Comp AI CRM
trycompai · CRM
GitHub stars
11,002Stats updated
What is Comp AI CRM?
Comp AI CRM puts ongoing research ahead of a conventional chat-over-records experience. Its agent runs separately from the browser, works through a durable queue, books follow-ups, and stops at a research budget. Observations carry source identity, while uncertain findings remain visible for review rather than being treated as established facts.
Comp AI CRM is an open-source CRM designed around an autonomous research agent. The agent keeps its own work queue and schedule, records observed evidence, and leaves weaker evidence for human review.
Core capabilities
- Run durable research sessions from a scheduled work queue
- Record observations and source evidence about people and companies
- Surface uncertain evidence as suggestions for a human to settle
The agent operates the CRM's research loop
The agent leases due work from a queue, chooses the next research step, uses authored CRM and enrichment tools, and stops when its budget is exhausted. Deployments, skills, schedules, and sessions give that work a durable lifecycle outside a single conversation.
- agent
- queue
- schedule
- research
Evidence stays attached to every conclusion
Tools report observations and their source identity, and the evidence ledger decides whether they support a record update or a suggestion for human review. External research sources are optional: the agent can also work from CRM history. This is an evidence-handling design, not a guarantee that every resulting customer fact is correct.
- evidence
- provenance
- review
- CRM
Where it fits
Use cases
- 01
Research a person or company
Assign an investigation to the agent, let its tools gather evidence from configured sources, and retain the resulting observations with provenance inside the CRM.
- 02
Keep research moving after sign-off
Use the durable queue, deployment, schedule, and research budget to continue due work without depending on an open browser tab or a manual chat session.
- 03
Review evidence before changing records
Allow strong observed evidence to update CRM facts while routing weaker or conflicting findings into a visible suggestion that a human can accept or reject.