IPscreener

IPscreener is a self-hosted, semantically driven patent research platform. Its AI Agent runs novelty and freedom-to-operate searches from a plain-language invention description, while adjacent tools cover invention drafting, claim charts, license evidence, and monitoring.

IPscreener is a Swedish patent intelligence platform built around a semantic AI engine that runs on its own infrastructure. The core idea is simple: instead of spending days constructing boolean queries and sifting through result lists, you describe an invention, a product, or a patent in plain text and the system plans and executes the research for you. The new AI Agent handles Novelty Search, Freedom to Operate screening, validity search, and invention landscape mapping end to end, producing a source-grounded report that can be exported to PDF or Excel. The company reports more than 20,000 cases analysed and over 5,000 active users, with logos from Philips, ABB, HP, Novartis, and Roche on its homepage. The platform is more than a single search box. Invention Manager turns raw disclosure text and supporting files into a structured, filing-ready draft using templates and live AI insight on prior art. Search Case starts from an invention text, a document, or a patent and gives you list, full-text, and image views with smart highlighting, comments, ranking, and a passage finder. Creative Studio is a chat interface for exploring the prior art landscape and generating reports or trend summaries. License Search maps patent claims to captured web pages as evidence of use, and Claim Chart builds a visual relevance table that links features to documents. Monitoring adds automated monthly delivery profiles and custom workflows with routing and archiving. The tool is aimed at R&D teams, patent attorneys, IP strategists, and startups that need to validate an idea quickly without reinventing the wheel. The privacy story is a differentiator: the AI is 100 percent in-house, hosted in a Swedish data center, with end-to-end encryption and no data sent to OpenAI, Google, or other third-party model providers. That matters for legal and enterprise buyers who cannot send confidential invention disclosures to external APIs. Practically, the platform sits between traditional patent databases (which demand query expertise) and general-purpose chatbots (which lack source grounding and patent-specific workflows). It competes on speed of first-pass analysis, leaving final legal judgement to the professional. Pricing is not published on the homepage beyond a free start with no credit card and a demo booking option, so tier costs need to be confirmed with sales. Some capabilities are labelled Pro, which suggests feature gating by plan. The workflow still requires human review, and the quality of output depends on how well the invention is described.