AIDGE
Role
Founder
Year
2026
Category
AI / Fintech
Context
AIDGE Inc.
01.
Problem
A fund desk needs sector analysts, PMs, a CIO, compliance, and traders. Coverage grows with headcount. Most AI tools in finance stop at research. They draft memos. People still own the pitch, the size, the risk check, and the order. The rest lives in other tools with scattered logs.
I built a multi-agent stack modeled on a real desk. The system runs 60+ LLM agents: researchers, PMs, and a CIO. News and macro agents scan filings, feeds, and transcripts, then send names to sector pods. Inside each pod, fundamental, technical, and news analysts work the same name in parallel. A sector PM writes a pitch: thesis, sizing ask, catalysts, risks. A CIO agent ranks pitches and sizes the book against the mandate. Compliance checks every order against restricted lists and limits. Failed orders never leave. An execution agent sends the order, watches the fill, and writes one log from scan to fill. Backend: Python, FastAPI, Redis Streams, Postgres. Next.js for the public site.
02.
Approach
Gallery
