By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

How Solve Intelligence Is Building the Future of AI Patent Law

Solve Intelligence (S23) builds AI software for patent attorneys. More than 700 IP teams use it, including firms like Perkins Coie and DLA Piper and in-house teams at Intel and Siemens.

Diana HuhostChrisguestSanjguest

Nov 10, 202631mWatch on YouTube

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

Great conversations run long. Their best ideas don't have to.

uListen is free on iPhone, iPad, Android, Chrome, and Claude.

CHAPTERS

Jump to publisher timestamps

Chapter summaries

From PhD side project to Solve Intelligence’s founding team
Diana Hu introduces Chris, Angus, and Sanj and frames Solve Intelligence as AI software for patent attorneys used by hundreds of IP teams. Chris recounts finishing his AI PhD early, starting a side project with Angus, and how his exposure to patents came from being on the inventor side and from his girlfriend being a patent attorney.
- •Diana Hu introduces the founders’ roles and what Solve Intelligence builds for patent workflows
- •Chris describes starting as a side project after finishing his AI PhD early
- •Early familiarity came from experiencing patent processes as inventors and seeing pain points
- •A personal connection to patent law (Chris’s girlfriend being a patent attorney) helped reveal the software gap
A two-week-old idea: applying to YC and pivoting through customer calls
The founders describe how they applied to YC on a whim when the idea was extremely new, expecting low odds. Chris explains that the YC application process forced them to talk to patent attorneys for the first time, clarify what they were building, and pivot within the patent space before submitting.
- •Chris says they thought their chances of getting into YC were very small
- •Sanj urged the team to notice the YC deadline and apply quickly
- •Filling out the YC application pushed them to talk to patent attorneys and refine the idea
- •Chris turned down a planned AI research job offer after becoming convinced the startup could be big
Leaving quant jobs and choosing ambition over salary
Diana asks about taking the leap from quant roles. One founder reflects that salary was the main draw of the quant job, while starting a company felt more satisfying due to direct user feedback; Sanj adds his motivation to build something ambitious and the importance of acting quickly when you find strong co-founders.
- •A founder says the quant job was mostly about salary and that building a startup is more satisfying
- •They describe being longtime YC fans and credit Sanj with pushing them to apply
- •Sanj cites inspiration from stories of DeepMind emerging from a PhD lab and wanting to ‘dream big’
- •Sanj emphasizes not hesitating when you find co-founders you’re impressed by
Early revenue and the “practical + deep tech” balance
Diana comments on Solve’s scale relative to DeepMind at the time it was sold, prompting reflection from a speaker on how surreal the growth feels. A speaker highlights a blend of deep thinkers and practical, customer-focused execution, arguing that getting revenue early created freedom and stability.
- •A speaker calls the company’s growth ‘surreal’
- •A speaker describes balancing deep technical thinking with practical customer focus
- •A speaker says early revenue gave the company freedom and independence
- •A speaker contrasts this with the challenge of lacking revenue and needing more ‘options’
Why patents: technical complexity, greenfield software, and a big opportunity
Diana asks why they chose patents despite not coming from law backgrounds. Chris explains patents are technical, have large spend across the process, and had little technology adoption, making them a compelling target after improved model capabilities; the team also notes their lack of preconceptions forced them to learn directly from customers.
- •Chris says model performance improvements made previously-unsolvable applications feel possible
- •Chris argues patents are deeply technical and a big industry with little tech adoption
- •Chris cites the combination of greenfield space, demand, and opportunity
- •A founder says having no legal background removed assumptions and forced customer conversations
- •Chris recalls skepticism that AI could handle multimodal patent work, and says their naivety helped them push forward
San Francisco and YC forcing functions: launching fast and learning through churn
Diana probes how moving to San Francisco during YC increased their speed, including getting a paid customer quickly. Chris describes SF’s ambient ambition and YC’s push to launch early; he shares how an aggressive launch deadline forced them to ship, make a first sale, and then learn rapidly when the customer churned.
- •Chris describes SF’s tech-saturated environment and higher ambition compared with London
- •Chris says YC guidance to launch early drove them to ship before they felt ready
- •Chris tells the story of deciding to launch despite ‘no product’ and using it as a forcing function
- •They made a first sale quickly, then rushed to operationalize the product
- •They lost that early customer, but used the feedback to improve
Reaching millions in revenue without salespeople: paid-only feedback loops
Diana asks how they grew to several million in annual revenue without hiring sales. Chris explains none of them had sales experience, but they launched early, avoided free trials, and relied on paying customers as the strongest feedback signal—tracking growth and churn until expansion and inbound compounded.
- •Chris says none of the founders had sold anything before starting the company
- •They chose not to offer free trials, using paying customers as the key signal
- •Chris describes tracking growth and churn and watching churn decline over time
- •As churn improved, customer expansion and inbound interest increased
- •Chris says they likely should have hired earlier to grow faster, but it wasn’t required to reach millions in revenue
Why not just use ChatGPT: domain depth, long feedback cycles, and execution rigor
