We built the resume tool we wanted.

Most resume tools fall into one of two camps. The free ones are templates and surface-level "ATS optimizers" that flag missing keywords and call it done. The paid ones promise miracles and deliver $39/mo of generic advice you could find on a Reddit thread.

We wanted something different: a tool that actually parses your resume the way an ATS does,tells you exactly which lines fail, and rewrites them in place while keeping your facts honest.

What's different

• Real ATS rules, not vibes. We model the parse and scoring rules common to Workday, Greenhouse, Lever, Taleo. The score is deterministic — same resume, same number.
• The agent doesn't lie. If a bullet has no metric, the AI asks you for one instead of hallucinating a number. Your facts stay yours.
• Every change is reversible. The History panel keeps the last 90 days of revisions. Restore any version with one click.
• Provider-agnostic. We default to a fast, cheap model. Pro users opt into higher-quality models for nuanced rewrites. Swap with one env var.

What we don't do

We don't sell resumes. We don't sell data. We don't lock parseability behind a paywall. The parsing, scoring, and heatmap are free forever — that's the part that's deterministic and useful even without an LLM in the loop.

Built with

Next.js 16, Vercel AI Gateway, Neon Postgres, Drizzle, Polar. Single-region, edge-runtime scoring, fluid functions for streaming chat. Bills under $0.01 per active user. We built the stack so the costs scale with you, not against you.

Get in touch

Bug reports, feature requests, hate mail: hello@resumes.repair. We read every message.