Chekable: Generative AI SaaS for patent professional

Chekable builds prior art search and generative patent drafting software for patent professionals. The Advent of LLMs provided a novel way for patent practitioners to search for nuanced claims in prior art and leverage generative properties to reduce mechanical writing jobs within a patent application. Out-of-the-box LLMs performed poorly while drafting legally accurate patent text therefore a specially configured agentic AI app is required to draft a patent application as per best practice in the industry.

Customer Discovery

• Number of participants: 12 patent professionals

• Customer segment: Big Law, in-house, mid-size practice, and small/solo practice

• Key insights

  • Prior Art Search requires a special skill set and professionals use at least 3 search tools and are open to any other new tool that promises better search results

  • Patent professionals are highly concerned about sensitive intellectual property data being used for training any models and how such training could be guaranteed not to be performed on their data.

  • Some wanted claims never to be written with an AI and some were open to ideating with AI on claims. They needed a way for the system to use manually written claims.

  • For drafting, professionals did not care about most of the sections. They only cared about the detailed description specification that needs to be referred to with drawings.

  • Patents are written differently for each subject matter such as S/W, mechanical devices, drugs, biotech, etc. They will need different LLM rules to cover each subject.

Wireframes and conceptual exploration

Design system and branding - Atomic level design comPonents

 Product design

Usability testing and product analytics