie.com/embed/8Et2k-z56kc 2026-07-27T00:00:00.000Z weekly 0.8 https://heym.run/blog/human-in-the-loop-ai-agents AI Drafts Your Support Replies. You Decide What Gets Sent. Email Agent | Heym https://i.ytimg.com/vi/OyrorEWCM5g/maxresdefault.jpg An AI agent can draft every reply in a busy support inbox. What matters is who decides what actually reaches the customer. In this 90-second demo, a Heym workflow reads a new customer email, writes a reply and then pauses. Nothing is sent until a person reviews it. You can approve the draft as it is, fix a sentence before approving, or reject it so the customer never sees it. What you will see: • A new email arrives and the HITL Support Reply workflow picks it up • The agent drafts a reply and requests a human review, so the run waits • The reviewer gets a notification with a link to a review page showing a summary and the full draft • Accept sends the draft as written • Edit & Continue sends it with your changes • Refuse stops it, and the agent records that no reply was sent Chapters 0:00 A full inbox and a new email 0:16 Heym reads the email and drafts a reply 0:28 The workflow waits for you 0:46 Fix one sentence, then approve 0:58 Reject it and nothing is sent 1:12 Nothing goes out until you approve Heym is a source-available, self-hostable platform for building AI workflows on a visual canvas, with agents, RAG, MCP and human-in-the-loop review built in. Try the template: https://heym.run/templates/hitl-support-reply-agent Human-in-the-loop docs: https://heym.run/docs/reference/human-in-the-loop Website: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #HumanInTheLoop #CustomerSupport https://www.youtube-nocookie.com/embed/OyrorEWCM5g 2026-07-25T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-browser-automation Browser Automation That Heals Itself | Heym https://i.ytimg.com/vi/z-KB_UZCfvM/maxresdefault.jpg Your browser automation breaks the moment a website ships a redesign. Heym fixes it while the run is still going: auto heal finds a sturdier locator, retries the step and saves the fix for every run after. In 90 seconds, an MCP client calls a Heym workflow that drives a real browser with Playwright, hands the page data to an AI agent, survives a portal redesign mid run, and restores the login session from saved cookies so the next run finishes in seconds. What you will see • Triggering a workflow from any MCP client and watching it run live on the canvas • Response control with the Output node, plus execution logs for every node • Screenshots captured at every step of the browser run • The Playwright code generated from visual steps, and the hardened sandbox for custom code • Execution history, and bringing a past run back to the canvas • Agent runs with every tool call and decision on record • Auto heal repairing a broken selector at run time • Cookie restore that skips the login entirely How auto heal works When a selector fails twice, Heym sends the page HTML, a screenshot and the failed step to your model, gets back a text or role based locator such as role=button[name='Export CSV'], and retries. The working step is stored with your saved steps, so the next run needs no LLM call at all. How cookie restore works Keep a Playwright storageState or a cookie list in a global variable. The Playwright node restores the session before the first step, checks an authenticated selector, and only runs your login fallback steps when the session has actually expired. Chapters 0:00 Run a workflow from an MCP client 0:15 Response control and execution logs 0:25 Screenshots at every step 0:35 The Playwright code behind your steps 0:45 Execution history and replaying a run 0:55 Agent runs, tool calls and decisions 1:05 Auto heal fixes a broken selector 1:16 Cookie restore for faster runs Try Heym: https://heym.run Browser automation with Heym: https://heym.run/solutions/browser-automation Playwright node docs: https://heym.run/docs/nodes/playwright-node GitHub: https://github.com/heymrun/heym The supplier portal in this video is a demo site built for the walkthrough. #BrowserAutomation #Playwright #AIAgents #MCP #Heym https://www.youtube-nocookie.com/embed/z-KB_UZCfvM 2026-07-21T00:00:00.000Z weekly 0.8 https://heym.run/blog/background-coding-agent 2026-07-19T00:00:00.000Z weekly 0.8 https://heym.run/blog/agentic-kanban-board Agentic AI Workflow Automation with Heym Boards https://i.ytimg.com/vi/oDoTfbU4O_M/maxresdefault.jpg Heym Boards turn a task board into an agentic workflow execution system. In this demo, we create an e-commerce sales board, add campaign tasks, connect AI workflows to board stages, run a campaign brief workflow, inspect status and outputs, use comments and files as extra context, add Human Review, and complete the task with a final workflow action. Heym is a self-hosted AI workflow automation platform for agentic workflows, RAG, MCP, human-in-the-loop automation, traces, dashboards, portals, and production AI systems. Try Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIWorkflowAutomation #AIAgents #AgenticAI #WorkflowAutomation #HumanInTheLoop #SelfHostedAI #SalesAutomation https://www.youtube-nocookie.com/embed/oDoTfbU4O_M 2026-07-14T00:00:00.000Z weekly 0.8 https://heym.run/blog/best-vibe-coding-tools-2026 2026-07-11T00:00:00.000Z weekly 0.8 https://heym.run/blog/codex-ai-agent 2026-07-06T00:00:00.000Z weekly 0.8 https://heym.run/blog/prompt-injection 2026-07-02T00:00:00.000Z weekly 0.8 https://heym.run/blog/heym-second-month-changelog Heym Dashboards: Build Interactive Dashboards with AI https://i.ytimg.com/vi/UAtFnCXmIQw/maxresdefault.jpg In this video, we walk through the Dashboard feature in Heym. Heym Dashboards let you create different dashboard experiences for workflows, teams, and business use cases. You can add visual elements such as logos, hyperlinks, images, and text, then organize them using the Tidy Up option. Each dashboard can also be explored within a selected time range, helping users focus on a specific period and inspect the related data in more detail. You can create new widgets using the Heym AI Assistant by simply describing what you need. After the widget is generated, you can fine-tune it to improve the layout, content, and data presentation. For more control, you can open the related canvas and edit the underlying workflow or widget logic directly. Heym Dashboards are not static reporting pages. They are connected to the workflows behind them. Create dashboards visually. Generate widgets with AI. Fine-tune the result. Edit the underlying logic. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We welcome contributions, feature ideas, workflow templates, documentation improvements, and pull requests. #AI #Automation #Workflow #Agents #Heym #Dashboards https://www.youtube-nocookie.com/embed/UAtFnCXmIQw 2026-06-29T00:00:00.000Z weekly 0.8 https://heym.run/blog/agentic-workflows Watch an AI Agent Workflow Run, Step by Step, in Heym https://i.ytimg.com/vi/f22KcSSgAyA/maxresdefault.jpg A draft scores 0.96 on an AI-writing check. We hand it to a Heym workflow and follow the run from the first node to the final response. It comes back at 0.27. The workflow is the Decision Model AI Text Humanizer. An agent rewrites the text five sentences at a time, a decision model scores each batch, and a loop keeps the best of five rounds. Everything it does along the way stays visible. In this video: • The run on the canvas, node by node • The Execution Log with status, timing and output for each node, and the final response • Inside the agent: every checkAi tool call, its score, and the rewrite it triggered • The decisions the loop made across five rounds • The execution timeline and the full LLM trace, with prompt, tool calls, tokens, cost and latency • A failing step, automatic retries, and Ask AI explaining the error • Pinned data, so the next run skips a slow step • Run history with Bring to Canvas, and edit history with diffs, preview and revert Chapters 0:00 A draft scored 0.96, run on the canvas 0:20 Execution Log and the response 0:30 Inside the agent: tool calls and decisions 0:42 Traces: timing, tokens and cost 0:52 Debugging: retries, errors and Ask AI 1:02 Pinned data and run history 1:18 Edit history Try the workflow: https://heym.run/templates/decision-model-ai-text-humanizer Heym: https://heym.run Docs Execution Log and pinned data: https://heym.run/docs/reference/canvas-features Traces: https://heym.run/docs/tabs/traces-tab Run history: https://heym.run/docs/reference/execution-history Edit history: https://heym.run/docs/reference/edit-history Source code: https://github.com/heymrun/heym #Heym #AIAgents #WorkflowAutomation https://www.youtube-nocookie.com/embed/f22KcSSgAyA 2026-07-20T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-workflow-automation-examples Extract Invoice Data from Any PDF Layout, Automatically | Heym https://i.ytimg.com/vi/MeBoKY9HgMA/maxresdefault.jpg Invoices arrive as PDFs, and someone usually copies