Open-Sourcing Comp AI: A Sales Agent That Rebuilds CRM with an Evidence Ledger
aigclink · x · 2026-08-03
Comp AI has open-sourced an agentic-first sales research CRM. The core idea is that traditional CRMs are just databases with forms, whereas this product centers on a durable agent that independently schedules and conducts customer research.
Four engineering highlights stand out:
- Independent Scheduling: The agent uses its own queue with FOR UPDATE SKIP LOCKED to prevent overlapping tasks, with automatic lease release on crashes.
- Zero Intelligence in API: NestJS only logs events into the queue; the agent decides what they mean, preventing logic drift between services.
- No Confidence Scores: To avoid sycophantic hallucinations, the system uses an evidence ledger. Strong evidence updates records directly, while weak evidence becomes a suggestion for human review.
- Strict Sandboxing: The sandbox denies all outbound network access and database credentials to prevent leaks, running web scraping at the app level.
The system includes 18 custom tools and 4 markdown skills, capable of functioning without external API keys by using local data, though it is heavily tied to the Vercel ecosystem.
Related event: Comp AI Open-Sources Native Agentic CRM Under MIT License(6 posts)→
More from coding & agent
- Connected steps are not a workflow: nobody designs the handoff — usehive · 2026-10-16
- Making Blender animations with Opus, Astra and Argon: how you prompt still matters — chongdashu · 2026-10-16
- Running PolyStrata + Cline coding agent on a $0-fancy laptop at ~90 tok/s — rpwoerk · 2026-10-16
- How do you attribute agent gains: model training or infrastructure? — jrxna · 2026-10-16
- Feed Claude your books, blogs and top engineers' code for claude.md — 'it now writes better code than me' — blaizedsouza · 2026-10-16
- Open-source /label plugin auto-names and colors Claude Code sessions in one step — hermanschutte · 2026-10-16