FeaturedCustomer Support#Decision Model#Noul#Choice#Score#Support Triage

Decision Model Support Triage

Use one Decision Model to assess urgency, choose a support team, and score business impact. Route tickets using all three answer types.

Workflow at a glance

The full canvas, before you import it

Click any node to see its config.

#Decision Model#Noul#Choice#Score#Support Triage

Click a node to select it β€” same as the Heym editor; the panel shows its settings.

8 nodes Β· Free & source-available

Decision Model Support Triage

A customer reports that checkout is failing while a campaign is live. Another asks for an invoice copy. These requests need different handling. This template uses a single Decision node to evaluate both with three typed questions, then routes each ticket to a priority review or the appropriate support queue.

Three question types in one model call

TypeQuestionHow the workflow uses it
NoulDoes this ticket need immediate attention?A yes probability of at least 0.8 triggers priority review.
ChoiceWhich support team should own the ticket?Billing and technical requests get separate queues. Other requests go to general review.
ScoreHow much business activity is affected?A score of at least 2 on a 0 to 3 rubric also triggers priority review, even without urgent wording.

What this workflow does

  1. SupportTicket supplies an editable customer message, with a checkout outage as the default example.
  2. TicketDecision asks all three questions in one request to the configured decision model.
  3. PriorityGate checks urgency and business impact. Either threshold sends the ticket to PriorityReview, with the suggested team and all three answers attached.
  4. For tickets below both thresholds, TeamRouter reads the Choice answer and selects BillingQueue, TechnicalQueue, or GeneralReview.
  5. The selected Output node displays the routing result, original message, urgency probability, team, and impact score.

Setup

  1. Copy the template into the Heym canvas.
  2. Create a Decision Model credential in Credentials using your provider's base URL and API key if required. Select it on TicketDecision.
  3. The example model is jev-latest. Change it if your endpoint uses a different model name. An LLM credential is not used by this template.
  4. Run with the default message, then replace the text input to try the examples below.

The final nodes display queue recommendations. They do not send notifications or create external tickets. Once you have checked the routing, you can connect your own Slack, email, or helpdesk actions to the branches.

Try different tickets

Sample messageIntended demonstration
Our checkout has returned payment errors for every customer for 25 minutes. A paid campaign is live and orders have stopped. Please help immediately.High urgency and high impact should lead to priority review, with technical support suggested.
Could you send a copy of last month's invoice? Our account works normally and there is no deadline.A routine billing request should reach the billing queue.
The export button has a visual alignment issue. Exports still work and nobody is blocked. No rush.A low-impact technical issue should reach the technical queue.
For your awareness, the nightly sync has stopped for all teams. Nobody can access today's records and there is no workaround.Business impact can trigger priority review even when the wording is calm.
Can your team speak at our community meetup next month?A request outside billing and technical support should reach general review.

These are illustrative routing expectations. Actual probabilities and scores depend on your model and the message.

Read the answers

  • Noul: $TicketDecision.answers.is_urgent.noul is a probability from 0 to 1, not an urgency rating.
  • Choice: $TicketDecision.answers.team.choice is billing, technical, or other. The model also returns a probability distribution and confidence.
  • Score: $TicketDecision.answers.business_impact.score uses four levels: 0 for no disruption, 1 for minor friction, 2 for blocked work, and 3 for widespread disruption. The result is probability-weighted and can fall between levels.

The thresholds are editable examples. Compare the routing against your own support tickets before connecting external actions. Keep node labels stable so the expressions continue to resolve.

How to import this template

  1. 1Click Import β†’ Copy JSON on this page.
  2. 2Open your Heym and navigate to a workflow canvas.
  3. 3PressCmd+V/Ctrl+Vβ€” nodes appear instantly.
  4. 4Add your API keys in the node config panels and click Run.
More workflow templates
View all templates
Heym
incident analysis Β· production AI
Observed across 100s of AI rollouts

AI workflows don't fail because of prompts.
They fail because of orchestration.

symptom Β· glue code01
5 tools
Scripts, vector DB, approval bot, tracing, browser runner β€” none of them talk.
symptom Β· visibility02
~0%
Observable behavior across the stack. Debugging is guesswork.
with heym Β· one runtimeβ–²
1 canvas
Agents, RAG, HITL, MCP, traces & evals. Self-hosted. Observable.
AI-Native RuntimeProduction-Grade
github.com/heymrun/heym