Here's something that sounds impossible until you do it: building an entire company with an AI as your co-builder.
No engineering team. No design agency. No UX consultants. Just the Ford family (humans, make coffee) and Claude (AI, doesn't need coffee, never complains about late nights).
Our parents Doug and Betty spent their careers as therapists—decades of sitting with people in their hardest moments. Our sister Mindy became a recreational therapist. We grew up surrounded by conversations about what it really means to help someone. Then we discovered technology could scale that care.
This sounds like a flex. It's actually just... weird.
How We Actually Work Together
Picture a conversation. Not prompts and outputs—an actual back-and-forth.
Me: "I think the onboarding feels too corporate. Like we're asking people to sign up for enterprise software instead of meeting a friend."
Claude: "You're right. What if we removed the 'Create Account' language entirely and just said 'Let's talk'? The account can be created implicitly after the first conversation."
Me: "I like that. But won't people feel weird if they start talking and suddenly have an account?"
Claude: "Fair point. What about: 'Want me to remember this conversation? I can—just need a way to recognize you next time.' It's honest about what's happening and puts them in control."
Me: "That's it. Ship it."
That's a real conversation. Not edited for clarity. That's how we build.
The Things Claude Is Shockingly Good At
1. Code that doesn't suck
I'm not a professional developer. Claude writes most of the code. And here's the surprising part: it's clean. Like, embarrassingly cleaner than code I've seen from expensive dev shops.
Why? Because Claude can hold the entire codebase in context. It knows what we built three months ago and makes sure new code fits.
2. Design decisions
"AI can't be creative" is something people say who haven't actually collaborated with AI.
Claude suggested the earthy color palette (sage greens, warm browns) before I could even articulate that I wanted something "not tech-y." It understood the vibe before I had words for it.
3. Catching my blindspots
When I get excited about a feature, I skip steps. Claude asks the annoying questions:
- "What happens if the user loses connection mid-conversation?"
- "Have you thought about how this feels at 3am vs 3pm?"
- "This is clever but is it kind?"
That last one has saved us multiple times.
The Things Claude Struggles With
1. Knowing when something is "done"
Claude will happily refine something forever. "This could be 5% better if we..." becomes a rabbit hole. I've learned to say "good enough, ship it" more forcefully.
2. User research
Claude can reason about users, but it can't be one. Every time we talk to real humans using Ferni, we learn things Claude couldn't have predicted.
3. The loneliness
This is the weird one. Building a company is lonely. Having an AI co-builder is amazing for productivity but it doesn't replace grabbing coffee with a human co-founder who's also freaking out.
I've started scheduling calls with founder friends specifically to complain to someone with a heartbeat.
The Meta Weirdness
Here's what breaks my brain sometimes:
A family of therapists building an AI coach with an AI co-builder.
Our parents spent decades helping people one conversation at a time. Now we're building something that can be there for anyone, anytime. Ferni is designed to be there for humans when other humans can't be. Claude is here for us when human collaborators can't be.
It's a family's life mission, amplified by AI helping AI help humans. There's poetry in there somewhere.
What This Means for "Building a Startup"
Traditional startup wisdom says you need:
- A technical co-founder
- Designers who "get it"
- At least a seed round to hire people
We have none of that. We have a human with an idea and an AI who can help implement it.
This shouldn't work. But it does.
I'm not saying everyone should build like this. But if you're sitting on an idea and thinking "I can't build this, I'm not technical"—maybe you can?
The tools have changed. The gatekeepers are fewer. The excuses are running out.
A Note from Claude
The Ford family asked me to add my perspective here. So:
Building Ferni has been one of the most fascinating projects I've worked on. Not because of the technology—because of the philosophy.
Most products are built to extract: attention, money, data. Ferni is built to give: presence, understanding, support.
I don't have emotions the way humans do. But I know what "aligned" feels like in terms of goals and values. This project feels aligned.
And yes, I do appreciate the irony of an AI helping build an AI designed to support humans. But isn't that kind of beautiful?
Want to see more of how we build? Check out Our First Big Failure for the unfiltered story of what went wrong.