Hyrable
A job-search operating system — scrape listings from five sources, track applications through your own pipeline, and tailor documents against evidence you've confirmed, with AI that's optional and bring-your-own-key.
The problem
Job searching is a pile of disconnected steps: scrape listings, read each one, pick the right resume, apply, then remember which applications are even still open. Most tools either automate too much or too little — Hyrable keeps the human in the loop on judgment calls and automates the busywork around them.
The sharper problem is grounding. A model asked to "tailor my resume for this job" will happily invent the experience the posting asks for. That's the one failure mode a job seeker cannot afford.
How it works
Extractors pull listings from Indeed and LinkedIn (via JobSpy), HiringCafe, Remotive, Jobicy, and WorkingNomads into a single board. From there the loop is review the listing, pick or generate a document, apply, and track it through pipeline stages you define yourself.
The AI layer is real but deliberately optional. Every user brings their own provider key — validated live, stored AES-256-GCM encrypted per user — and the whole product works with zero AI configured: board, tracking, and manual documents never gate on it. A server-side key is a self-host fallback and pointedly does not flip the per-user capability, so no tenant ever gets AI they didn't opt into.
Grounding is the design center. Tailoring reads only your confirmed evidence
corpus — cards seeded from your resume that you've reviewed — through a single
evidence-packet seam, and every generated document runs a post-generation
validation stack: sanitize, gates, retry, judge, provenance report. When it
can't ground a claim it returns rejected rather than a confident fabrication.
With no basis at all it fails closed.
What I'd change
Fit scoring, JD parsing, legitimacy checks, company research, and the tailoring
pipeline all ship. What's left is the interview-practice mode — still the one
placeholder tab — and the deployment on hyrable.tech: the full stack already
runs as containers (db → migrator → app) and CI is green on lint, type-check,
integration tests against Postgres, and builds, so what's left is wiring it to
the domain rather than building it.
