Find the connections, the conflicts, and the confidence in a corpus
Genesis is an investigative research workspace that maps how documents relate, surfaces where they disagree, and scores how certain each claim really is. All analysis runs locally, in your browser.
The seeded dossier is entirely fictional. Genesis does not use external AI services and needs no API keys.
Connections
A knowledge graph links entities, documents, and claims, so you can see how the pieces of a dossier fit together and trace the evidence behind any claim.
Conflicts
A deterministic heuristic finds contradictions: negation, opposing claims, incompatible numbers and dates, and hedged uncertainty, each with a confidence score and its evidence.
Confidence
A local TF-IDF pipeline ranks documents and passages by meaning and shows the shared terms, so you can judge how well the evidence actually supports a claim.
How it works
Ingest
Import text files or use the seeded corpus. Every document is parsed locally and added to a searchable queue.
Analyze
Genesis builds the graph, computes TF-IDF similarity, and runs the contradiction heuristic over the claims.
Investigate
Form a hypothesis, collect evidence, and review the conflicts that touch your investigation.
See it working in under a minute
Run the guided demo to walk the whole workflow: corpus, search, graph, conflicts, investigation, and challenges. Then reset and explore on your own.
The dossier, analyzed
This is what Genesis found when it ran over the seeded NovaGrid corpus. Open the workspace to reproduce every result and dig into the evidence.
Documents
7
Entities
8
Claims
16
Conflicts
13
One source frames this as positive and the other as negative (approval versus rejection).
NovaGrid's revenue grew 40% last year. but NovaGrid's revenue actually fell 12% over the same period.
One source frames this as positive and the other as negative (increase versus decrease).
The transaction is expected to close by the third quarter of 2026. but Closing has been delayed to 2027.