"The contradiction between the official statement and the document was flagged automatically. That was the story, and we had it documented before we published."
A six-month investigation generates hundreds of documents, dozens of sources, and a tangle of claims that confirm, contradict, and complicate each other. Spreadsheets and notebooks don't find contradictions. AI chat tools don't remember last month.
Drop in government filings, corporate reports, leaked documents, official statements, court records. Stratified chunking handles 300-page PDFs, always indexing the structural sections so nothing important gets missed regardless of where it sits in a long document.
Every entity in your evidence base, person, organisation, document, event, appears as a node in the knowledge graph. Edges connect them by subject-verb-object relationships extracted from every source. Click any entity to see everything Epistamate knows about it across your entire investigation.
When an official statement claims entity A did X, and a document in your knowledge base shows entity A did the opposite, Epistamate flags it as a contested claim with both sources named. Triplet-based, same subject, opposing predicate object. The kind of contradiction that's the story gets surfaced automatically.
Before filing the story, log the decision. Epistamate records every verified finding, every contested claim, every source, every gap that remained, timestamped, immutable, exportable. If legal challenge follows publication, you have the complete record of what you knew and what you acknowledged not knowing.
Corporate governance investigation, financial irregularities in a listed entity. Each claim traceable to source, with contradictions surfaced automatically.
Fully local, nothing leaves your machine. Contradiction detection across all sources. Story graph showing entity relationships. Decision Log for defamation defence.
Get in touch →