Receipted

• 

• 

  recruiter

Inspiration

Resumes reward confidence, not truth. Recruiters waste time guessing who actually has the skills a job requires, and candidates inflate language because that’s what the system incentivizes. We wanted to flip that dynamic by making proof matter more than claims.

What it does

Receipted lets recruiters paste a job description, automatically extracts the core skills, and matches candidates based on evidence from real work stories. It flags suspicious or inflated claims and surfaces the best-fit candidate with clear explanations, not resume bullets.

How we built it

We built Receipted using FastAPI, Neo4j, and a lightweight frontend. You.com is used as a live “BS detector” to ground vague job requirements and claims against real-world definitions and sources. Akash provides a distributed compute option for running verification jobs reliably. Composio is used to generate and send personalized outreach emails to matched candidates.

Challenges we ran into

The biggest challenge was avoiding surface-level AI output. We had to design scoring logic that rewards consistency across multiple stories, penalizes sudden skill inflation, and integrates verification signals without unfairly punishing candidates.

Accomplishments that we're proud of

We built a recruiter-first workflow that goes from job description to ranked candidates in minutes, with explainable scores and evidence. We also designed a system where verification actually affects scoring, not just UI badges.

What we learned

AI is most useful when it constrains exaggeration instead of amplifying it. Grounding, transparency, and simple explanations matter more than flashy models.

What's next for Receipted

We plan to expand verification depth, support ATS integrations, and give recruiters configurable trust thresholds based on their hiring needs.

Built With

• cursor
• python