Hey HN, we’re Soami, David, and Achyut, co-founders of Extract. Extract parses documents into structured data (text, tables, and figures). Teams use it for RAG, feeding llms, and populating databases and forms. Today we’re launching our first OCR model, now used in Extract.

You can try some examples here or upload your own (no signup required) to test it out: https://extract.page/demo

We built Extract out of YouLearn, where we were processing 70m+ pages and slow parsing was the bottleneck. We started with a purely algorithmic pipeline that pulled native text straight from the document and only ran OCR on pages that needed it. It was cheap and fast, but once we put it in front of our Extract customers and their hardest documents, it hit an accuracy ceiling. We wanted to keep the speed and cost while improving accuracy, so we trained our own VLM for the cases that broke. It also provides element level bboxes, so each result points back to its exact place on the page.

That took one customer from 71% to 92% text accuracy in under a week, at the same speed and cost. We can do this because of our synthetic data generation pipeline that recreates the messy, real-world documents the model gets wrong, so we can retrain on those exact cases without having to hand-label data.

To see how this holds up against other providers, we benchmarked Extract against AWS Textract, Extend, Reducto, LlamaParse, and Unstructured on 130 human labeled pages from difficult real-world documents. Extract is #1 on text accuracy (81.9%) and word-overlap F1 (84.5%), second on grounded accuracy, and competitive on layout IoU, while running at least 2x faster than every parser we tested.

Here are the benchmarks: https://extract.page/bench

Extract is $3 per 1000 pages and about 5x cheaper than AWS Textract (layout + table enabled). To see how it performs on your own docs, feel free to send us a few and we’ll run a benchmark on them. We’ll get back to you with the results in a few days once we receive the docs: https://cal.com/team/youlearnai/extract-intro

Thanks for reading this post! It's our first version of the model and we're shipping further improvements to handwritten, multilingual, and table-heavy documents. We know there are documents it won't handle well yet. If you have one, we'd love to see it.

Logicomix and 6 months at my new job

Fixing my ridiculous fridge with a tiny Funnel site

Companies Are Using Reddit to Manipulate ChatGPT and Google AI Search

Why are so many Show HNs being flagged?

Show HN: Testbump – automated test driven versioning

Dotnet-slopwatch – detect when AI coding agents "fix" problems by cheating

A victory for digital common sense: Bavaria's deal with Microsoft falls through

SemiAnalysis: TCO of Space Datacenters

CI caching is not one cache

GPS satellites have broadcast a "numbers station" in their public signals

Show HN: Division Swarm, the OS for Multi-Agent Systems

Signal Struggles on Trains Widespread

Qwen 3.7 Plus

Partial Graphics Programs

AgilitySDK 721 Preview and an addition to LinAlg in Shader Model 6.10

I built an API that stops AI hallucinating colour

Show HN: Nib, collaborative font editor on the web

Intel's new CEO cut management layers in half. The stock is up nearly 500%

(PewDiePie) any LOCAL AI model close to Claude Code?

Do we fear the serializable isolation level more than we fear subtle bugs?

Meta enters enterprise AI race with new business agent

Show HN: NoiR Code – because QR sounds similar to "noir"

AI Has Ruined the Job Market

AI to double data centre power and water consumption by 2030, UN researchers say

Running a full blockchain stack locally (2024, 22 minutes) [video]

Majorana 2, made more reliable with Microsoft Discovery agentic AI

Show HN: Extract (YC P25) – Fast, accurate document parsing

Tilt: A toolkit for fixing the pains of microservice development

DaVinci Resolve 21

Thoughts on Building with Agents

Show HN: Curatube: a distraction free interface for YT playlists