We recently announced the acquisition of Casetext by Thomson Reuters for $650M in cash, pending regulatory approval. 

I had the pleasure of being interviewed by a member of the Thomson Reuters family, Krystal Hu of Reuters, at Collision Conference in Toronto the day after the deal was announced. You can watch the full interview here.

While there’s no shortage of news about shiny investments in AI, I thought it would be useful to go behind the scenes of building (and exiting) a great AI company—and how we at Canvas work with our portfolio companies and CEOs every step along the way. Companies like Casetext are not a quick build. While GPT4 was released less than four months ago, Casetext has spent a decade gathering a unique dataset, understanding customer use cases, and building credibility in the highly skeptical legal market. 

Below are six of my takeaways from working with Casetext over the past six years, starting with leading the Series B in 2017.

1) The best investments emerge out of a strong thesis

Coming out of law school at Berkeley, I saw firsthand that the $6B legal research industry needed to be disrupted. Legal cases are in the public domain and free, and yet law firms paid millions of dollars to search them. The search query structure was Byzantine. Tolls were charged for each search, and the software itself was not intelligent—it did not help lawyers avoid mistakes, nor did it use prior searches or knowledge to improve future results. The best it had to offer was Shepardization, which despite being a critical component of any legal research platform, simply validates the current judicial posture on each key legal point in a case. In mapping the legal research space, we saw that the landscape was limited: companies were focused on lead-gen, content management or narrow segments such as patents. Casetext stood out because of its early understanding that AI would be the game-changing technology.

2) My favorite founders have untraditional backgrounds and a chip on their shoulder

Casetext’s CEO Jake Heller is the type of "spikey" founder I like. Jake is both a lawyer and developer. His parents didn't graduate from college—and his dad didn’t graduate from high school for that matter. Jake didn't have plans for college initially, but he excelled at debate in high school. He received a scholarship to Pitzer, graduated with honors, and went on to Stanford Law School. Jake is entirely self-made. 

I remember asking Jake what his vision was for the company. When I shared my concern, I remember the look of astonishment on his face. “You don't understand,” Jake responded. “It isn’t like I have tons of ideas for startups. I am a lawyer. This is my One Thing. It has to be big.”

3) A prepared mind leads to fortuitous meetings

When we first met Casetext, they had no paying customers. We set out to help them change that. Our focus at Canvas is helping companies with their go-to-market strategy. Often that means introducing them to the right person who can help them. Before we’d even invested in Casetext, and while I was on a family vacation in Kauai, I met Jeroen Plink as I was chasing after my children at a resort. In between applying sunscreen on the kids, I learned that Jeroen had led the sales efforts at Practical Law Group, a legal company that sold to Thomson Reuters. I thought he'd be the perfect person to help Casetext. I introduced Jeroen to Jake, and they hit it off. Jeroen joined the board as an independent board member, and he set to work helping the company commercialize their product. Just a few months later, I heard from Jeroen that the company now had three law firms as paying customers.

4) To truly understand a company’s viability, you have to talk to customers directly

My background is in engineering and product, and I learned early in my tenure at Procter & Gamble that my personal opinion does not matter. It’s far more important what the customer says. For me, diligence means personally getting on the phone to conduct consumer interviews. As with most early products, Casetext’s customers—law firms in this case—liked but did not love the product. But law firms were willing to nibble, and I believed they would eventually bite. We moved forward and led Casetext’s Series B in February 2017. 

5) Without customers, AI is just a shiny new technology. With customers, AI can supercharge sales

Casetext launched its CoCounsel product this March with GPT4. They had been in the sandbox with OpenAI building together for months. To underscore the step change in AI from GPT3.5 to GPT4, Casetext sponsored a study to prove that GPT4 could pass the bar exam. While GPT3.5 did not pass the bar, GPT4 scored in the 90th percentile. This isn’t just a passing score—GPT4 did better than most associates! CoCounsel was ready for primetime. 

Prior to GPT4 and the launch of CoCounsel, the company was on a path to profitability. By CoCounsel's launch, every large law firm already knew who Casetext was. The company had 10,000 paying customers—a constituency of law firms and companies that believed in them and were primed to want more. 

In the first 45 days of CoCounsel’s launch, Casetext added another 1,000 customers from their self-serve product alone. Suddenly Casetext’s enterprise contracts had another “zero” behind them. Fortune100 companies opened up. By midyear, they’d nearly doubled their revenue from the start of 2023. The growth-round term sheets came in.

6) Engage a banker well before you are even considering a sale

I always encourage companies to have a banker by their side who is an expert in the particular sector and who gets to know the company over time. Long before a company sells, they assist with strategic introductions, and, if a liquidity event materializes, they hit the ground running. Months ago, we decided to bring in a banker who was deep in the legal sector and knew all the key potential acquirers. The banker in this case knew Thomson Reuters well, and helped Casetext to negotiate and perfect the deal so it was a win for all parties—investors, executives, and employees. In addition to providing content Casetext wouldn’t otherwise have access to, Thomson Reuters could sell the product into every law firm in the country.

What is next? Casetext was the perfect poster child for GPT4. We see Casetext continuing to revolutionize the world of legal technology. Attorneys will need to embrace the world of AI, or risk being consumed by it.

In the meantime, I’m off searching for my next deal. If you’re a company that’s building in an industry where AI can supercharge your product, please don’t hesitate to reach out!