Coop detects the company on any page, researches it, scores the fit, and keeps your pipeline honest. When you ask Claude a question, it already has your emails, meetings, and full history loaded.
No backend. No account. Coop sends nothing anywhere - only your own Claude talks to Anthropic.
Detection, research, scoring, drafting, memory - the whole job, handled while you get on with your day. Here’s the running list.
Three of them are scored - Match, Conversation health, and Offer likelihood.
Click through the real flow. This is the product, not a storyboard.
Folded in 2 new meetings, rescored - Match 7.8 → 8.2 · now B+
Drafts land as drafts. Coop never sends.
How’s Mercury Labs really going?
Well, with one caveat. Health is 74 and the HM is warm, but their comp band tops out near your floor - anchor before the panel.
Where do I actually stand this week?
Mercury Labs is your strongest live thread - panel Thursday, health 74 and climbing. Lumen went quiet 9 days ago; a nudge is drafted and waiting for you.
Logged the recruiter reply and moved Lumen to Screening Call.
Land on any company page and he pulls it together: who they are, who runs it, what people say about working there. Nothing fires until you click.
5 great applications beat 50 mediocre, and he tells you which 5. Six months in, you should be glad you took the job you took.
The intelligence officer, not the autopilot. He briefs you on everything he sees. You make the call.
Your judgment lives in a settings box Coop actually reads - plain sentences, not config. Change one line and every score, draft, and brief downstream changes with it.
I built Coop during my own search. Five weeks from idea to daily use. It runs my search every day.
Private beta. The form takes about 60 seconds. I review every application personally.
Get beta access