PRIOR ART RESEARCH
The SIPI team has been performing prior art searches for decades, mainly in support of litigation. We also perform prior art searches occasionally for our Technically Enhanced Prosecution (TEP) services. This has pushed us to find efficient and effective methods to perform this service.
Often we will receive a search from our clients prior to starting a project and are usually disappointed with the deliverable. Unfortunately, Prior Art Searching has become commoditized, usually performed by offshore entities racing to the bottom to beat each other on cost. This is a recipe for compromising quality in any marketplace.
As our clients reach out to us more and more to help them with "Litigation Grade" prior art searching it is obvious there is a growing need to increase quality while still maintaining competitiveness. With the growing functionality of LLMs, our software programming team can see a way to disrupt the market.
As AI permeates our culture, we are discovering how hazardous "enthusiastic hallucinations" can be. With any software tool that uses LLMs, trash in equals trash out. The real challenge is understanding what is trash. As we move toward Agentic AI, this will change and evolve, but there will likely always be a need for an "arbiter of truth". Our SMEs function as the arbiter of patent truth.
DIFFERENTIATION
Writing a software tool to use LLMs capable of searching a database of patents is relatively straightforward (generative AI can do this). The real challenge is searching non-patent literature. There are many different approaches, but at the core of the challenge is the "database" the LLM searches. This is one of our differentiators. We have engaged in thousands of RE projects and thousands of litigation-grade claim charts over the last couple of decades. This is a very powerful resource.
Another critical differentiator is who is generating the search terms for the LLM to process. We are blessed with some of the most experienced Subject Matter Experts on the planet. Every one of our team members have the licensing, claim interpretation, claim construction, and reverse engineering expertise to shape those LLM inputs in a way that only a highly experienced Subject Matter Expert can.
Lastly, our search reports focus on impactful data generation. Nobody has time to digest data that is not actionable. We know how to invalidate patents