Hey Product Hunt - I am Gabriel, one of the builders behind Repo. I started working on Repo because every serious AI agent eventually runs into the same problem: it can sound useful, but it does not reliably know what the company knows, where that knowledge came from, or whether it is allowed to use it. Repo is our attempt at a company memory layer for agents. Instead of giving an agent one giant prompt, the agent asks Repo for context. Repo checks scope, searches approved sources, and returns source-backed context with evidence, freshness, citations, and limitations. We are early and still building in public. The first version focuses on connectors, semantic retrieval, API access, and our Context Contract: the structured response format agents receive when they ask for company context. The bigger vision is simple: agents should not just answer confidently. They should answer with receipts. I would love feedback from people building agents, internal tools, support workflows, sales ops, or enterprise AI systems. What context do your agents struggle to access safely today?

Repo

Company memory for production AI agents

Company memory for production AI agents