LogicFlow
All results

AlidadeAI integration

Let the founder run his platform through Claude, without AI doing the math

An MCP server that gives AI safe access to surveys, templates and distributions.

The full Alidade story
Alidade: AI integration

  • 95%

    smaller payload for template updates

  • 1

    MCP endpoint with full template CRUD

  • 0

    numbers calculated by the AI

logicflow · alidade · the work
before

The challenge

  • The platform UI is dense by necessity: matrix questions, branch logic, permission trees, template content cells.
  • The founder and new lab members could not operate it efficiently.
  • When AI did the math itself, it produced wrong numbers.
  • An MCP layer is an engineering project of its own: auth, schema versioning, payload size, safety guards.
what we built

The solution

  • An MCP connector at /api/v2/mcp for surveys, templates, distributions and filter calculations.
  • AI calls server-side calculations instead of computing numbers itself.
  • A safety guard: AI can edit only the templates it created.
  • Payload optimization and a backward-compatible schema for a growing tool surface.

after

The results

  • Achieved a production MCP endpoint with full template CRUD (live March 2026), without exposing human-authored templates to AI edits.
  • Achieved correct numbers in AI-generated reports, without the AI doing any math.
  • Cut MCP payload size for template updates by 95%.
  • The founder now uses Claude as the main interface to operate his own platform.
“The MCP is a huge leap forward to have AI built in with a chatbot. I was thinking that would be a way later stage.”
Dr. Todd Hall, Founder, Alidade

built with

  • Model Context Protocol
  • Claude
  • Node.js
  • Oracle

More from Alidade

All 9 Alidade stories
logicflow · your build

Thinking about your own build?

Get a clickable prototype in 48h

Free. 2 prototypes a month. Yours to keep.