Agatha Career Intelligence

Agatha Career Intelligence is the product I point at when someone asks what "shipping the thing" looks like. It is not a resume template pack with a chatbot glued on. It is a career platform that takes a messy PDF, turns it into structured profile data, matches roles, rewrites documents for each posting, and can run the application loop while the candidate sleeps.

The job, not the demo

Job search is a queue of repetitive work: parse the CV, scan twenty boards, rewrite bullets for ATS, draft a cover letter that is not generic, and actually submit. Agatha owns that queue.

• Upload a CV and get a queryable profile, not a blob of text.
• Match against live openings instead of a static job dump.
• Tailor the resume per role — keywords, structure, and ATS score in the same pass.
• Generate a cover letter that cites the posting, not a paragraph of adjectives.
• Auto-apply through a connected inbox when the candidate wants the agent to move.

The live product is at agathalabs.app. That URL is the case study.

What I actually built

The architecture is a Next.js / TypeScript full stack with background workers for the slow jobs — crawling, matching, document generation — so the UI never waits on a model. Profile extraction turns unstructured PDFs into a schema a matcher can query. Skill inference is the unglamorous piece: "built a React app" has to mean hooks, state, and ES6+, or the matching layer is theatre.

ATS quality is measured, not claimed. Templates are recruiter-vetted. The score is computed against the posting, not a vibes check. When the agent applies, Gmail is connected over OAuth. Credentials stay with the candidate. Nothing is trained on their CV.

Why this is the featured case

Agatha is the standard I hold the rest of the studio to: a live URL, a real workflow, and a handoff a client can run a meeting with. If you want the build notes, I wrote them up in Shipping Agatha Career Intelligence.