LLR supports innovation and is making available a new patent search engine powered by AI
We are pleased to announce the launch of a new free service on our website: an AI-powered patent search engine, freely accessible on the page AI Patent Search.
This new offering is the result of a technical integration carried out with TWENTY YEARS, an AI-powered patent search and analysis platform created by Étienne Gabella, French and European Patent Attorney. Étienne Gabella has been working with LLR for around twelve years and has our full support in connection with this initiative.
What this new tool enables
In practical terms, any visitor to our website can:
- Provide a description of an invention or product in natural language: a simple sentence, a paragraph, a full document (up to 90,000 characters, equivalent to a 40-page document);
- Obtain within seconds a list of relevant patents or patent applications published by the European Patent Office (EPO), the USPTO (United States), and the CIPO (Canada) over the past 20 years;
- View the key information for each publication: publication number, current owner, legal status, validation territories, as well as the scope of protection defined by the closest independent claim.
No keywords are required. That is precisely what makes the tool accessible to inventors and companies discovering the world of patents.
How it works
When a user enters a description:
1. TWENTY YEARS’ proprietary language models extract technical features;
2. These features are structured to form candidate claims;
3. These candidate claims are compared by semantic similarity against more than 30 million real patent claims indexed in the TWENTY YEARS database;
4. The Twenty Years engine produces a relevance score for each publication whose claim is close to one or more of the candidate claims;
5. The tool returns the 10 publications with the highest respective relevance scores
This approach goes beyond the limits of traditional keyword searches: a relevant patent using terminology different from that used by the user will still be identified.
Exemplary confidentiality at the heart of the approach
For us, this is an essential point, and what fundamentally sets TWENTY YEARS apart from other AI solutions available on the market.
A user entering an invention description into this engine may potentially be describing an invention that is not yet protected.
Yet:
- Unlike mainstream AI tools (such as ChatGPT, Perplexity, Gemini), whose terms of use often allow the exploitation of input data
- And unlike certain professional tools that themselves rely on third-party AI providers…
…TWENTY YEARS uses its own proprietary language models, hosted on its own servers within the European Union (currently in Germany, through a provider governed by European law).
No data entered via the search engine is transmitted to OpenAI, Anthropic, Google, or any other third party. The search is strictly confined to TWENTY YEARS’ infrastructure, under a European legal framework.
A tool designed for fast and relevant searching
Our objective in integrating this tool into our website is simple: to enable our visitors (inventors, business leaders, innovation managers) to carry out an initial patent search themselves that is relevant, fast, and fully confidential.
The LLR teams remain fully at your disposal to complement such searches and explore your matters in greater depth: protection strategy, infringement analysis, freedom to operate, filings, litigation, opposition, and all industrial property issues.
For those who wish to access more advanced features (additional results, search refinement, automated preliminary freedom-to-operate analysis, etc.), visit https://twentyyears.ai.
→ Access the search engine: https://www.llrip.com/en/ai-patent-search
→ Contact us for in-depth support: info@llrip.fr
→ Learn more about TWENTY YEARS: https://twentyyears.ai