AI prior-art search finds patents and papers that describe the same idea as an invention even when they use different words, and returns each result with the passage that matched so an attorney can judge relevance from the source rather than from a score.
All IP & Patent Firms →
Inventors, applicants and academic authors describe the same mechanism in different vocabularies, and a novelty search that relies on Boolean strings and classification codes inherits that blind spot. The document that sinks an application is often a paper or an older patent in a neighbouring field, phrased in a way no searcher thought to try.
Semantic search compares meaning rather than words, so it finds those documents. It also finds a lot of near-misses, and a ranked list with a similarity score is not a search report. What makes the output usable is evidence: for each hit, the passage that matched, mapped to the claim element it bears on, with the queries and databases recorded so the search can be repeated and defended.
We test on the firm’s closed matters, where the art an examiner later cited is known. For each, the system searches from the original disclosure as if it were new, and we measure whether the cited documents appear and how far down. That gives an honest view of recall in your technical fields, which vary a great deal: a search that performs well in mechanical engineering may not in biotechnology. Fields that underperform stay on the firm’s existing process until they are fixed.
Yes. Google Patents offers semantic search, and open projects such as PQAI provide natural-language prior-art search over patents and papers. They are useful for a first look. Professional searches usually need broader non-patent coverage, confidential handling of the disclosure, and a record of the search that can be repeated, which is where a firm’s own system or a commercial platform comes in.
It depends heavily on the technical field and the sources searched, which is why we measure it on your own closed matters before relying on it. Semantic search tends to find differently worded art that keyword search misses, and also surfaces more near-misses. It complements a skilled searcher rather than replacing one.
Queries to public search sites can, in principle, reveal what a firm is working on, and pasting a disclosure into a consumer AI tool raises retention and training concerns. We run searches from infrastructure under terms that forbid retention and training, or entirely on self-hosted models, with queries logged per matter.
What we built, what stage it is at, and a link to the live thing wherever there is one.
yourOS is an operating system with an AI at its heart. You talk to it like a person, out loud or typed. It makes your calls, plans your days and works on your projects with a crew of agents, on your own private server. Our own product, built from the first commit; the waitlist is open.
Website, backend and app assets for Sonora, a free AI sound-therapy app on iOS and Android published by Cognitive AI.
Brand, website, prototype, apps on five platforms, the AI backend and the algorithm-intelligence pipeline for Podswap, a social growth engine that runs without a human in the loop.
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Live at coherence.com: brand identity generated with AI agents we directed, the motion system and website for the group behind Sonora, AboutAI, WakeUp and MudShadows. Agent work remains in build.
Live at domainbooth.com for DomainBooth, a premium domain brokerage: a new animated mark, every page redesigned and rebuilt with designed mobile layouts, and a custom CMS for its sales record, featured names, press and enquiries.
We test against art an examiner already cited, before any live disclosure is searched.
30-minute technical call · no deck