
AI platform combining 1,500+ prebuilt connectors, custom Python MCP servers attached from the UI, and an n8n-style visual workflow builder. Users build knowledge bases and AI pipelines on any connector, while a harness-engineered agent reasons across connectors, MCP tools, knowledge bases and live web search. FastAPI backend with Microsoft Entra ID OAuth, PostgreSQL, Azure AI Search and Redis. Rolled out to HR, DevOps and infrastructure teams: 65%+ productivity lift.
The problem
Teams wanted AI that works with their own tools, but every new integration meant custom code.
What I built
Ship 1,500+ connectors and let users attach their own Python MCP servers from the UI.
Add an n8n-style workflow builder and knowledge bases on top of any connector.
One harness-engineered agent reasons across connectors, MCP tools, knowledge and web search.
How it works
Connect: 1,500+ connectors, Custom MCP servers, Live web search
Build: Knowledge bases, Visual workflow builder
Reason: Harness-engineered agent, Tool selection, Memory
Serve: FastAPI backend, HR · DevOps · Infra teams
Built-in controls: Microsoft Entra ID OAuth, PostgreSQL · Redis, Azure AI Search.
Result
Rolled out to HR, DevOps and infrastructure teams with a 65%+ productivity lift.
My role: AI Engineer & Backend Architect
0 likes · 1 view · Nov 13, 2026
Teams wanted AI that works with their own tools, but every new integration meant custom code. Result: Rolled out to HR, DevOps and infrastructure teams with a 6