the numbers into a spreadsheet by hand. This 90-second video shows a Heym workflow doing that job instead. Drop in a PDF invoice and an AI agent reads it, pulls out the vendor, invoice number, date, total and line items, and writes them as a clean row in a table. When the next invoice uses a completely different layout, the same columns fill in, with dates and amounts normalized to one format. How the workflow works: • InvoiceURL takes the link to the PDF • InvoiceExtractor, an AI agent, sends the PDF to LlamaParse and reads the result • It returns structured JSON: vendor_name, invoice_number, invoice_date, due_date, subtotal, tax, total, currency and line_items • InvoiceData passes that JSON on to a DataTable, QuickBooks, Xero or your next step The template is free. Open it, add your LlamaParse key and a model credential, and run it on your own invoices: https://heym.run/templates/invoice-pdf-extractor Chapters 0:00 An invoice arrives as a PDF 0:10 Drop it in, the fields land in a table 0:34 Another invoice, a different layout 0:46 The same fields fill in 0:58 Nobody copies the numbers 1:10 The invoice arrives, the row is ready Heym is an AI workflow automation platform with a visual editor: https://heym.run The invoices, vendors and amounts in this video are fictional. #invoiceautomation #aiagents #workflowautomation https://www.youtube-nocookie.com/embed/MeBoKY9HgMA 2026-06-24T00:00:00.000Z weekly 0.8 https://heym.run/blog/llm-guardrails 2026-06-19T00:00:00.000Z weekly 0.8 https://heym.run/blog/agentic-rag Ask Your Company Files a Question, Get an Answer You Can Check - Self Hosted | Heym https://i.ytimg.com/vi/8Et2k-z56kc/maxresdefault.jpg Your team asks the same questions every week. How many remote days do we get? What is the travel allowance? The answers are already in your policy documents, but nobody wants to dig through a PDF to find them. This 90-second video shows the Chat with Drive Docs workflow in Heym. You upload your company PDFs to Drive and ask questions in a chat. The agent finds the right file, reads it, and answers with the exact sentence it used, so anyone can check where the answer came from. When the files do not cover a question, it says so. Ask whether you can bring your dog to the office and it searches all four policies, finds nothing, and tells you it does not know instead of making something up. How the workflow works: • userQuery takes the question from the Chat Portal • docAgent lists your Drive files with the listDriveFiles tool • readDriveFile fetches the PDF, and the pdf_base64_to_text skill extracts the text • answer sends the reply back to the chat The template is free to use. Open it, choose a model credential, upload your PDFs to Drive, and start asking: https://heym.run/templates/drive-chat-with-docs Chapters 0:00 Put the company files in Drive 0:12 Ask a question 0:32 The answer, with the line from the file 0:48 A question the files do not answer 1:00 It says it does not know 1:12 From the file, not from the chat Heym is an AI workflow automation platform with a visual editor: https://heym.run The company, policies and files in this video are fictional. #aiagents #documentai #workflowautomation https://www.youtube-nocookie.com/embed/8Et2k-z56kc 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/agentic-design-patterns Watch an AI Agent Workflow Run, Step by Step, in Heym https://i.ytimg.com/vi/f22KcSSgAyA/maxresdefault.jpg A draft scores 0.96 on an AI-writing check. We hand it to a Heym workflow and follow the run from the first node to the final response. It comes back at 0.27. The workflow is the Decision Model AI Text Humanizer. An agent rewrites the text five sentences at a time, a decision model scores each batch, and a loop keeps the best of five rounds. Everything it does along the way stays visible. In this video: • The run on the canvas, node by node • The Execution Log with status, timing and output for each node, and the final response • Inside the agent: every checkAi tool call, its score, and the rewrite it triggered • The decisions the loop made across five rounds • The execution timeline and the full LLM trace, with prompt, tool calls, tokens, cost and latency • A failing step, automatic retries, and Ask AI explaining the error • Pinned data, so the next run skips a slow step • Run history with Bring to Canvas, and edit history with diffs, preview and revert Chapters 0:00 A draft scored 0.96, run on the canvas 0:20 Execution Log and the response 0:30 Inside the agent: tool calls and decisions 0:42 Traces: timing, tokens and cost 0:52 Debugging: retries, errors and Ask AI 1:02 Pinned data and run history 1:18 Edit history Try the workflow: https://heym.run/templates/decision-model-ai-text-humanizer Heym: https://heym.run Docs Execution Log and pinned data: https://heym.run/docs/reference/canvas-features Traces: https://heym.run/docs/tabs/traces-tab Run history: https://heym.run/docs/reference/execution-history Edit history: https://heym.run/docs/reference/edit-history Source code: https://github.com/heymrun/heym #Heym #AIAgents #WorkflowAutomation https://www.youtube-nocookie.com/embed/f22KcSSgAyA 2026-07-03T00:00:00.000Z weekly 0.8 https://heym.run/blog/migrate-to-heym Heym AI Convert: Migrate Workflows from Other Platforms https://i.ytimg.com/vi/AgA5vg3d2VU/maxresdefault.jpg In this video, we walk through AI Convert on the Heym website. AI Convert helps you create a new workflow from a natural language description or migrate an existing automation from another workflow platform into Heym. First, we describe the workflow we want to build. For this example, we generate a workflow that finds relevant videos based on a given topic. Once the workflow is generated, we can preview the structure, copy it, or download it as a file. Then we paste it directly onto the Heym canvas, where the nodes and connections are created automatically and the workflow becomes ready to configure and run. We also show how AI Convert can detect the source platform of an existing workflow, analyze its nodes and connections, and convert the automation into a Heym-compatible workflow. This makes it easier to migrate existing automations without rebuilding every step manually. With Heym AI Convert, you can: * Describe a new workflow in natural language * Paste an existing workflow from another platform * Detect the original workflow platform * Convert nodes and connections into Heym format * Preview the generated workflow * Copy or download the result * Paste it directly onto the Heym canvas * Continue editing and running it inside Heym Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We welcome bug reports, feature ideas, workflow templates, documentation improvements, and pull requests. #AI #Automation #Heym #Workflow #Converter #Agents #WorkflowBuilder https://www.youtube-nocookie.com/embed/AgA5vg3d2VU 2026-09-13T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-customer-support-agent Importing Heym Workflow Templates in One Minute https://i.ytimg.com/vi/pfW3DQMEk-g/maxresdefault.jpg This step by step tutorial shows how to import free AI workflow templates from the Heym website into your own self hosted workspace. You will see how to open the templates page, pick a template, use Copy JSON, and paste it into the canvas so all nodes and connections appear instantly. We finish by editing node settings, adding API keys, and running the workflow so you have a complete template to production path you can repeat for any use case. #AI #AIAutomation #AIWorkflows #Workflow #Automation #NoCode #SelfHosted #OpenSource #Tutorial #HowTo #Developer #Heym #AITools #JSON #AIAgent https://www.youtube-nocookie.com/embed/pfW3DQMEk-g AI Drafts Your Support Replies. You Decide What Gets Sent. Email Agent | Heym https://i.ytimg.com/vi/OyrorEWCM5g/maxresdefault.jpg An AI agent can draft every reply in a busy support inbox. What matters is who decides what actually reaches the customer. In this 90-second demo, a Heym workflow reads a new customer email, writes a reply and then pauses. Nothing is sent until a person reviews it. You can approve the draft as it is, fix a sentence before approving, or reject it so the customer never sees it. What you will see: • A new email arrives and the HITL Support Reply workflow picks it up • The agent drafts a reply and requests a human review, so the run waits • The reviewer gets a notification with a link to a review page showing a summary and the full draft • Accept sends the draft as written • Edit & Continue sends it with your changes • Refuse stops it, and the agent records that no reply was sent Chapters 0:00 A full inbox and a new email 0:16 Heym reads the email and drafts a reply 0:28 The workflow waits for you 0:46 Fix one sentence, then approve 0:58 Reject it and nothing is sent 1:12 Nothing goes out until you approve Heym is a source-available, self-hostable platform for building AI workflows on a visual canvas, with agents, RAG, MCP and human-in-the-loop review built in. Try the template: https://heym.run/templates/hitl-support-reply-agent Human-in-the-loop docs: https://heym.run/docs/reference/human-in-the-loop Website: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #HumanInTheLoop #CustomerSupport https://www.youtube-nocookie.com/embed/OyrorEWCM5g 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/opentelemetry-ai-agents Tracing AI Workflows in Heym https://i.ytimg.com/vi/RyMdJDihzbo/maxresdefault.jpg In this video, we walk through the tracing experience in Heym using a simple travel workflow. One agent finds activities to do in the city. The other agent finds local food ideas and restaurant recommendations. From the canvas, you can inspect each agent node, open and collapse execution details, and use the timing view to understand how long each node took. When you select a node from the timing view, Heym opens the related trace. In the Traces tab, you can inspect the full LLM call behind that node, including the request, response, timing details, and pricing information. For teams building production AI workflows, this visibility matters. You need to know what ran, how long it took, what was sent to the model, what came back, and how much the model call cost. Heym Traces help you debug, monitor, and improve AI workflows from one visual workspace. