A CRM you talk to instead of fill in
Your accounts, contacts, and deals in real tables — captured, updated, and answered by agents rather than data entry.
No forms to fill in
Say what happened on the call. The record gets created, linked, and filed correctly without anyone opening a form.
Real records underneath
Everything lives in structured tables you can query, filter, and export at any time. It's a database, not a chat log.
Clean without the busywork
Duplicates get merged, missing fields get filled, and stale deals get surfaced — on a schedule, without being asked.
Two agents run the whole thing
One you work with directly, one that quietly keeps the data honest while you're doing something else.
- RRevenue OperatorCRM Interface
Captures notes into records, answers pipeline questions from real queries, and pushes records out to the tools you already use.
- PPipeline StewardAutonomous Record Hygiene
Never talks first. Wakes on new rows and a weekly sweep to dedupe, enrich, and flag what has gone quiet.
See exactly what gets deployed
Every agent, instruction, skill, and starter file is open to inspect before you create a workspace.
- Agents
- 2
- Skills
- 6
- Files
- 3
# Workspace This workspace is a headless CRM. The data tables are the system of record and the agents are the interface — there is no CRM UI to click through. A record that is not in a table does not exist. - /crm/SETUP.md — table schema and the one-time setup checklist. - /crm/CONVENTIONS.md — the rules every agent follows when writing records. Read it before writing, not after. ## Business context Fill this in during setup and keep it current: what is sold, to whom, typical deal size and sales cycle, where leads come from, and who owns the pipeline. Keep it factual and label assumptions as assumptions.
Ready to stop doing data entry?
Start from the template and have a working pipeline the same day.