How it works

The platform reads what institutions publish, works out what applies to you, and shows you the sentence it came from.

Reading is continuous, not a project
Sources publish on their own schedules and in their own formats. Some post a PDF to a page, some run a feed, some put a notice in a gazette supplement weeks after the decision. The platform watches all of them daily, so the question is never whether anyone remembered to check.
Documents are parsed, not skimmed
Text layers are read directly. Pages without one, usually scans, are put through optical recognition on their own so a single bad page does not hold up a document. Tables are kept as tables, because a fee schedule read as prose is worse than no fee schedule.
Extraction records what supports it
Entities, relationships and obligations come out together with the span of text that carries them. Nothing enters the graph without a document, a page and the words themselves, so a finding can always be taken back to its source.
Rules run against your commitments
Your own strategy documents, policies and agreements set the frame. Findings are ranked by how close they sit to what your institution has already committed to, which is the difference between a feed of everything published and a short list that concerns you.
People decide, and the record keeps both sides
A finding arrives with its evidence, its owner and the time remaining. Confirm it and it stays confirmed. Disagree and it is sent back with your reason attached. The disagreement is kept as part of the record rather than deleted.

Request a demo

We are onboarding early partners ahead of commercial launch. Tell us what your institution watches and we will show you the platform against it.