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, documentation improvements, templates, and pull requests. #Heym #AIWorkflowAutomation #AIAgents #LLMObservability #WorkflowAutomation #AITracing #LLM #Automation #SelfHosted #DeveloperTools https://www.youtube-nocookie.com/embed/RyMdJDihzbo 2026-06-09T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-cost-optimization Heym Analytics: Monitor and Improve Your AI Workflows https://i.ytimg.com/vi/oWAV-_JJSKk/maxresdefault.jpg Heym Analytics helps you monitor how your AI workflows perform over time. In this video, we walk through the Analytics tab in Heym and show how you can track workflow activity from a single dashboard. You can filter analytics by all workflows or select a specific workflow, choose a time range, and review key metrics such as total executions, success rate, error rate, and average latency. The Analytics tab also helps you inspect workflow usage, identify failed executions, open error details, and understand where improvements are needed. For teams building AI automations, visibility is critical. Workflows are not just prototypes. They become production systems that need monitoring, reliability tracking, and continuous improvement. With Heym Analytics, you can build, run, monitor, and improve your automation stack in one place. Product: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #aiworkflowautomation #workflowautomation #aiagents #automation #analytics #observability #llm #selfhosted https://www.youtube-nocookie.com/embed/oWAV-_JJSKk Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-code-review 2026-09-25T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-evaluation Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ Heym Analytics: Monitor and Improve Your AI Workflows https://i.ytimg.com/vi/oWAV-_JJSKk/maxresdefault.jpg Heym Analytics helps you monitor how your AI workflows perform over time. In this video, we walk through the Analytics tab in Heym and show how you can track workflow activity from a single dashboard. You can filter analytics by all workflows or select a specific workflow, choose a time range, and review key metrics such as total executions, success rate, error rate, and average latency. The Analytics tab also helps you inspect workflow usage, identify failed executions, open error details, and understand where improvements are needed. For teams building AI automations, visibility is critical. Workflows are not just prototypes. They become production systems that need monitoring, reliability tracking, and continuous improvement. With Heym Analytics, you can build, run, monitor, and improve your automation stack in one place. Product: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #aiworkflowautomation #workflowautomation #aiagents #automation #analytics #observability #llm #selfhosted https://www.youtube-nocookie.com/embed/oWAV-_JJSKk 2026-05-31T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-observability Tracing AI Workflows in Heym https://i.ytimg.com/vi/RyMdJDihzbo/maxresdefault.jpg In this video, we walk through the tracing experience in Heym using a simple travel workflow. One agent finds activities to do in the city. The other agent finds local food ideas and restaurant recommendations. From the canvas, you can inspect each agent node, open and collapse execution details, and use the timing view to understand how long each node took. When you select a node from the timing view, Heym opens the related trace. In the Traces tab, you can inspect the full LLM call behind that node, including the request, response, timing details, and pricing information. For teams building production AI workflows, this visibility matters. You need to know what ran, how long it took, what was sent to the model, what came back, and how much the model call cost. Heym Traces help you debug, monitor, and improve AI workflows from one visual workspace. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, documentation improvements, templates, and pull requests. #Heym #AIWorkflowAutomation #AIAgents #LLMObservability #WorkflowAutomation #AITracing #LLM #Automation #SelfHosted #DeveloperTools https://www.youtube-nocookie.com/embed/RyMdJDihzbo Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ 2026-05-29T00:00:00.000Z weekly 0.8 https://heym.run/blog/heym-first-month-changelog Heym Dashboards: Build Interactive Dashboards with AI https://i.ytimg.com/vi/UAtFnCXmIQw/maxresdefault.jpg In this video, we walk through the Dashboard feature in Heym. Heym Dashboards let you create different dashboard experiences for workflows, teams, and business use cases. You can add visual elements such as logos, hyperlinks, images, and text, then organize them using the Tidy Up option. Each dashboard can also be explored within a selected time range, helping users focus on a specific period and inspect the related data in more detail. You can create new widgets using the Heym AI Assistant by simply describing what you need. After the widget is generated, you can fine-tune it to improve the layout, content, and data presentation. For more control, you can open the related canvas and edit the underlying workflow or widget logic directly. Heym Dashboards are not static reporting pages. They are connected to the workflows behind them. Create dashboards visually. Generate widgets with AI. Fine-tune the result. Edit the underlying logic. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We welcome contributions, feature ideas, workflow templates, documentation improvements, and pull requests. #AI #Automation #Workflow #Agents #Heym #Dashboards https://www.youtube-nocookie.com/embed/UAtFnCXmIQw 2026-06-29T00:00:00.000Z weekly 0.8 https://heym.run/blog/context-engineering Persistent Memory and Second Brain for AI Agents https://i.ytimg.com/vi/bvUqTVu5gUc/maxresdefault.jpg In this video, we look at Heym’s Agent Memory Graph, a second brain for AI agents that need persistent memory over time. We use an orchestrator and sub-agent example to show how agents can keep, inspect, and share memory across workflows. With persistent memory enabled, each agent can store long-term context. Heym visualizes this memory as a graph, where teams can search entities, inspect relationships, and understand how different pieces of information are connected. This is useful for long-running agents, multi-agent systems, and workflows that need context over time. Instead of starting from zero in every run, agents can learn, remember, and reuse knowledge. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIAgents #AgentMemory #SecondBrain #AgenticAI #WorkflowAutomation #GraphMemory https://www.youtube-nocookie.com/embed/bvUqTVu5gUc 2026-08-05T00:00:00.000Z weekly 0.8 https://heym.run/blog/vibe-coding Build Your First AI Agent in One Sentence | Heym https://i.ytimg.com/vi/wp9TY_OQUKw/maxresdefault.jpg Describe the job in one sentence and Heym's AI Assistant lays out the workflow for you. In this video, a single sentence becomes a working AI agent that turns a messy note into a clean bullet summary. What happens in the video: 1. Start with an empty canvas. 2. Type one sentence: "Summarize this note into bullets." 3. The assistant adds three steps: YourMessage, ChatAgent and AgentReply. 