Skip to main content

N°01 · Multi-Tenant SaaS Automation Platform

13 specialised LLM agents for proposal generation, discovery chat and competitive analysis, using tool calling to fetch CRM and knowledge-base data.

What I built

  • →RAG with embeddings + vector search grounds output in client documents; agent flows modelled as LangGraph-style stateful graphs.
  • →A2A orchestration with a policy-gated human-approval lifecycle: pending → dispatched → awaiting_approval.
  • →JWT middleware enforces 100% tenant data segregation; agent tracing and eval checks on every proposal.

Outcome

60%

faster release deploys

Stack

  • FastAPI
  • React 19
  • Supabase
  • Docker
  • Gemini 2.5 Pro

Tell me what you are trying to fix, and I will tell you how I would approach it.