On the TechTO stage ×1
First seen on a TechTO stage in 2026. Every TechTO talk is searchable — ask the archive about Matt ↗
In their words
After what we called sort of drunkenly stumbling through the jungle trying to find a recipe that worked, we were inspired by GitHub Copilot at the time.
Investors have different incentives. They have different business models and different ways of winning. So incentives aren't always aligned when valuations are maxed out, and it can create an environment where it's actually harder to succeed.
We did that until we knew very deeply that we had product market fit. And if you think you might have product market fit, then you don't.
Quick answers
What does Spellbook do?
Spellbook makes AI contract workflow software for lawyers at both law firms and in-house counsel. Co-founder Scott Stevenson's insight, before ChatGPT launched, was that large language models could give lawyers what GitHub Copilot's autocomplete gave engineers. Inovia Capital led the company's Series A after an earlier, smaller follow-on check.
How did Spellbook grow out of Rally?
For about five years the company operated as Rally, a legal-tech product with decent metrics but never deep product-market fit. In summer 2022 Scott Stevenson coded a Spellbook prototype on nights and weekends as a lead-magnet side project. Customers reacted more strongly than to anything the team had shipped before, so over roughly three months they transitioned the entire operation and raised their next round on that success.
How fast does he think founders should experiment before product-market fit?
As quickly as possible: design each experiment around the least amount of work and time required to get data, since an experiment that takes a year to prove out is probably the wrong one. He credits much of Spellbook's success to the tolerance for rapid experimentation the team built during five years of constant pivots at Rally.