4. Paste a messy note and press Run. 5. A clean summary comes back. The workflow is the Build Your First AI Agent template. Open it, connect an OpenAI-compatible credential to ChatAgent, and change the system instruction to fit your own job. Already have Brave Search API or Website Loader workflows? Both are Heym templates. Describe the workflow you want in Chat, and it can run them to search the web and read pages before it builds. Chat uses the workflows you already have, so add them first. Chapters 0:00 An empty canvas and one sentence 0:18 The assistant lays out the steps 0:38 Paste a messy note and run it 0:56 A clean summary comes back 1:08 Web research from Chat Links Template: https://heym.run/templates/first-ai-agent Brave Search API template: https://heym.run/templates/brave-search-api Website Loader template: https://heym.run/templates/jina-web-fetcher Heym: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #WorkflowAutomation #Heym https://www.youtube-nocookie.com/embed/wp9TY_OQUKw 2026-07-11T00:00:00.000Z weekly 0.8 https://heym.run/blog/agentic-ai-vs-generative-ai 2026-06-27T00:00:00.000Z weekly 0.8 https://heym.run/blog/what-is-mcp Heym MCP Server: Expose Workflows as Tools for Claude, Cursor and OpenAI https://i.ytimg.com/vi/rsLKKzMXsAI/maxresdefault.jpg In this video, we walk through Heym’s built-in MCP server feature. Heym can expose existing workflows as MCP tools, so they can be called directly from MCP-compatible clients such as Claude, Cursor, and other AI tools. In the demo, we first connect a Heym workflow to Claude using a custom connector. Claude can then call the workflow through MCP, Heym executes it in the background, and the result comes back directly inside Claude. Then we show how to connect multiple Heym workflows to Cursor. The selected workflows appear as tools in Cursor, and each tool can call the related workflow inside Heym. This makes Heym more than a workflow builder. It becomes an execution layer for your AI toolchain. You can build workflows visually in Heym, expose them as MCP tools, and run them from the clients your team already uses. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, workflow templates, documentation improvements, and pull requests. #ai #automation #workflow #agents #heym #mcpserver #mcpclients #mcpworkflows #mcptools https://www.youtube-nocookie.com/embed/rsLKKzMXsAI 2026-05-16T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-vs-chatbot Heym Portal: Expose AI Workflows as Chat Interfaces https://i.ytimg.com/vi/ig3oCbwReVs/maxresdefault.jpg Heym Portal lets you expose AI workflows as chat interfaces. In this demo, we run an AI Provider Pricing Monitor and a city-based workflow through Portal. Users can send inputs, receive streamed responses, continue conversations, copy messages, and download results without opening the workflow editor. Product: https://heym.run GitHub: https://github.com/heymrun/heym Support Heym by starring the repo. Contributions are welcome. #Heym #AIWorkflowAutomation #AIAgents #WorkflowAutomation #LLM #Automation #SelfHosted #DeveloperTools #AIPortal #NoCode https://www.youtube-nocookie.com/embed/ig3oCbwReVs 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/rag-vs-fine-tuning Heym Evals: Test Every Prompt Against Every Model https://i.ytimg.com/vi/P7-eK8u-UlQ/maxresdefault.jpg You changed one line in your prompt. Did your agent get better, or does it just feel better? Heym Evals lets you test every prompt against every model and see the answer in numbers. In this video: • Build a suite around the system prompt you want to test, and let Optimize Prompt suggest a sharper version. • Write test cases by hand or let a model generate them. Each case pairs an input with the answer you expect. • Pick several models and run them side by side, as many times per test case as you need. • Score with Exact Match, Contains or LLM-as-Judge. The judge can be an OpenAI, OpenAI-compatible or Gemini model, or a Decision Model that scores from the probabilities it computed on a five-level rubric. • Compare accuracy, latency and tokens per model, follow accuracy run after run, reopen any past run and export it as JSON. Chapters 0:00 Did the prompt get better? 0:14 Build a suite 0:24 Test cases 0:34 Models and scoring 0:44 Run every model 0:58 How the judge scores 1:10 Results, history and export Heym is a source-available, self-hostable automation platform purpose-built for AI workflows, with a visual canvas, multi-agent orchestration, built-in RAG, MCP support and built-in evals. Website: https://heym.run Evals docs: https://heym.run/docs/tabs/evals-tab GitHub: https://github.com/heymrun/heym #LLM #AIAgents #PromptEngineering https://www.youtube-nocookie.com/embed/P7-eK8u-UlQ 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/text-to-workflow From Prompt to Workflow: Heym AI Assistant Demo https://i.ytimg.com/vi/5ijD2lGz7zQ/maxresdefault.jpg Build AI workflows faster with Heym AI Assistant. In this demo, we show how to create a workflow from scratch, generate nodes with natural language, edit the workflow through chat, and run it directly from the canvas using workflow builder. Product: https://heym.run GitHub: https://github.com/heymrun/heym From Prompt to Workflow: Heym AI Assistant Demo #Heym #AIWorkflowAutomation #AIAssistant #AIAgents #WorkflowAutomation #Automation #LLM #SelfHosted #OpenSource #vibecoding https://www.youtube-nocookie.com/embed/5ijD2lGz7zQ 2026-05-09T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-lead-qualification-agent 2026-05-15T00:00:00.000Z weekly 0.8 https://heym.run/blog/prompt-chaining 2026-05-04T00:00:00.000Z weekly 0.8 https://heym.run/blog/llm-orchestration 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/how-to-build-ai-agent Persistent Memory and Second Brain for AI Agents https://i.ytimg.com/vi/bvUqTVu5gUc/maxresdefault.jpg In this video, we look at Heym’s Agent Memory Graph, a second brain for AI agents that need persistent memory over time. We use an orchestrator and sub-agent example to show how agents can keep, inspect, and share memory across workflows. With persistent memory enabled, each agent can store long-term context. Heym visualizes this memory as a graph, where teams can search entities, inspect relationships, and understand how different pieces of information are connected. This is useful for long-running agents, multi-agent systems, and workflows that need context over time. Instead of starting from zero in every run, agents can learn, remember, and reuse knowledge. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIAgents #AgentMemory #SecondBrain #AgenticAI #WorkflowAutomation #GraphMemory https://www.youtube-nocookie.com/embed/bvUqTVu5gUc 2026-08-05T00:00:00.000Z weekly 0.8 https://heym.run/blog/how-to-connect-two-apis-workflow 2026-04-27T00:00:00.000Z weekly 0.8 https://heym.run/blog/best-mcp-servers-2026 Heym MCP Server: Expose Workflows as Tools for Claude, Cursor and OpenAI https://i.ytimg.com/vi/rsLKKzMXsAI/maxresdefault.jpg In this video, we walk through Heym’s built-in MCP server feature. Heym can expose existing workflows as MCP tools, so they can be called directly from MCP-compatible clients such as Claude, Cursor, and other AI tools. In the demo, we first connect a Heym workflow to Claude using a custom connector. Claude can then call the workflow through MCP, Heym executes it in the background, and the result comes back directly inside Claude. Then we show how to connect multiple Heym workflows to Cursor. The selected workflows appear as tools in Cursor, and each tool can call the related workflow inside Heym. This makes Heym more than a workflow builder. It becomes an execution layer for your AI toolchain. You can build workflows visually in Heym, expose them as MCP tools, and run them from the clients your team already uses. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, workflow templates, documentation improvements, and pull requests. #ai #automation #workflow #agents #heym #mcpserver #mcpclients #mcpworkflows #mcptools https://www.youtube-nocookie.com/embed/rsLKKzMXsAI 2026-06-13T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agents-vs-rpa Extract Invoice Data from Any PDF Layout, Automatically | Heym https://i.ytimg.com/vi/MeBoKY9HgMA/maxresdefault.jpg Invoices arrive as PDFs, and someone usually copies the numbers into a spreadsheet by hand. This 90-second video shows a Heym workflow doing that job instead. Drop in a PDF invoice and an AI agent reads it, pulls out the vendor, invoice number, date, total and line items, and writes them as a clean row in a table. When the next invoice uses a completely different layout, the same columns fill in, with dates and amounts normalized to one format. How the workflow works: • InvoiceURL takes the link to the PDF • InvoiceExtractor, an AI agent, sends the PDF to LlamaParse and reads the result • It returns structured JSON: vendor_name, invoice_number, invoice_date, due_date, subtotal, tax, total, currency and line_items • InvoiceData passes that JSON on to a DataTable, QuickBooks, Xero or your next step The template is free. Open it, add your LlamaParse key and a model credential, and run it on your own invoices: https://heym.run/templates/invoice-pdf-extractor Chapters 0:00 An invoice arrives as a PDF 0:10 Drop it in, the fields land in a table 0:34 Another invoice, a different layout 