Back to resources
RepoAI & machine learningSenteLabsAI/OpenExecutive

OpenExecutive

A virtual executive team of 8 specialist Claude agents behind one coherent persona — CSO, CFO, People, GC, COO, CMO, CPO, Board Comms. FastAPI + Next.js, RAG over your own company docs.

by SenteLabsAI

OpenExecutive (SenteLabsAI) is eight specialist Claude agents — CSO, CFO, People, General Counsel, COO, CMO, CPO, Board Comms — behind a single orchestrator (Sonnet 4.6) that fans out the query in parallel and merges the replies into one voice instead of eight. Each specialist pulls from ChromaDB seeded with MBA-level knowledge plus whatever company docs you upload; episodic memory in SQLite carries past decisions across sessions, and prompt caching keeps persona + company profile cached (~85% hit rate) while the RAG context stays uncached in the user turn so answers don't go stale.

Backend is Python 3.11 + FastAPI with ChromaDB and SQLite, frontend is Next.js 15, and the deep-reasoning roles (CSO, CFO, GC, Board) escalate to Opus 4.7. Ships with web, Slack, email, Telegram, Google Chat, Discord, and CLI front ends. Drop ANTHROPIC_API_KEY in .env and make dev — first boot is slow while it pulls the embedding model. I keep it bookmarked as a reference for running several specialist agents behind one orchestrator without the illusion breaking.

Open resource