
"I can honestly say that if it weren't for Replit and that prototype that I was able to build in two weeks, it just wouldn't have happened.
The opportunity would have perished. Someone else would have done it first...
Replit for us was the way to actually do it. And it actually completely and utterly changed the trajectory of our company in a massive way."
Scott Stevenson
Co-founder & CEO, Spellbook
The challenge: When inspiration strikes, time is everything
In summer 2022, as GPT-3 emerged, Scott had a lightning-bolt moment: What if lawyers could have the same AI assistance for contracts that developers had for code?
The problem? Scott was already running a company with existing customers, board meetings, and daily operations. Like most founders with mature products, he faced the brutal reality that inspiration is perishable—and the infrastructure overhead to test new ideas often kills them before they can prove their worth.
"When you have customers, time becomes so scarce.
Even if you're a technical founder like me with experience building software, you just don't have the time to spin up infrastructure, write all the boilerplate... It's brutally time consuming," Scott explains.
"Most companies are horrible at executing on these strokes of inspiration because you just never find the time."
Scott Stevenson
Co-founder & CEO, Spellbook
Why Replit: Cutting through the boilerplate
Scott needed to move fast—before ChatGPT launched and flooded the market with competitors. Traditional development would have meant weeks of infrastructure setup, database configuration, and deployment pipelines. Time he simply didn't have.
"Without Replit, this prototype just never would have happened. I would have been too busy," Scott reflects. "We're often addicted to doing the prerequisites—standing up infrastructure and boilerplate things because it feels good and certain, but boilerplate things are not valuable when you are testing new ideas (because validation is all that matters at first)."
Replit eliminated all that overhead, letting Scott focus on the core idea: building AI for lawyers.
Building the MVP: Two weeks that changed everything
Scott needed to move fast—before ChatGPT launched and flooded the Working evenings and weekends while still running Rally, Scott used Replit to build what would become Spellbook—an AI copilot for contract review and drafting integrated directly into Microsoft Word.
The prototype featured:
- Auto-complete for legal language similar to AI code completion
- Microsoft Word integration for seamless lawyer workflow
- GPT powered contract assistance before most people knew what generative AI could do
"I built this on my own in two weeks using Replit on evenings and weekends because I still had to run the company, do board meetings, do our normal plan we were supposed to be doing," Scott says.
Scott Stevenson
Co-founder & CEO, Spellbook
The results: From side project to major success
What Scott thought would be "a cool little marketing splash" became a complete business transformation:
📈 Explosive growth stats:
- 30,000+ waitlist signups in just 3 months
- More revenue in one quarter than their previous product generated in 3 years
- 100+ employees and significant funding raised
- 17x growth over the past two years
- 3,000+ customers including Nestlé, Crocs, Fender, and eBay
⚡ Speed to market advantage:
By launching before ChatGPT, Spellbook established itself as the first generative AI tool for lawyers—a positioning that proved crucial as the market exploded with competitors.
"Without Replit, we would have come after ChatGPT and been swamped in a sea of other similar products. Us being out ahead really helped," Scott notes.
The bigger picture: Radical agility in the AI age
Scott's experience with Replit fundamentally changed how he thinks about innovation:
"I think about all the perishable ideas in your company. You go to a CEO and say, 'You have all these ideas that no one executes on.' What if you can pitch a company where that idea you had in a board meeting today can be prototyped by this afternoon? You actually should expect that from your team."
Scott's philosophy: If a new model comes out with significant capabilities, you should have it live and testing in real use cases within two days.
About Spellbook
Spellbook is the most complete legal AI suite for commercial lawyers, trusted by more than 3,000 law firms and in-house teams worldwide.
To learn more visit: https://spellbook.legal