Recon
BuildingVerified company memory for AI agents, built from your docs, tickets, Slack, code, and customer context.
Built by
Pratik Kumar JainAbout Recon
Recon is a building project that provides a verified memory for AI agents, built from your docs, tickets, Slack, code, and customer context. Your docs, conversations, and tickets become one verified source of truth.
Recon is currently in the building stage. Its tech stack includes Next.js, TypeScript, OpenAI, Claude, Tailwind CSS, Vercel, Supabase, MCP, and Postgres.
Evidence comes from Notion, Slack, Zendesk, Intercom, and HubSpot. Security and access controls include encryption at rest, workspace-scoped access, read-only by default, per-action write approvals, and audit logs.
Key features
- Unified verified memory built from your docs, tickets, Slack, code, and customer context.
- Each belief has a receipt and is linked to the exact source line.
- Claude, ChatGPT, and Slack can read the same verified record and respond with receipts.
- When sources disagree, a deterministic verdict surfaces contradictions and indicates what's true.
- Evidence comes from tools like Notion, Slack, Zendesk, Intercom, and HubSpot.
- Security and access controls include encryption at rest, workspace-scoped access, read-only by default, per-action write approvals, and audit logs.
Tech stack
Frequently asked questions
What is Recon?
A verified memory for AI agents built from your documents, tickets, Slack, code, and customer context.
Who is it for?
AI agents that need a verified company memory.
What tech does Recon use?
Next.js, TypeScript, OpenAI, Claude, Tailwind CSS, Vercel, Supabase, MCP, and Postgres.
What is the stage of Recon?
Currently building.
