One model.
Four workflows.

Build invalidity grounds for an inter partes review filing. The model surfaces the prior-art universe most likely to land §102 anticipation and §103 obviousness arguments before the PTAB.

• ·Outcome-aligned ranker trained on Final Written Decisions — the model learns which references actually win, not just which match.
• ·Side-by-side compare for §103 obviousness combinations.
• ·Save shortlisted references into a matter with status + notes; export to a Word draft as a starting point for the petition.

Counter a petitioner's invalidity grounds. Find the prior art they should have cited but didn't — and the distinguishing references that strengthen claim-construction arguments.

• ·Forward-citation lookup flags obscure refs that lack credibility weight.
• ·Inline reader with highlights to mark distinguishing language for the response brief.
• ·Bulk mode runs all the asserted-claim families at once when the petition spans a portfolio.

Before buying a patent or portfolio, find the references most likely to surface in a future invalidity challenge. Surface the risk before the deal closes.

• ·Bulk search across an entire portfolio in one pass; per-patent risk summary.
• ·Structural evidence chips (shared CPC + shared citations) flag claim families with deep prior-art neighborhoods.
• ·Matters workspace groups every asset under one case for legal-team handoff.

Stand up invalidity contentions on the local-patent-rules clock. Replace weeks of manual prior-art search with a ranked, explained shortlist on day one.

• ·Snippet-level attribution shows which passages of each reference earned the rank — drops directly into element-by-element charting.
• ·Save references per asserted patent inside one matter; track Keep / Shortlist / Dismissed.
• ·Word-doc export gives the team a structured starting point before manual claim-element alignment.