0:46 The same fields fill in 0:58 Nobody copies the numbers 1:10 The invoice arrives, the row is ready Heym is an AI workflow automation platform with a visual editor: https://heym.run The invoices, vendors and amounts in this video are fictional. #invoiceautomation #aiagents #workflowautomation https://www.youtube-nocookie.com/embed/MeBoKY9HgMA 2026-04-25T00:00:00.000Z weekly 0.8 https://heym.run/blog/how-to-build-rag-pipeline Ask Your Company Files a Question, Get an Answer You Can Check - Self Hosted | Heym https://i.ytimg.com/vi/8Et2k-z56kc/maxresdefault.jpg Your team asks the same questions every week. How many remote days do we get? What is the travel allowance? The answers are already in your policy documents, but nobody wants to dig through a PDF to find them. This 90-second video shows the Chat with Drive Docs workflow in Heym. You upload your company PDFs to Drive and ask questions in a chat. The agent finds the right file, reads it, and answers with the exact sentence it used, so anyone can check where the answer came from. When the files do not cover a question, it says so. Ask whether you can bring your dog to the office and it searches all four policies, finds nothing, and tells you it does not know instead of making something up. How the workflow works: • userQuery takes the question from the Chat Portal • docAgent lists your Drive files with the listDriveFiles tool • readDriveFile fetches the PDF, and the pdf_base64_to_text skill extracts the text • answer sends the reply back to the chat The template is free to use. Open it, choose a model credential, upload your PDFs to Drive, and start asking: https://heym.run/templates/drive-chat-with-docs Chapters 0:00 Put the company files in Drive 0:12 Ask a question 0:32 The answer, with the line from the file 0:48 A question the files do not answer 1:00 It says it does not know 1:12 From the file, not from the chat Heym is an AI workflow automation platform with a visual editor: https://heym.run The company, policies and files in this video are fictional. #aiagents #documentai #workflowautomation https://www.youtube-nocookie.com/embed/8Et2k-z56kc 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-memory Persistent Memory and Second Brain for AI Agents https://i.ytimg.com/vi/bvUqTVu5gUc/maxresdefault.jpg In this video, we look at Heym’s Agent Memory Graph, a second brain for AI agents that need persistent memory over time. We use an orchestrator and sub-agent example to show how agents can keep, inspect, and share memory across workflows. With persistent memory enabled, each agent can store long-term context. Heym visualizes this memory as a graph, where teams can search entities, inspect relationships, and understand how different pieces of information are connected. This is useful for long-running agents, multi-agent systems, and workflows that need context over time. Instead of starting from zero in every run, agents can learn, remember, and reuse knowledge. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIAgents #AgentMemory #SecondBrain #AgenticAI #WorkflowAutomation #GraphMemory https://www.youtube-nocookie.com/embed/bvUqTVu5gUc 2026-08-05T00:00:00.000Z weekly 0.8 https://heym.run/blog/ai-agent-use-cases Importing Heym Workflow Templates in One Minute https://i.ytimg.com/vi/pfW3DQMEk-g/maxresdefault.jpg This step by step tutorial shows how to import free AI workflow templates from the Heym website into your own self hosted workspace. You will see how to open the templates page, pick a template, use Copy JSON, and paste it into the canvas so all nodes and connections appear instantly. We finish by editing node settings, adding API keys, and running the workflow so you have a complete template to production path you can repeat for any use case. #AI #AIAutomation #AIWorkflows #Workflow #Automation #NoCode #SelfHosted #OpenSource #Tutorial #HowTo #Developer #Heym #AITools #JSON #AIAgent https://www.youtube-nocookie.com/embed/pfW3DQMEk-g 2026-06-07T00:00:00.000Z weekly 0.8 https://heym.run/blog/multi-agent-ai-systems Persistent Memory and Second Brain for AI Agents https://i.ytimg.com/vi/bvUqTVu5gUc/maxresdefault.jpg In this video, we look at Heym’s Agent Memory Graph, a second brain for AI agents that need persistent memory over time. We use an orchestrator and sub-agent example to show how agents can keep, inspect, and share memory across workflows. With persistent memory enabled, each agent can store long-term context. Heym visualizes this memory as a graph, where teams can search entities, inspect relationships, and understand how different pieces of information are connected. This is useful for long-running agents, multi-agent systems, and workflows that need context over time. Instead of starting from zero in every run, agents can learn, remember, and reuse knowledge. Heym: https://heym.run GitHub: https://github.com/heymrun/heym #Heym #AIAgents #AgentMemory #SecondBrain #AgenticAI #WorkflowAutomation #GraphMemory https://www.youtube-nocookie.com/embed/bvUqTVu5gUc 2026-08-05T00:00:00.000Z weekly 0.8 https://heym.run/blog/what-is-agentic-ai 2026-04-10T00:00:00.000Z weekly 0.8 https://heym.run/blog/best-ai-agent-builders-2026 2026-06-22T00:00:00.000Z weekly 0.8 https://heym.run/blog/how-to-build-mcp-server Heym MCP Server: Expose Workflows as Tools for Claude, Cursor and OpenAI https://i.ytimg.com/vi/rsLKKzMXsAI/maxresdefault.jpg In this video, we walk through Heym’s built-in MCP server feature. Heym can expose existing workflows as MCP tools, so they can be called directly from MCP-compatible clients such as Claude, Cursor, and other AI tools. In the demo, we first connect a Heym workflow to Claude using a custom connector. Claude can then call the workflow through MCP, Heym executes it in the background, and the result comes back directly inside Claude. Then we show how to connect multiple Heym workflows to Cursor. The selected workflows appear as tools in Cursor, and each tool can call the related workflow inside Heym. This makes Heym more than a workflow builder. It becomes an execution layer for your AI toolchain. You can build workflows visually in Heym, expose them as MCP tools, and run them from the clients your team already uses. Product: https://heym.run GitHub: https://github.com/heymrun/heym If you find Heym useful, please support us by starring the repository. We are open to all kinds of contributions, including bug reports, feature ideas, workflow templates, documentation improvements, and pull requests. #ai #automation #workflow #agents #heym #mcpserver #mcpclients #mcpworkflows #mcptools https://www.youtube-nocookie.com/embed/rsLKKzMXsAI 2026-09-25T00:00:00.000Z weekly 0.8 https://heym.run/blog/what-is-ai-workflow-automation Heym Chat Control Your AI Workflows with Natural Language https://i.ytimg.com/vi/Umimz8ovb4I/maxresdefault.jpg Heym Chat is more than a Q&A interface. In this short demo, we show how the Chat tab in Heym works as a conversational control layer for AI workflows. With Heym Chat, you can choose your provider and model, use or edit Quick Chat items, describe what you need in natural language, attach documents for extra context, and get results directly from the chat interface. Instead of navigating through multiple screens, you can talk to your automation platform and let it execute. Product: https://heym.run GitHub: https://github.com/heymrun/heym What you will see in this video: Provider and model selection Quick Chat items Natural language requests Document attachment support Chat as a control layer for AI workflows #Heym #AIWorkflowAutomation #aiagents #workflowautomation #llm #automation #RAG #MCP #SelfHosted #opensource https://www.youtube-nocookie.com/embed/Umimz8ovb4I 2026-04-07T00:00:00.000Z weekly 0.8 https://heym.run/templates/categories/marketing-seo 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/sales-crm 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/customer-support 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/legal-compliance 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/hr-recruiting 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/finance-ops 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/document-ops 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/data-analytics 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/dev-it-ops 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/app-integrations 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/personal-productivity 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/education-learning 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/healthcare 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/ai-agents 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/multi-agent 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/operations-automation 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/categories/agent-skills 2026-07-15T00:00:00.000Z weekly 0.78 https://heym.run/templates/html-status-page 