In September 2023, I was preparing to shut down Colabra.
We'd started the company building software for scientific R&D. Nobel laureate Shinya Yamanaka's team was using it every day to run a lab in San Francisco from Japan. I had flown to Kyoto on a day's notice to win that contract and kept visiting the lab after we signed it. We'd built something people valued, but selling it was difficult. Small organizations had limited budgets, and larger ones could like the product without making buying it a priority.
When a private equity firm buying up R&D software companies approached us with an acquisition offer, we took the call. Those conversations introduced us to data rooms, where companies collect the documents involved in a deal. The software we'd built to organize research documents and manage access could also help teams reviewing an acquisition.
We decided to raise more money to pursue that pivot. Getting the financing nearly killed us.
After hundreds of investor pitches, we engaged a dissolution adviser. We told the team what was happening and prepared to help customers move their data and employees find new jobs. At the same time, we continued building and fundraising.
This is the period I think about when I hear founders talk about chewing glass. We had to leave enough time to wind down responsibly while continuing to ask investors to back the company.
In October, we found an investor. We were still waiting for the money when I wrote to the dissolution adviser:
Definitely sweated a bit there and a lot of crawling through the mud for an extended period of time, but hopefully this makes Colabra stronger in the long-run.
The financing came through. We went on to build and deploy AI software for private-equity customers that analyzes tens of thousands of financial and legal documents. The work we'd done for research labs gave us a foundation for a very different business.
We had to rethink the product again as AI improved. With Claude Code and then Cowork, the people I was talking to started expecting much more from software. Some wanted to build their own tools. Others wanted everything tailored to the way they worked, down to how information appeared in a presentation for a particular deal.
We kept adding dashboards to accommodate these requests, until Colabra itself was becoming harder to use. One user described it to me as starting with a data dump and ending with another data dump inside Colabra. That one really got to me, because I could see what they meant.
So we started moving toward a headless product. Colabra could process and analyze the documents, and people could work with the results in Claude or ChatGPT. We'd spent years building an application for people to use. Now the same work could help customers build tools of their own, tailored in ways we couldn't have anticipated.
I've kept Colabra moving forward by knocking on doors. I kept talking to investors and customers, flying to meet potential customers, and taking red-eyes to conferences. I built relationships with people across different industries who could make warm introductions to the companies we wanted as customers.
I also kept iterating on the product, rewriting our messaging, and trying different ways to reach customers. When our approach wasn't working, I changed the product or the pitch and went back out to sell again.
I'm proud of the product, the team, and my part in getting us here. We've built software that people use for work as consequential as running a research lab or evaluating an acquisition.
AI is improving quickly, and many people still haven't seen how much it could improve their work and lives. I'm excited to keep building Colabra to help more people put those capabilities to use.