What Is Solve Intelligence? AI Patent Drafting Software

What is Solve Intelligence?

Solve Intelligence is a company that makes AI software for patent attorneys. It was founded in 2023 by Chris (CEO), Angus (CTO) and Sange (chief research officer), and went through Y Combinator (YC) in the summer 2023 batch. According to the founders, more than 700 intellectual property (IP) teams now use it, including law firms and in-house teams at companies like Intel and Siemens. They report $55 million raised in total. Its main focus is helping attorneys draft patent applications.

Who started Solve Intelligence, and how?

None of the founders came from a law background. Chris was finishing a PhD in AI at UCL and had a few months free. He started a side project with his brother Angus, who worked at a quantitative trading firm. They had both been on the inventor side of patents, which means they knew how much time and money the process takes. They also knew there was little software for it. Large language models (LLMs, the technology behind ChatGPT) had just taken a big step forward in 2023, and that made the idea feel timely.

Chris had planned to take a job doing AI research in Singapore. Then he talked to Sange, who was going all in on AI and had also worked in quant finance. Chris was asking for career advice. He ended up showing Sange what he and Angus were building, and the two teams began working together at Sange's apartment the next day.

The YC application came almost by accident. Sange noticed the deadline was about a week away. They figured they might as well fill in the form. Answering the questions pushed them to talk to real patent attorneys for the first time. That led them to pivot the idea, staying in patents but changing what they built. By the end of the week, Chris says, he was convinced enough to turn down the Singapore job. Angus left his quant job. They described the application as a long shot, and they got in.

The founders say they poached the name, which is a nod to DeepMind. Sange liked the idea of an ambitious name pointing at fundamental problems, not a narrow, micro one.

What does Solve Intelligence actually do?

The interview centers on one job: helping attorneys draft patent applications. A patent application is a long technical document. It describes an invention in detail and includes claims, which are the precise legal statements defining what the patent protects.

The founders also describe the difficulty that shaped the product. Patent work involves multimodal inputs and outputs. That means more than text: chemical structures, technical drawings and similar material. Early customers told them AI would never handle this. The founders took that as a challenge, not a verdict.

Patent practice includes other tasks beyond drafting. Attorneys respond to rejections from patent examiners (called office actions) and compare patent claims against products or prior art (often done in claim charts). The interview excerpt doesn't walk through those features, so check Solve Intelligence's own site for its current feature list before assuming what it covers.

Who uses it, and how big is the company?

Per the founders, the user base includes:

• Large law firms that handle patent work for clients.
• In-house teams at large companies, with Intel and Siemens named.
• More than 700 IP teams in total.

On funding, they say they have raised $55 million in total. They put the market at over $300 billion spent every year across the patent process, with very little technology in use. Treat that as the founders' own estimate. It explains why they saw a large, underserved market.

How did it grow so fast?

Several choices stand out, and they apply well beyond patents.

They launched before the product existed. In the first week or two of YC, Chris proposed launching the next day. They had nothing to ship, but announcing it forced them to build, host and sell something. They made their first sale that week, then spent the weekend getting the product running.

They charged from day one. There were no free trials. Chris says that because they knew nothing about patents, they needed a real signal, and paying customers gave a clearer one than free users. They tracked growth and churn (customers leaving) monthly. As churn fell, customers expanded their use, spoke well of the product and sent inbound leads.

They set public goals. YC group office hours had founders state a revenue target for the next two weeks. Missing it in front of peers was a useful pressure.

They sold without a salesperson. None of the founders had sales experience. They say they reached several million dollars in annual revenue without hiring one. In hindsight, Chris says they should have hired earlier and might have grown faster.

They had no assumptions. Not knowing how patent attorneys should work meant they had to listen to customers.

Why can't a lawyer just use ChatGPT for patents?

The interviewer asked this directly, but the transcript provided cuts off before the answer. So here is what the founders did say, plus well-established general reasons.

From the founders: patent drafting involves multimodal material such as chemical structures and technical drawings. A chat window with a text box is a poor fit for that.

General reasons patent work resists a generic chatbot:

• Precision matters. Claim wording defines legal scope. One loosely chosen word can narrow or widen protection.
• Consistency across a long document. The claims, description and drawings must agree with each other.
• Confidentiality. An invention generally must not be publicly disclosed before filing, so how a tool handles data is a serious question for any firm.
• Workflow. Attorneys work in drafts, edits and reviews. They need a tool that fits that process, not a one-off answer to paste in.
• Accountability. A licensed attorney remains responsible for the filing, so output has to be easy to check.

None of this means a general chatbot is useless for patents. It means a purpose-built product can address these specifics, which is the bet Solve Intelligence is making.

What does this mean for you?

If you work in a law firm or IP department: AI drafting tools are now in use at large firms and corporate teams. Evaluate them on your own documents. Ask about data handling, how outputs are reviewed, and whether the tool handles drawings and technical content in your field.

If you're an inventor, engineer or researcher: your attorney may be using tools like this, which can change how they ask you for input. Clear invention descriptions help whichever tool is used.

If you work in any document-heavy profession: the pattern is the useful lesson. A specialized product, built with deep customer feedback, got traction where general tools fell short. Ask where your own work has high-precision, repeatable documents.

If you're thinking about starting something: the founders applied with a two-week-old idea, talked to customers during the application, and charged early.

Could you build a small invention-tracking tool of your own?

Most teams around a patent process never touch drafting. They handle the surrounding admin: collecting invention disclosures from engineers, tracking who has reviewed what, and watching deadlines. A shared spreadsheet usually does that badly. With Remy, you can describe an invention disclosure form and a status tracker in plain English, and it builds and deploys a full-stack app with sign-in and a database. Engineers submit ideas through the form, and reviewers see each submission's status and owner. It would not draft patents, but it can replace the spreadsheet that surrounds that work.