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/workflow-change-audit-log 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/hitl-support-reply-agent AI Drafts Your Support Replies. You Decide What Gets Sent. Email Agent | Heym https://i.ytimg.com/vi/OyrorEWCM5g/maxresdefault.jpg An AI agent can draft every reply in a busy support inbox. What matters is who decides what actually reaches the customer. In this 90-second demo, a Heym workflow reads a new customer email, writes a reply and then pauses. Nothing is sent until a person reviews it. You can approve the draft as it is, fix a sentence before approving, or reject it so the customer never sees it. What you will see: • A new email arrives and the HITL Support Reply workflow picks it up • The agent drafts a reply and requests a human review, so the run waits • The reviewer gets a notification with a link to a review page showing a summary and the full draft • Accept sends the draft as written • Edit & Continue sends it with your changes • Refuse stops it, and the agent records that no reply was sent Chapters 0:00 A full inbox and a new email 0:16 Heym reads the email and drafts a reply 0:28 The workflow waits for you 0:46 Fix one sentence, then approve 0:58 Reject it and nothing is sent 1:12 Nothing goes out until you approve Heym is a source-available, self-hostable platform for building AI workflows on a visual canvas, with agents, RAG, MCP and human-in-the-loop review built in. Try the template: https://heym.run/templates/hitl-support-reply-agent Human-in-the-loop docs: https://heym.run/docs/reference/human-in-the-loop Website: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #HumanInTheLoop #CustomerSupport https://www.youtube-nocookie.com/embed/OyrorEWCM5g 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/codex-pr-fix-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/opencode-go-pr-fix-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/agent-codex-pr-dispatcher 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/reddit-subreddit-get-tool-call 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/batch-llm-status-tracker 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/imap-support-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/file-upload-intake-metadata 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/tesseract-ocr-pdf-to-text 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/jina-web-fetcher 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/brave-search-api 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/cursor-post-notifier 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/gemini-image-creator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/pdf-docx-translation 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/claude-blog-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/open-meteo-weather-snapshot 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/github-release-radar 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/research-writer-pipeline 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/webdev-blog-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/zenquotes-random-quote 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/jsonplaceholder-user-profile 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/event-analytics-clickhouse 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/set-remap-fields 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/urgent-line-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/inbox-tldr-summarizer 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/wait-debounce-handoff 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/slack-ai-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/slack-postgres-mcp-answer-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/daily-telegram-google-analytics-mcp 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/cloudback-mcp-backup-review 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/discord-incoming-webhook 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/batch-url-fetcher 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/rag-document-ingest 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/rag-qa-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/switch-language-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/resilient-http-error-handler 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/google-sheets-ai-enricher 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/playwright-visual-ai-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/telegram-faq-auto-reply 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/realtime-websocket-alert-relay 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/async-subworkflow-json-api 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/drive-share-link-mailer 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/grist-to-bigquery-sync-log 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/rabbitmq-delayed-publisher 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/self-stopping-status-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-content-repurposer 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/cold-email-ai-outreach 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-lead-qualification 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/payment-webhook-handler 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/daily-briefing-email 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/social-media-post-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/crm-contact-sync 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/seo-keyword-opportunity-finder 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/adversarial-code-review 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/binance-websocket-price-alert 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/youtube-rss-csv-exporter 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/drive-chat-with-docs Ask Your Company Files a Question, Get an Answer You Can Check - Self Hosted | Heym https://i.ytimg.com/vi/8Et2k-z56kc/maxresdefault.jpg Your team asks the same questions every week. How many remote days do we get? What is the travel allowance? The answers are already in your policy documents, but nobody wants to dig through a PDF to find them. This 90-second video shows the Chat with Drive Docs workflow in Heym. You upload your company PDFs to Drive and ask questions in a chat. The agent finds the right file, reads it, and answers with the exact sentence it used, so anyone can check where the answer came from. When the files do not cover a question, it says so. Ask whether you can bring your dog to the office and it searches all four policies, finds nothing, and tells you it does not know instead of making something up. How the workflow works: • userQuery takes the question from the Chat Portal • docAgent lists your Drive files with the listDriveFiles tool • readDriveFile fetches the PDF, and the pdf_base64_to_text skill extracts the text • answer sends the reply back to the chat The template is free to use. Open it, choose a model credential, upload your PDFs to Drive, and start asking: https://heym.run/templates/drive-chat-with-docs Chapters 0:00 Put the company files in Drive 0:12 Ask a question 0:32 The answer, with the line from the file 0:48 A question the files do not answer 1:00 It says it does not know 1:12 From the file, not from the chat Heym is an AI workflow automation platform with a visual editor: https://heym.run The company, policies and files in this video are fictional. #aiagents #documentai #workflowautomation https://www.youtube-nocookie.com/embed/8Et2k-z56kc 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/first-ai-agent Build Your First AI Agent in One Sentence | Heym https://i.ytimg.com/vi/wp9TY_OQUKw/maxresdefault.jpg Describe the job in one sentence and Heym's AI Assistant lays out the workflow for you. In this video, a single sentence becomes a working AI agent that turns a messy note into a clean bullet summary. What happens in the video: 1. Start with an empty canvas. 2. Type one sentence: "Summarize this note into bullets." 3. The assistant adds three steps: YourMessage, ChatAgent and AgentReply. 