Asserting · Plaintiff side
Every claim element maps to source evidence in the accused product. AI proposes the mapping; counsel reviews, refines, and controls every chart before it's relied on. And because the same engine reads your patent's own prosecution history, you see the estoppel and disclaimer the other side will raise — before they raise it.
How it works
Contist structures claims into elements, maps them against targets, and generates evidence-backed analysis — all connected to the underlying prosecution history.
Independent and dependent claims are parsed from the patent and resolved into a structured hierarchy with the full dependency chain intact.
Each claim is decomposed into preamble, body, and wherein elements. Means-plus-function limitations are flagged automatically for attorney confirmation.
Accused products, one column of the matrix each. Upload PDFs, DOCX, or URLs. Prior art goes in the same way when you want to pressure-test your own claims before asserting them.
AI proposes met / partially-met / not-met per (element, target) with citations and confidence. Counsel reviews every cell in the matrix, refines the rating, and adds notes.
Narrative generation is gated: it only runs on cells an attorney has rated met or partially met. Nothing gets argued until the underlying triage is ratified.
Capabilities
Long-form narrative generation is per-cell and rating-gated. It only runs on cells an attorney has rated met or partially met. The rating decides whether to generate; the LLM still argues from evidence — the gate prevents narrative on cells no one has reviewed.
Attorneys can also flag their notes as soft guidance for the narrative. Off by default (notes stay private); on-by-checkbox when counsel wants the argument to follow a specific theory.
Claims × elements × targets. Each cell carries an AI-proposed rating (met / partially met / not met), source citations, and confidence. Reviewed one cell at a time through a slide-over panel.
Three-column charts add per-element construction under Phillips v. AWH Corp.: plain meaning to a PHOSITA, specification support with pinpoint citations, and constructed meaning. Cached per claim term.
Dependent claims are resolved through their full parent chain back to the independent root. LLM prompts get structured chain context; shared ancestors are deduplicated across sibling claims.
Every claim element is linked to its prosecution history. See which amendments narrowed the claim, which arguments distinguish prior art, and how scope moved from filing to grant.
Read the essay: Patent Prosecution History Is a First-Class IP AssetEvery mapping traces back to source evidence. Every claim traces back to prosecution events. Every construction traces back to spec passages. Outputs are reproducible on demand and reviewed by counsel before they are relied upon — the same chart, re-derived, says the same thing.
View the underlying prosecution history
Claim charts are powered by structured prosecution intelligence.