How a Beer With a Patent Attorney Led to Compass

I didn’t set out to build a product. I was just having a beer with a buddy.

He’s a patent attorney, and he was griping about how he couldn’t use ChatGPT at work. I asked why not – seemed like the perfect use case for someone drowning in dense documents all day. His answer was simple: “I’m dealing with confidential information. I can’t just upload client IP into some third-party platform.”

Fair point.

So I told him I’d see what I could do. I went home, duct-taped together some open source software, and put it directly on his laptop. No cloud. No third-party anything. Just a local model he could throw documents at without worrying about where that data ended up.

He went away and used it for a few weeks. When he came back, the first thing he said was “this saves me eight hours a week.” The second thing he said was “we need this for my whole firm.”

Now, at Barefoot Solutions – my custom software development shop – that’s music to our ears. We scoped out a proper project for them. A locally hosted large language model, a RAG database so it could actually reference their documents intelligently, and a clean interface their attorneys could use without needing a CS degree. We got the budget approved. Nice project for us.

Then I Read About Goldman Sachs

Right around that time, I came across an article about Goldman rolling out GSAI – their internal AI platform. I did a little digging and thought, huh, that’s a lot like what we’re about to build for this law firm.

So I kept pulling the thread. JPMorgan was building one. BAE Systems was building one. Cisco was building one. Basically every Fortune 100 company was standing up internal AI platforms to boost the productivity of their knowledge workers. Same core idea: give people access to powerful AI, but keep the data inside the walls.

And that’s when it clicked.

Mid-market companies are going to want the exact same thing. They have the same confidentiality concerns, the same productivity gaps, the same knowledge workers spending hours on tasks that AI handles in minutes. But they don’t have Goldman’s internal engineering team. They don’t have a hundred-person AI division to build it from scratch.

Someone was going to have to build it for them.

I went back to the law firm and said, “I think there’s a product here. Internal AI platforms for mid-market companies in highly regulated industries – legal, finance, defense, healthcare. I want to build this on my own dime and license it back to you for much cheaper than the custom project we quoted.”

They heard the word “cheaper” and said great, let’s do that.

From One Firm to a Company

So that’s how Compass was born. We built it out, deployed it for the law firm, and watched their attorneys reclaim those eight hours a week across the board. Then we took it to market as Barefoot Labs – a separate company focused entirely on private AI platforms for organizations that can’t afford to be careless with their data.

The funny thing is, none of this started with a grand vision or a pitch deck. It started with a friend complaining over a beer, me cobbling something together on a weekend, and then paying attention when the Fortune 100 started validating the exact same idea at enterprise scale.

I think that’s how the best products get built, honestly. Not from market research or trend forecasting, but from solving a real problem for a real person and then realizing you’re not the only one with that problem.

The law firm still uses Compass. They were our first customer and they’re still one of our best references. And every time I grab a beer with my buddy, he reminds me I owe him a royalty.

I tell him the discount was his royalty.

If your organization is sitting on sensitive data and wondering how to give your team AI tools without the compliance headaches, we should talk. No pitch – just a conversation about what this could look like for you.

Share this article