Case Studies

Automated Patent Analysis

Technologies

LLM
NLP
AI

Client’s Need

Searching for patents and analysing their relevance to the subject under investigation is a time-consuming and frustrating activity for trademark and patent specialists. Some documents purposefully avoid using relevant terminology to remain hidden from keywords-based searches. On the other hand, relaxing the search criteria often returns too many documents to be analysed timely and accurately.
Aim: Quickly provide experts with a high-quality shortlist of documents and a summary of their content.

Approach

Combine semantic search with automatic summarisation for a fully fledged RAG-AI system. Using a dedicated LLM trained for specific subject areas allows to limit the CAPEX and OPEX of a proprietary tool while maintaining control over the behaviour of the AI and the input documentary sources.

Main Outcomes

  • The tool is still under development and allows users to set the breadth and depth of their search.
  • Searches can be refined interactively with follow-up questions.
  • The solution can run on customer’s hardware, while ensuring complete confidentiality of search topic and retrieved information.