Diana asks why lawyers wouldn’t just use ChatGPT. Sanj argues patents require deep technical understanding and have delayed ‘reward signals’ (like acceptance or litigation outcomes), and he describes their approach as deeply understanding customer needs, scaling evaluation and data, and operating with strong efficiency.
- •Sanj says writing patents requires deep understanding of the underlying technology
- •Sanj highlights long feedback loops: outcomes can be far in the future
- •Sanj emphasizes customers specifying needed outputs as a key advantage in building the product
- •Sanj discusses scaling evaluation and data sources to make the system work reliably
- •Sanj describes focusing on efficiency and correctness for customers seeking valuable patents
What makes patent workflows technically distinct: multimodal inputs and specialized UX
Chris explains that patents differ from typical ‘legal tech’ like contracts or M&A because they are deeply technical and often multimodal. He outlines the variety of data types involved and notes that patent workflows require customized workflows and UX beyond generic chat tools and even beyond general legal products.
- •Chris distinguishes patents from other legal tech areas like contract review or M&A
- •He lists multimodal inputs/outputs involved: structures, drawings, sequences, video, equations, tabular data
- •Chris names patent-specific workflows they support, including drafting and portfolio analysis
- •He notes integrating many sources requires specialized UX and workflow ‘harnesses’
- •Chris argues this specialization makes the niche defensible versus generic chat interfaces
Hypergrowth and the ‘next model’ mindset: excitement, agentic UX shifts, and stickiness
Diana highlights their rapid headcount growth and multiple offices, asking what the ride has felt like. Chris describes the moment as uniquely fast-moving and shares a heuristic: feel excited—not afraid—about new model releases; the founders discuss evolving from simpler interactions to more agentic systems and increasing product stickiness via organizational adoption.
- •Chris says the current period feels uniquely accelerated and creates pressure to move fast
- •Chris shares a heuristic: excitement vs fear about upcoming model releases
- •Chris says they use multiple model families and see ‘jagged’ performance across tasks
- •A founder says AI products require reworking UX frequently as capabilities change, shifting toward agentic behavior
- •Sanj emphasizes organizational sharing and deployments that make the product ‘sticky,’ while noting confidentiality considerations
Where to build: SF ambition, London talent, and staying close to customers
Diana asks how to navigate location decisions given visas, customers, and hiring. Chris recommends spending time in San Francisco for its ambition but argues a generational company can be built in London, citing advantages in accessing strong technical talent—while also stressing the need to travel frequently to meet customers in key markets.
- •Chris recommends going to San Francisco to experience its ambition and meet talented builders
- •Chris says you don’t have to stay in SF to build a major company
- •He argues London offers untapped technical talent, though local comparison can lower standards
- •Chris notes the US is their largest market and main revenue source despite global selling
- •He emphasizes frequent travel and in-person customer engagement as part of the job
Advice to a younger self: avoid the ‘delayed life plan’ and build useful things
Diana asks each founder what advice they’d give their younger self, especially for people in big labs, big tech, or quant. Chris contrasts academic incentives with the startup forcing function of revenue; another founder warns against delaying entrepreneurship due to comfort; Sanj adds that joining an early-stage startup can be a valuable stepping stone.
- •Chris says much academic research isn’t useful and can be driven by benchmark chasing
- •Chris values startups because revenue forces you to build something people will pay for
- •Chris urges going all in sooner rather than following a ‘delayed life plan’
- •A founder echoes that good salary can trap you into delay if your dream is to start a company
- •Sanj says joining an early-stage startup can teach you quickly and lead to founding later
Audience Q&A: learning sales, landing the first customer, and correcting the ICP
In Q&A, Chris reframes sales as curiosity-driven discovery and clear demos rather than aggressive pitching. He then describes their first paid customer—a solo inventor who paid, received a refund, and taught them to focus on selling to lawyers—followed by a repeatable outbound approach that brought their first lawyer customer who stayed.
- •Chris says much of sales is asking questions to understand workflows and pain points
- •He describes building, then presenting the output clearly so users can evaluate it
- •Chris explains developing a repeatable process once you learn common customer questions
- •Their first sale came from a LinkedIn post and a solo inventor who didn’t want a lawyer
- •They refunded that inventor and learned to target lawyers; Chris outlines LinkedIn outbound leading to a lawyer sale that didn’t churn
Biggest mistake and early-startup ownership: hire earlier and take responsibility
Chris answers a question about setbacks by describing constant new challenges and naming a core regret: not scaling hiring earlier despite being lean and reaching profitability. Sanj answers how to learn while working at a startup by embracing chaos, learning the whole business, and taking “ridiculous ownership” to drive outcomes like closing customers and improving product.
- •Chris says every stage brings new, unfamiliar challenges without prior pattern recognition
- •He cites not hiring and scaling the team earlier as a key mistake and missed acceleration
- •Chris emphasizes that after co-founders, the most important factor is the talent you bring in
- •Sanj describes enjoying startup chaos and learning by shipping and seeing the whole business
- •Sanj urges taking strong ownership—questioning why customers aren’t closing and pushing product improvements

Publisher timestamps

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.