Cross-checks by multiple AIs
Evidence, Logic, Citation, and Counterargument personas independently scrutinize each finding before it advances. Counterarguments surface design-around risks early.
4 checking roles
by IP RichPatentOS Enterprise · For corporate IP departments
Every year, valuable patents are abandoned because the maintenance call is made on a hunch. PatentOS shows A/B/C ranks, Japan revenue of products that may infringe, and infringement candidates you had not found, each with its evidence.
What's at risk
Most corporate IP teams review many patents each year on partial information. Maintenance decisions get made under deadline, and nobody has the time to check whether products on the market may infringe your own patents.
How it works
Drop in a CSV of patent numbers, yours, your acquisition target's, or a watchlist. JP and US patents in one workflow.
Separate AI roles break down the claims, look for products that may infringe, keep agreement and dissent on record, and estimate revenue exposure with cited sources.
Anything ranked A (high likelihood of infringement, high revenue) can go to your own counsel, or to a partner patent attorney you engage directly, before you act.
A heatmap, a rank, a revenue estimate and an evidence package, per patent, so you can make retain, sell, license and abandon decisions with the evidence in hand.
Built differently
Procurement teams ask: "why this system?" PatentOS has separate AI roles check evidence, logic, citations and counterarguments, then leaves a reviewable decision record your legal team can inspect.
Evidence, Logic, Citation, and Counterargument personas independently scrutinize each finding before it advances. Counterarguments surface design-around risks early.
4 checking rolesContradictions are retained as reviewable notes instead of being averaged into false confidence. Counsel can see where the system was uncertain.
Reviewable uncertaintyExact, semantic, and source-specific retrieval work together so important assertions point back to product evidence, filings, or technical material.
Cited evidenceOne AI role combines the results into a decision record with sources, uncertainty and escalation notes.
One review recordJPO claim structures and FI/F-term classifications handled as filed. USPTO means-plus-function detection. Japanese utility models supported. No translation layer that drops nuance.
Both jurisdictionsHigh-conviction findings move forward, ambiguous findings route to review, and weak findings are rejected with the reason preserved.
Every important decision keeps reasoning notes, source citations, and timestamps. Months later you can reproduce why a maintenance call was made.
Maps a foreign parent company to its Japanese subsidiary automatically. This is the basis for Japan revenue estimates and is slow to do by hand at volume.
Every correction from a patent attorney feeds into how we evaluate accuracy, with accuracy targets set per technology area. Your data, corrections included, is never used to train shared models.
Use Cases
Four scenarios where PatentOS replaces guesswork with evidence, and surfaces value your team didn't know was there.
Triage a large portfolio at once. A/B/C ranking with revenue evidence helps you keep what's valuable, drop what's not, and lower the risk of abandoning an A-ranked patent by mistake.
Ongoing checks for products that may infringe your patents, across Japanese e-commerce, manufacturer websites and corporate filings, with source-cited evidence.
Valuation by patent and for the whole portfolio, with the assumptions and evidence your finance team needs.
Evaluate a target company's portfolio before the LOI: strong patents that are easy to miss, weak claims and prior-art exposure.
Sample Output
Every analysis ships as a portfolio heatmap, an A/B/C ranking, a revenue estimate, and a claim chart per high-rank patent. See the sample output.
View sample outputs →Built by former in-house IP staff
PatentOS's product team includes operators from Fuji Xerox's IP department and patent attorneys with practice in Japan, the US and China. We've sat in the seat where the maintenance call is yours and the budget pressure is real.
A-rank findings can go to your own counsel, or to a partner patent attorney you engage directly.
Each customer's data is handled in an isolated environment. Your data is never used to train shared models.
Built for both jurisdictions from day one. JPO and USPTO claim structures handled natively.
Every AI decision logged with reasoning chain. Defensible if a maintenance decision is ever questioned.
Pricing
We will invite 50 companies from among those who pre-register to join as early adopters. They get early-adopter pricing and a seat at our roadmap sessions.
FAQ
Bring your highest-stakes patent, or your annual maintenance shortlist. We'll run PatentOS on it live with your team. If it doesn't surface something you didn't already know, you walk away.