Idea: Platform for Patent Search

NOTE: This is still very much a draft of the one-pager.

Idea

Create either a better platform for patent search or a middleware API for patent data

Core Issue

Tech and R&D teams spend a lot of time and money developing new products. However, if they don’t have the appropriate patent/the patent has already been made, then they will have wasted — or will waste — a bunch of dollars. In any case, the issue of searching patents to find one that is relevant to your potential intellectual property is a big problem.

Key Issues

• Patent data is non-standardized: there do exist some databases such as from the US PTO, but these are still hard to use, and most platforms (that are open access) still use a relational database and a basic SQL query
• Searching patents can be expensive: there are some existing players out there — albeit, I don’t think any of them are that big — but a lot of firms actually still go to patent search companies. These companies use a mix of automation + manual/expert due diligence in order to find the patents that are most relevant to a company’s IP.

Solution

There’s a couple of ideas here, but overall, I think there’s something here.

• Stripe for Patents: at the most basic level, given potential customers like patent search firms and companies building AI-powered patent search, we could just build the middleware for patents. This could involve scraping all of the existing patents from U.S. companies, cleaning them ourselves/automating that process, and running an API on a server that users can query.
• The Patent Search platform itself: in a more complex level, build all of the infra. This might be a crowded space, but I also don’t think it’s like AI/financial technology/SaaS, where there has been a winner take all (COULD BE VERY WRONG). At the very least, I also think there are a ton of interesting problems:
  • HCI related to building a good search interface
  • Data mining problems related to performing optimal search (do we give natural language queries? Do we use sliders and others? Etc.)
  • Systems level problems related to surfacing results (how can we provide in-depth, comprehensive insights in as low latency as possible?)

Right to Win:

• I think this is one of the first places where the idea is specific and the market is huge. I also think it doesn’t suffer from an industry like machine learning right now where there’s a lot of hype around LLMs and ML infra but not too much signal on what the winners will be (i.e., there will have to be a graveyard of startups somewhere).
• We have the skillset to build this product. Yes, I think that the law space will be tricky — will have to think about selling into this space — but it seems like there is traction on this front and we could out execute. We clearly have the necessary technical skills, from the HCI knowledge, to the ML knowledge, to the core systems knowledge to make shit run fast.
  • I also think we will enjoy these problems. It’s not just doing full-stack stuff: we also get to be in the nitty gritty of solving a real-life problem but with cool tech and algorithms. -My dad works in the IP tech space, and can connect us with people. We can also get advisors on the tech side (i.e., SAIL people?) to folks on the law side (i.e., SLS?). Either way, I think we can leverage the Silicon Valley connections.

Unknowns:

• What is the state of the law space right now? There might be a chance that this could be a multiple winners take all type of market — lots of law firms might have their own preference, and it’s more of a sales game. Nonetheless, we’d have superior product + eng quality, esp. in velocity (i.e., what fucking Citadel or Motional dev is gonna work on law tech?).
• Who do we sell to? In other words: what do we build? There’s clearly some low hanging fruit and some other cool opportunities here, but we want to think about where in this process the value accrues. Are AI-powered patent search platforms already big that we can actually do more of the data cleaning step? What about patent search firms — can they benefit from our tooling?