LightfieldAI-native CRM that builds itself and does work for you
NeuroForge
One AI OS, six specialized agents for every workflow
10 followers
One AI OS, six specialized agents for every workflow
10 followers
NeuroForge AI is a multi-agent AI OS for real engineering & knowledge work. Six agents: Engineer AI (LangGraph Planner→Coder→Tester→Debugger), Research AI, Education AI, Automation AI, Conversational AI, plus an Admin/RAG Studio with org-level knowledge isolation (ChromaDB + Gemini embeddings). Built with FastAPI, LangChain, LangGraph, MongoDB, Groq, Docker. One unified OS instead of five separate tools. Built solo — would love feedback on agent orchestration & RAG design!
How does the org-level knowledge isolation actually hold up when multiple agents pull from overlapping documents at the same time, especially with the ChromaDB + Gemini embedding setup under load?
how does the agent orchestration actually work in practice when an engineer task needs input from the research agent mid-run, do they hand off context cleanly or does it get messy?
Shipping six coordinated agents as a solo build is genuinely impressive, and packaging them as one OS rather than scattered tools is a smart move. Curious how you're handling context handoffs between the Engineer and Research agents without losing state mid-task.