An AI answer vetted by a council of models

One question goes to several models, they score each other's answers, and you get one vetted result by email.

  • Fewer flukes from a single model
  • Answers are compared and ranked against each other
  • The final result arrives by email
  • You just ask, as usual, in chat
Chat
Set it up for yourselfScriptera

How it works

To connect sendPulse and Chat, you don't need a developer: a ready-made scenario links them in minutes.

  1. Starts when: When chat message received
  2. Then: Initialize Variables
  3. Then: Split by Model
  4. Then: Stage 1: Query Model Answers
  5. Then: Stage 1: Extract Responses
  6. Then: Stage 1: Aggregate Responses
  7. Then: Stage 2: Create Anonymized Responses
  8. Then: Stage 2: Build Ranking Prompt
  9. Then: Stage 2: Re-Add councilModels
  10. Then: Stage 2: rankingPrompt per LLM
  11. Then: Stage 2: Query Model Answer Rankings
  12. Then: Stage 2: Extract Rankings
  13. Then: Stage 2: Aggregate
  14. Then: Stage 3: Calculate Aggregate Rankings
  15. Then: Stage 3: Build ChairmanPrompt Part 1
  16. Then: Stage 3: Build Chairman Prompt Part 2
  17. Then: Stage 3: Chairman Synthesis
  18. Then: Stage 4: Format Output Part 1
  19. Then: Stage 4: Format Output Part 2
  20. Then: Stage 4: Send Result per Email

You can launch this sendPulse + Chat integration in Scriptera: describe the task in plain words — the scenario is built, launched and monitored for you.

Similar scenarios