Bob Michaels/ai
All projectsCase study 02 of 10

A multi-tenant SaaS, built end to end

Content Lab

A multi-tenant SaaS that measures how AI engines recommend an agency's clients, then writes compliant content for them.

Agencies see how ChatGPT, Perplexity, Gemini and Grok recommend each of their clients, then generate compliant content, chatbots and leads from the findings. Each tenant's data is isolated by row-level security.

When
Mar to Aug 2026
Status
Deployed to production
commits since March
2,190
lines of Python and TypeScript
135k
API routers over 87 services
38
database migrations
~100

What this does not claim

Revenue and customer counts are not recorded on disk, so none are claimed here.

01

The problem

Buyers now ask AI engines who to hire. Agencies had no way to see what those engines say about their clients, and no way to act on it at scale.

02

What I built

A tenant-isolated platform that queries the major AI engines with buyer questions, scores who gets recommended and why, and drafts compliant content, chatbots and leads from the findings. About 22,000 lines of the code are tests.

It grew out of the GEO research engines I first built in n8n. Content Lab is where that research became a product other agencies could use.

03

What it produced

The platform is deployed to production. The chatbot work inside it became the Grounded Chatbot Platform.

Measured from the repository

2,190 commits, about 76k lines Python and 59k TypeScript, 87 services, 38 API routers, about 100 migrations.

The stack

  • Python and FastAPI
  • Next.js
  • Supabase Postgres with row-level security
  • Vercel
  • Inngest