Parse a newly released public document trove into structured data and build a consumer search product on top of it
When a large set of messy, redacted, scanned public documents is released, most people cannot read them. Parse the whole corpus into structured data with an agentic document-processing pipeline, then build an interface (e.g. an inbox-style email browser) that makes it searchable and browsable for the public.
online B2B Online Directories & Lead Generation
From AI Engineer — Parsing the Epstein Files for JMail — Palak Agarwal & Omar Alhait, Reducto at 02:02
Problem: Millions of pages of scanned PDFs, handwriting, complex forms and broken redactions are unreadable; nobody will read three million pages, so the data is inaccessible until it is extracted and structured.
For: General public / journalists / researchers interested in a high-profile document release
Products from this video
JCal JFlights JMail Reducto Reducto Documentation Reducto Studio
Examples
- JMail — Gmail-style interface where users log into the inbox as if they were Jeffrey Epstein; built in five hours, grew to over 450 million visits and 18 million users by February; Reducto parsed the 3+ million pages behind it
Other takes on Directory sites
- Curated local city directory with paid business listings and optional printed city guides
- Build a portfolio of niche directory websites that aggregate hard-to-find information and monetize organic traffic
- A deeply enriched niche directory built from individual business websites
- A searchable directory for estate sale cleanout and senior downsizing businesses
- Build a niche directory website that organizes listings by category or location and sends visitors to detailed listing pages where businesses can receive conversions.
- Evergreen, location-based directory websites that curate useful local information