4. Paste a messy note and press Run. 5. A clean summary comes back. The workflow is the Build Your First AI Agent template. Open it, connect an OpenAI-compatible credential to ChatAgent, and change the system instruction to fit your own job. Already have Brave Search API or Website Loader workflows? Both are Heym templates. Describe the workflow you want in Chat, and it can run them to search the web and read pages before it builds. Chat uses the workflows you already have, so add them first. Chapters 0:00 An empty canvas and one sentence 0:18 The assistant lays out the steps 0:38 Paste a messy note and run it 0:56 A clean summary comes back 1:08 Web research from Chat Links Template: https://heym.run/templates/first-ai-agent Brave Search API template: https://heym.run/templates/brave-search-api Website Loader template: https://heym.run/templates/jina-web-fetcher Heym: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #WorkflowAutomation #Heym https://www.youtube-nocookie.com/embed/wp9TY_OQUKw 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/telegram-personal-life-manager 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/invoice-pdf-extractor Extract Invoice Data from Any PDF Layout, Automatically | Heym https://i.ytimg.com/vi/MeBoKY9HgMA/maxresdefault.jpg Invoices arrive as PDFs, and someone usually copies the numbers into a spreadsheet by hand. This 90-second video shows a Heym workflow doing that job instead. Drop in a PDF invoice and an AI agent reads it, pulls out the vendor, invoice number, date, total and line items, and writes them as a clean row in a table. When the next invoice uses a completely different layout, the same columns fill in, with dates and amounts normalized to one format. How the workflow works: • InvoiceURL takes the link to the PDF • InvoiceExtractor, an AI agent, sends the PDF to LlamaParse and reads the result • It returns structured JSON: vendor_name, invoice_number, invoice_date, due_date, subtotal, tax, total, currency and line_items • InvoiceData passes that JSON on to a DataTable, QuickBooks, Xero or your next step The template is free. Open it, add your LlamaParse key and a model credential, and run it on your own invoices: https://heym.run/templates/invoice-pdf-extractor Chapters 0:00 An invoice arrives as a PDF 0:10 Drop it in, the fields land in a table 0:34 Another invoice, a different layout 0:46 The same fields fill in 0:58 Nobody copies the numbers 1:10 The invoice arrives, the row is ready Heym is an AI workflow automation platform with a visual editor: https://heym.run The invoices, vendors and amounts in this video are fictional. #invoiceautomation #aiagents #workflowautomation https://www.youtube-nocookie.com/embed/MeBoKY9HgMA 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/linkedin-candidate-ranker 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-video-tiktok-publisher 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/stripe-quickbooks-receipts 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/linkedin-sales-outreach 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/blind-eval-trio 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/coding-agent-github 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/google-analytics-mcp-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/google-search-console-mcp-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/blood-panel-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-provider-pricing-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/governed-web-research-agent-mcp 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/adversarial-pr-review 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/customer-survey-sentiment-slack-digest 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/sales-pipeline-risk-slack-standup 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/candidate-scorecard-sheet-screener 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/portfolio-diversification-slack-brief 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/contract-risk-review-queue 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/referral-intake-priority-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/competitor-launch-radar-slack 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/kpi-anomaly-slack-watch 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/weekly-focus-coach-from-sheets 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/property-inspection-escalation-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/employee-onboarding-orchestrator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/pto-request-approval-hitl 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/contract-renewal-expiry-tracker 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/gdpr-dsar-intake-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/patient-appointment-reminder 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/speed-to-lead-autoresponder 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/account-icp-scoring-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-sales-proposal-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/sales-call-notes-to-crm 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/renewal-upsell-opportunity-finder 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/meeting-notes-to-action-items 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/meeting-prep-brief-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/goal-to-weekly-plan 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-email-reply-drafter 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/daily-task-prioritizer 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/csv-stats-analyzer-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/html-table-to-markdown-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/pii-redactor-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/invoice-total-calculator-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/text-readability-scorer-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/employee-offboarding-checklist 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/job-description-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/employee-pulse-survey-analyzer 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/nda-generator-from-intake 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/contract-obligation-extractor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/compliance-checklist-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/insurance-prior-auth-prep 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/post-visit-followup-drafter 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/market-research-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/customer-feedback-triage-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/trip-planner-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/cease-desist-letter-pdf-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/demand-letter-pdf-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/privacy-policy-pdf-skill 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/offer-letter-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/interview-kit-generator 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/content-production-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/incident-response-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/rfp-response-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/due-diligence-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/hiring-committee-agent-team 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/discord-slash-command-faq-reply 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-incident-digest-to-discord 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/sentry-open-issue-digest-discord 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/care-coordination-intake-registry 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/finance-evidence-s3-archive 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/github-issue-intake-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/jira-bug-intake-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/deterministic-mcp-catalog-lookup 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/normalized-webhook-json-response 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-research-brief-supabase-archive 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/daily-github-issue-digest-discord 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/linear-bug-intake-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/notion-meeting-notes-action-page 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/csv-to-json-converter 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/python-latency-sla-report 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/clinical-trial-eligibility-intake-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/fhir-lab-result-release-qa-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/provider-credentialing-expiry-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/referral-leakage-network-optimizer 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/erp-crm-account-reconciliation-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/order-to-cash-exception-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/inventory-allocation-conflict-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/edi-asn-discrepancy-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/freight-invoice-audit-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/product-page-seo-catalog-sync-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/programmatic-seo-location-page-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/voice-of-customer-faq-schema-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/revops-enrichment-waterfall-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/support-contract-sla-credit-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/google-drive-to-heym-drive-backup 