Which Tech Center, which art unit, which CPC class. Get it wrong and you're not just off by a guess, you're setting the wrong prosecution strategy, quoting the wrong timeline, maybe filing claims that were worth rewriting first. Most classification tools answer that question by matching keywords against whatever art happens to be sitting in an index somewhere, and if that index hasn't been updated in years, neither has the prediction.
Live September 2026, Juristat’s Claims Analyzer update provides IP professionals with the most accurate analysis in the market, backed by evidence. This tool pairs keyword matching with a language model trained on recent patent text, so two applications describing the same invention in different words actually get recognized as similar, not just the ones that happen to share vocabulary. It's trained on the last two years of filings, refreshes every quarter, and a refresh only ships if it beats the version already in production.
Claims usually get written and rewritten in real time, often in a conversation with Claude, Co-Pilot, or ChatGPT, long before they ever reach a docketing system. Claims Analyzer now runs directly in your LLM via MCP connector, or through the Juristat API as a step in your own intake process. That changes what classification checking actually is: not a lookup you run once at the end, but something you check mid-draft and act on immediately.
Ask something like "revise claim 1 to avoid a 3600 art unit and re-run it" and get an answer in the same conversation, without switching tools or losing your place.
Claims Analyzer is live now in the Juristat web app, on the Juristat MCP server, and via the Juristat API. Reach out to your Juristat representative to talk through what this looks like for your firm or your portfolio, or sign up to see it live here.