Hiring teams waste hours reading resumes that a parser could rank in seconds. CVault uploads your CV stack, scores every candidate against your role (0-100 fit score), flags skill gaps, and exports clean JSON — in ~5 seconds per resume. Unlike keyword matchers, it understands spatial layouts (yes, even two-column Canva PDFs). Unlike raw LLMs, it never hallucinates your schema. Free tier: 10 parses/month forever. No credit card needed. 27k+ resumes parsed, 95%+ accuracy.
No reviews yetBe the first to leave a review for CVault AI
Maker
📌
Hey Product Hunt! I built CVault because I was tired of standard ATS parsers acting like glorified keyword matchers and raw LLMs constantly breaking JSON schemas at scale, and then charging you hundreds a month.
Between my AI bachelor classes, I've spent the last few months engineering a headless, hybrid parsing SaaS that actually understands spatial layouts (yes, even those terrible two-column Canva PDFs) and maps niche tech skills accurately without hallucinating. It beats TOP resume parsers on accuracy, speed and cost like: Affinda, Sovren and RChilli.
Here is what it does under the hood:
Spatial Awareness: Doesn't read left-to-right blindly; it understands layout so contact info and experience don't mix. It uses the top of the line OCR's to achieve this.
Deep Evaluation: Extracts text, evaluates skill depth, and returns candidate impact scores.
Uses custom made RAG systems, self learning Taxanomies and many more self learning features to constantly improve, and not make the same mistakes twice!
100% GDPR Compliant: Processes everything in-memory, then nukes it. Zero data retention.
Startup Friendly: Free for 10 parses/mo, then scales from $9.99/mo so solo devs don't have to buy bloated enterprise software. This is a mini ATS currently, with plans of making it something much bigger in the future!
We've successfully parsed over 27,000 resumes with a 99% read success rate, but I want to push the engine to its absolute limits.
If you’re building an internal hiring dashboard, job board, or AI recruiter—throw your absolute messiest, most chaotic PDFs at it. I would love your feedback on the architecture, and bonus points if you manage to break the extraction logic!
Report
This is actually interesting because you identified the real problem — not “parse text,” but “don’t lose signal because of layout + terminology.” The “candidate verdict / impact score” part is powerful but also risky — transparency on how that’s derived will matter a lot if this touches hiring decisions. $9.99 entry is smart. If it consistently beats keyword-based ATS parsing on messy PDFs, that’s a real wedge.
This is actually interesting because you identified the real problem — not “parse text,” but “don’t lose signal because of layout + terminology.” The “candidate verdict / impact score” part is powerful but also risky — transparency on how that’s derived will matter a lot if this touches hiring decisions. $9.99 entry is smart. If it consistently beats keyword-based ATS parsing on messy PDFs, that’s a real wedge.