Patent Watch Wants Your Dormant Patents to Start Paying Rent
The Stroe brothers built AI that reads a patent, finds the product infringing it, and drafts the claim chart in about twenty minutes. Y Combinator and a $2.8M seed round think the patent back office is ready for software.
There is a peculiar kind of waste in the world of intellectual property. A company will spend up to $50,000 filing and maintaining a single patent over its lifetime, hire lawyers, argue with examiners, wait years - and then, the moment the patent is granted, file it in a drawer and forget it exists. Patent Watch is a company built almost entirely around that drawer.
The Toronto startup, part of Y Combinator's Fall 2025 batch, makes AI software that reads a patent, scans the market for products that appear to use the same invention, and generates the document lawyers call a claim chart - the line-by-line comparison that proves infringement. What normally takes patent attorneys months, Patent Watch says it does in roughly twenty minutes.
The OriginA frustration at Philips
The company comes from two brothers, Alexander and Andreas Stroe. Andreas was doing research at Philips when he filed four patents of his own and got a close look at the machinery behind a corporate patent portfolio. Philips holds more than 50,000 patents. Defending them - knowing which ones are being used by competitors, and doing anything about it - turned out to be nearly impossible by hand.
That gap between owning intellectual property and being able to enforce it became the pitch. Alexander, who had spent time day-trading funded accounts and running large ad budgets before this, took the CEO seat. The two started Patent Watch in 2025 and were accepted into YC's Fall batch soon after.
The ProductWhat the software actually does
At its core, Patent Watch is a search-and-evidence engine. You give it a patent - or a whole portfolio - and it maps the claims against real products on the market, scoring how closely each claim element matches on a 0-to-100 scale. The output is a draft claim chart with the analysis already filled in, ready for a human expert to review and edit rather than build from scratch.
Claim chart, scored — illustrative example
Around that engine sit the rest of the tools: Product Mapping to find licensing and litigation targets across a portfolio, Patent Search over a large patent database, Prior Art and Invalidity Analysis for the defensive side, and Portfolio Analytics for a bird's-eye view. The same infringement engine feeds several jobs - drafting cease-and-desist letters, hunting counter-litigation angles, prepping M&A due diligence, or simply deciding what to sell.
Where the time goes
The CustomerWho's actually buying
The buyers are corporate IP teams, patent attorneys, and licensing firms - the people whose portfolios are large enough to be valuable and too large to police by hand. It is sold as B2B software, subscription-based, with a book-a-demo enterprise motion. Around launch the company reported early recurring revenue and pilots with larger tech firms, run by a team of about five.
The reframe Patent Watch leans on is worth sitting with: the hard part of a patent isn't getting it, it's finding out who is already using it. Enforcement, not invention, is where the money leaks out. A tool that closes that gap is selling leverage to a team that is almost always outnumbered by its own filings.
The MarketWhere it fits, and against whom
Patent intelligence is not an empty field. Incumbents like Clarivate - whose Derwent line added its own AI patent monitor - have sold to IP departments for years, and a wave of AI-native tools such as Patlytics and IPRally are chasing the same workflows. The traditional alternative is even simpler and more expensive: a law firm billing by the hour to chart claims manually.
Patent Watch's wager is that the full stack - detection, charting, invalidity, analytics - belongs in one product built for monetization rather than a pile of point tools stitched together after the fact. Whether that holds up against deep-pocketed incumbents is the open question, but the seed round suggests investors are willing to fund the attempt.
The funding so far
The BetBoring market, big budgets
Most founders chase shiny new markets. Patent Watch went the other way, straight into the paperwork behind intellectual property - reading claims, charting evidence, comparing products. It's unglamorous work, which is exactly why it has stayed manual and expensive for so long. If AI can do the grunt work accurately enough to survive a lawyer's review, the drawer full of forgotten patents starts to look less like dead weight and more like inventory.