FinePrint launches a consumer-first AI contract scanner with source-page citations

Available today on iPhone, iPad, Android, and Mac, the app reads everyday contracts and explicitly cites the page each finding came from, addressing AI hallucination concerns at the UX layer.

McLean, Virginia. June 1, 2026.

FinePrint, a new AI-powered document scanner built for everyday consumers rather than law firms, launched today on the App Store and Google Play. The app reads PDFs of consumer agreements, leases, employment contracts, offer letters, insurance policies, gym memberships, HOA rules, auto leases, and service contracts, and produces plain-language findings, each one cited to the exact page of the source document.

While most legal AI products focus on lawyers, law firms, or enterprise buyers, FinePrint targets the documents normal people sign every year and don't read: a lease, a job offer, a gym contract, an insurance policy. The category list is intentionally broad and document-type aware, the app applies different extraction rules for an offer letter (acceptance deadlines, signing-bonus clawbacks, equity vesting cliffs) than for a homeowner's insurance policy (named exclusions, sub-limits, claim-filing deadlines).

The trust design

Every finding cites a page number. Tap a finding to see the verbatim quote from the original PDF, and a "View on page N in original PDF" link that opens the source document at the cited page. The product positions this as a hallucination guardrail: if the model fabricates something, the user sees that the citation doesn't actually contain the claimed clause.

The disclaimer is explicit and front-and-center

FinePrint provides informational summaries only and is not legal advice. The app surfaces this every time a finding is opened, in the share/export flow, and at the bottom of every screen showing extracted content.

"Most legal-AI tools either oversell, 'a lawyer in your pocket', or disclaim themselves into uselessness. I wanted something honest: here's what the document appears to say, here's the page it says it on, and here's what might matter. The user decides what to do next."

Jitesh Nadimpalli, Founder of FinePrint

Nadimpalli is a solo founder who built FinePrint after losing money to gym-cancellation fees buried in his membership terms. The first real user was himself, running the prototype on his own apartment lease renewal, it caught half a dozen clauses he'd missed, including a notice-to-vacate window that had quietly shortened from 60 days to 30.

Technical posture worth noting

• Document-type-aware prompt modules: 10 supported document types each have their own extraction guidance, severity calibration, and "things to look for" checklists
• "Exhaustive extraction" mode for restrictive document types (leases, HOA rules, gym memberships) finds 30–50 distinct findings per document; "selective" mode for less rule-heavy types (employment, general)
• Server-side validation: any finding the model emits that fails the citation invariant (the quoted text must actually appear verbatim on the cited page) is dropped before reaching the user
• Backend on AWS (Lambda, DynamoDB, S3, Bedrock Claude), processed in us-east-1, never used for AI training

Availability

FinePrint is available today on the App Store and Google Play. Free tier requires no account. Learn more at fineprint.irohtechnologies.com.