Solve Intelligence

A patent attorney opens a browser tab. On one side sits an invention disclosure - a tangle of diagrams, half-sentences and an inventor's enthusiasm. On the other sits a blank specification that, by tradition, will take days to fill. Between them now sits Solve Intelligence, and the days have quietly become hours.

That is the scene in 2026 inside more than 400 intellectual-property teams - law firms and corporate departments scattered across six continents. They draft, respond to office actions, harvest inventions, and chart claims inside a single editor that looks deceptively like a word processor and behaves like a very well-read junior associate who never sleeps. The company behind it is three years old, profitable, and was last valued enough to pull $40 million out of investors in a single December round.

Above: a job that used to require a legal pad, a fresh pot of coffee, and a clear afternoon. Now it requires a tab.

The Problem They Saw

A $200 billion industry that runs on prose

Patents are strange documents. They are simultaneously legal contracts, technical manuals and strategic weapons - and every word can be litigated for a decade. Drafting one is slow, expensive, and bottlenecked by the small population of people qualified to do it. The global IP services market is worth north of $200 billion, and an embarrassing share of that value is locked up in the time it takes a highly trained attorney to type.

For years the obvious fix - automation - kept running into an obvious wall. Patent prose is unforgiving. A clumsy phrase can narrow a claim, hand a competitor a loophole, or get an application rejected. The earliest generative tools were happy to write confident nonsense, which in patent law is the most expensive kind. The attorneys, sensibly, did not trust them.

So the bottleneck stayed where it was. In-house teams watched patentable inventions slip by because nobody had the hours to harvest them. Litigators built claim charts by hand, cell by cell, the night before a deadline. The work was important, repetitive, and resistant to every shortcut anyone had tried. Which is a polite way of saying it was waiting for someone to take it seriously.

The Founders' Bet

Two brothers, three PhDs' worth of conviction

Solve Intelligence was founded in 2023 by brothers Chris and Angus Parsonson, alongside Sanj Ahilan. The bet was specific: that the way to win the trust of patent attorneys was not to replace them, but to build the tool with them - and to put serious machine-learning research underneath it rather than a thin wrapper over someone else's model.

Chris Parsonson did a machine-learning PhD at UCL and had passed through the Alan Turing Institute, InstaDeep and Dyson before deciding patents were the problem worth his decade. Ahilan, also a UCL machine-learning PhD, had written popular open-source code for fine-tuning large language models. The trio's instinct was to seat AI researchers next to practising patent attorneys and not let either group design in a vacuum.

A founding team where "we'll ask our patent attorney" means walking ten feet, not booking a consultant.

Y Combinator backed them in its summer 2023 batch with a $3 million seed. The pitch was unglamorous and exactly right: don't demo a chatbot, demo a faster Tuesday for a patent attorney. Investors who had watched a hundred "AI for X" decks noticed that this one had customers who paid and stayed.

The Product

An editor on the outside, research on the inside

What a user sees is calm: an in-browser document editor any attorney can open and use on day one, no migration project required. What sits behind it covers the whole patent lifecycle - invention harvesting, drafting, prosecution and office-action responses - with models that can be customised for the awkward realities of software, biotech and chemical inventions, including biological sequences and chemical structures.

In December 2025 the company stepped beyond the drafting desk and into the courtroom with Charts, a product aimed at litigation and high-volume analysis. It generates the customisable claim charts that litigators previously assembled by hand: invalidity and standard-essential-patent charts, freedom-to-operate and clearance reviews, infringement mappings, claim-construction analyses and portfolio-level comparisons. It was the moment Solve stopped being a writing tool and became an IP platform.

The Proof

Customers, cash, and the receipts

The numbers are the argument. Roughly 60% of customers are law firms - including DLA Piper, Perkins Coie and Banner Witcoff - and 40% are corporate IP departments such as Siemens and Avery Dennison. Annual recurring revenue grew more than tenfold in a year to eight figures, and the company says it is profitable, having generated more cash than it has ever raised. That last detail is rare enough in AI to deserve italics.

Strategic backers tell their own story. Thomson Reuters - which sells into the same legal market - increased its stake across both the Series A and Series B rather than treating Solve as a curiosity. Microsoft backed the Series A. The Series B was co-led by Visionaries and 20VC, with a cap table that reportedly includes angels behind Tinder, Canva, Deel, Ironclad and Hugging Face. When the incumbent data company keeps buying more, the upstart is doing something the incumbent can't.

A customer list where the law firms and the companies they advise both showed up. Awkward at a conference; excellent for a balance sheet.

The Mission

Less typing, more judgment

Solve's stated aim is to build an AI-native platform that accelerates every stage of the patent lifecycle for in-house counsel, law firms and inventors worldwide. The quieter ambition underneath it is a redefinition of the job. If software handles the laborious drafting, the attorney's value moves up the stack - to deciding what is worth patenting, how broadly, and how to turn a portfolio into leverage.

It is a comfortable story to tell a profession nervous about automation, and Solve has the unusual standing to tell it, because the people building the tool are partly the people it is built for. Attorneys report 60-90% time savings on their workflows - the kind of figure that, if even half true, reshapes how a firm staffs a matter.

Why It Matters Tomorrow

The shape of the next decade of IP

Patents protect the inventions that define each era - and the volume of inventions worth protecting is only climbing, especially in AI, biotech and chemistry, the very fields with the most punishing prose. If drafting and analysis get an order of magnitude cheaper and faster, more ideas get protected, smaller players can afford to play, and the strategic centre of gravity in IP shifts from who can type to who can think. That is the world Solve is betting on, and increasingly funding from its own revenue.

Back to the open tab. The attorney who once faced a blank specification and a lost afternoon now reviews a draft, sharpens the claims, and spends the saved hours on the question that actually pays: is this invention worth defending, and how hard. The blank page didn't disappear. It just stopped being the bottleneck. For an industry that has run on prose for two centuries, that is the part worth watching.