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-language-speaking-practice-coach 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/vocabulary-spaced-repetition-flashcard-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/pdf-assignment-rubric-feedback-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/academic-literature-review-research-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/standards-aligned-lesson-plan-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/lecture-notes-quiz-generator-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/lms-at-risk-student-intervention-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/corporate-training-module-builder-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/differentiated-instruction-accessibility-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/citation-bibliography-audit-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/reading-comprehension-worksheet-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/mobile-app-review-release-radar 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/personal-trainer-program-review-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/daycare-pickup-authorization-workflow 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ecommerce-product-recommendation-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/pet-vaccination-record-reminder 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/live-football-score-commentary-relay 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/anime-watchlist-research-agent 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/player-feedback-liveops-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/decision-model-rag-reranker 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/decision-model-smart-model-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/decision-model-agent-tool-gate 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/decision-model-x-post-extension-backend 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/decision-model-ai-text-humanizer Watch an AI Agent Workflow Run, Step by Step, in Heym https://i.ytimg.com/vi/f22KcSSgAyA/maxresdefault.jpg A draft scores 0.96 on an AI-writing check. We hand it to a Heym workflow and follow the run from the first node to the final response. It comes back at 0.27. The workflow is the Decision Model AI Text Humanizer. An agent rewrites the text five sentences at a time, a decision model scores each batch, and a loop keeps the best of five rounds. Everything it does along the way stays visible. In this video: • The run on the canvas, node by node • The Execution Log with status, timing and output for each node, and the final response • Inside the agent: every checkAi tool call, its score, and the rewrite it triggered • The decisions the loop made across five rounds • The execution timeline and the full LLM trace, with prompt, tool calls, tokens, cost and latency • A failing step, automatic retries, and Ask AI explaining the error • Pinned data, so the next run skips a slow step • Run history with Bring to Canvas, and edit history with diffs, preview and revert Chapters 0:00 A draft scored 0.96, run on the canvas 0:20 Execution Log and the response 0:30 Inside the agent: tool calls and decisions 0:42 Traces: timing, tokens and cost 0:52 Debugging: retries, errors and Ask AI 1:02 Pinned data and run history 1:18 Edit history Try the workflow: https://heym.run/templates/decision-model-ai-text-humanizer Heym: https://heym.run Docs Execution Log and pinned data: https://heym.run/docs/reference/canvas-features Traces: https://heym.run/docs/tabs/traces-tab Run history: https://heym.run/docs/reference/execution-history Edit history: https://heym.run/docs/reference/edit-history Source code: https://github.com/heymrun/heym #Heym #AIAgents #WorkflowAutomation https://www.youtube-nocookie.com/embed/f22KcSSgAyA 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/decision-model-support-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/google-ai-mode-visibility-drop-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/universal-commerce-protocol-checkout-readiness 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/a2a-agent-card-compatibility-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/mcp-oauth-token-audience-security-audit 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/eu-ai-act-article-50-transparency-ledger 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/c2pa-content-credentials-validation-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/sbom-vex-exploitability-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/post-quantum-tls-migration-inventory 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/owasp-agentic-excessive-agency-review 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/rag-knowledge-base-poisoning-sentinel 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/shadow-ai-saas-discovery-dlp-router 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/deepfake-vendor-payment-change-triage 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/gpu-inference-cost-per-token-anomaly-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-agent-trace-slo-burn-alert 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/ai-crawler-robots-policy-drift-monitor 2026-07-15T00:00:00.000Z monthly 0.75 https://heym.run/templates/community/torqon-mcp-memory-loop 2026-07-03T00:00:00.000Z monthly 0.6 https://heym.run/templates/community/meeting-notes-json-tasks-community 2026-07-03T00:00:00.000Z monthly 0.6 https://heym.run/company/adidas.com 2026-07-15T00:00:00.000Z monthly 0.65 https://heym.run/company/trendyol.de 2026-07-15T00:00:00.000Z monthly 0.65 https://heym.run/playground/browser-automation 2026-07-28T00:00:00.000Z weekly 0.8 https://heym.run/playground/ai-agent Build Your First AI Agent in One Sentence | Heym https://i.ytimg.com/vi/wp9TY_OQUKw/maxresdefault.jpg Describe the job in one sentence and Heym's AI Assistant lays out the workflow for you. In this video, a single sentence becomes a working AI agent that turns a messy note into a clean bullet summary. What happens in the video: 1. Start with an empty canvas. 2. Type one sentence: "Summarize this note into bullets." 3. The assistant adds three steps: YourMessage, ChatAgent and AgentReply. 4. Paste a messy note and press Run. 5. A clean summary comes back. The workflow is the Build Your First AI Agent template. Open it, connect an OpenAI-compatible credential to ChatAgent, and change the system instruction to fit your own job. Already have Brave Search API or Website Loader workflows? Both are Heym templates. Describe the workflow you want in Chat, and it can run them to search the web and read pages before it builds. Chat uses the workflows you already have, so add them first. Chapters 0:00 An empty canvas and one sentence 0:18 The assistant lays out the steps 0:38 Paste a messy note and run it 0:56 A clean summary comes back 1:08 Web research from Chat Links Template: https://heym.run/templates/first-ai-agent Brave Search API template: https://heym.run/templates/brave-search-api Website Loader template: https://heym.run/templates/jina-web-fetcher Heym: https://heym.run GitHub: https://github.com/heymrun/heym #AIAgents #WorkflowAutomation #Heym https://www.youtube-nocookie.com/embed/wp9TY_OQUKw 2026-07-28T00:00:00.000Z weekly 0.8 https://heym.run/playground/data-pipeline 2026-07-28T00:00:00.000Z weekly 0.8 https://heym.run/playground/weather-brief 2026-07-28T00:00:00.000Z weekly 0.8 https://heym.run/playground/incident-router 2026-07-28T00:00:00.000Z weekly 0.8 https://heym.run/playground/weekly-planner 2026-07-28T00:00:00.000Z weekly 0.8 https://heym.run/compare 2026-08-09T00:00:00.000Z monthly 0.82 https://heym.run/compare/n8n 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/zapier 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/make 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/pipedream 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/activepieces 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/power-automate 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/dify 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/flowise 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/langflow 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/node-red 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/windmill 2026-08-09T00:00:00.000Z monthly 0.76 https://heym.run/compare/gumloop 2026-08-09T00:00:00.000Z monthly 0.76