Written text — handwritten or not
Typed reports, scanned PDFs, printed forms, and handwriting alike. We pull out the entities, clauses and recurring structures, and keep every finding anchored to the page it came from.
Empirical Patterns finds the patterns hiding in unstructured material — written text typed or handwritten, video, and pictures. Drop it in, ask a question in plain language, get a structured answer back.
What we read
Real documents are messy. Scanned forms, a photo of a whiteboard, a note scribbled in a margin, an hour of footage. Empirical Patterns treats all of it as one corpus.
Typed reports, scanned PDFs, printed forms, and handwriting alike. We pull out the entities, clauses and recurring structures, and keep every finding anchored to the page it came from.
Walk-throughs, inspections, interviews, screen recordings. We read what happens on screen and what is said about it, then surface the moments that repeat — with the timestamp attached.
Photographs, diagrams, whiteboards, screenshots, charts. If a person could look at it and spot the pattern, so can we — across thousands of images at once.
How it works
No schema to design up front, no tagging project, no six-week onboarding.
Attach documents, images and video — mixed formats in the same batch. Nothing needs sorting first.
Recurring structures, entities, anomalies and relationships get recognised across the whole set, not one file at a time.
Question the corpus like a colleague who has read all of it. Answers come back with the source behind them.
Structured output through the API, so the patterns land in the systems your team already runs on.
Point Empirical Patterns at the material you already have and ask it the question you have been meaning to answer.