Intake call notes that become scorecards

The idea

The intake call is where a search is won or lost, and most recruiters compress it into bullet fragments in an ATS notes field. Three weeks later nobody remembers whether the client said Kubernetes was a must-have or a nice-to-have, and candidates get judged against vibes. The crowd has already declared generic AI meeting notes a commodity — the reference product charges $14/mo for summaries — so a summary alone is not a product.

The wedge is what happens after the intake: the call becomes a weighted scorecard, and every subsequent candidate screen is automatically graded against it with quoted evidence. A horizontal note-taker summarizes each call in isolation; it will never connect call five back to call one. Agencies already pay well above $14 per seat for tooling, and this artifact plugs straight into how they already sell shortlists to clients.

Who pays

The buyer is the agency owner or team lead at a 2-20 person recruiting shop, paying $29/seat/mo. They justify it as fewer bad submissions and a client-ready spec PDF that reduces spec disputes. Distribute through recruiting communities, LinkedIn content aimed at agency owners, and integrations marketplace listings for Bullhorn and Recruiterflow later.

MVP scope

• Recall.ai bot that joins intake and screen calls from a pasted link
• Claude extraction into an editable structured job spec
• Auto-generated weighted scorecard from the must-haves
• Candidate screens scored against the scorecard with transcript quotes as evidence
• Ranked candidate view per role and PDF export of the spec
• Skip for v1: ATS integrations, calendar auto-join, client portals, custom scorecard templates

Build IntakeDeck, a web app for recruiting agencies that turns client intake calls into job specs and reusable candidate scorecards. Stack: Next.js (App Router), Recall.ai for meeting bots, Claude API for extraction, Postgres via Prisma, deployed on Vercel.

Core flow: recruiter pastes a meeting link or connects their calendar; a Recall.ai bot joins the intake call and returns the transcript. Claude extracts a structured job spec: role summary, must-have requirements, nice-to-haves, dealbreakers, comp range, interview process, and urgency. The recruiter reviews and edits, then the app generates a candidate scorecard — each must-have becomes a weighted criterion with a 1-5 scale and evidence field.

The compounding feature: every subsequent candidate screen for that role also gets a bot. After each screen, Claude scores the candidate against the intake scorecard, citing transcript quotes as evidence for each rating, and produces a ranked candidate list per role.

Pages: roles dashboard, intake review editor, scorecard builder, candidate list with rankings, candidate detail with quoted evidence, team settings. Non-obvious details: store criteria as versioned rows so editing a scorecard mid-search does not corrupt earlier candidate scores; let recruiters export the job spec as a client-ready PDF for sign-off, which becomes the source of truth in disputes.

Pricing: $29/seat/mo, minimum 2 seats, 14-day trial, Stripe billing. Free tier: 2 roles, no candidate scoring.