In 2023 I sat in on a presentation by a well-known consulting firm pitching AI transformation to a mid-sized company. They showed 40 slides. They talked about "AI-readiness assessments," "change management," and "center-of-excellence models." They mentioned GPT exactly once, in passing. The engagement they were proposing was six months of strategy work before anyone would write a line of code.
The people in the room were smart. They knew something was wrong. But they didn't know enough about what AI could actually do to push back effectively, and the consultants knew enough jargon to sound authoritative.
I'd seen this pattern before, but this was the moment I decided to do something about it.
The actual problem
Large consulting firms sell strategy because strategy is scalable. You template a framework, train junior staff to apply it, and sell it to many clients at a high margin. The problem is that AI strategy without AI implementation is often worse than useless — it gives leadership the feeling of action without any of the learning that comes from building something real.
The companies that are winning with AI aren't the ones with the best AI strategy documents. They're the ones that built something, learned from it, and iterated. The learning curve is steep and you can only climb it by doing.
What I kept seeing was a gap between the advice being given and the ability to execute on it. Not because the companies were lazy or dumb, but because building AI systems requires a very specific combination of skills — ML, software engineering, product intuition, infrastructure knowledge — that large firms rarely concentrate in the people who talk to clients.
What Solvbit is
The answer I settled on was to stay small and senior. Solvbit is not a traditional consulting firm. We don't do strategy decks. We build systems.
A typical engagement: a company comes to us with a problem they think AI can help with. We spend a week or two scoping it — understanding the data they have, the infrastructure they're running on, the actual workflow that needs to change. We write a technical brief that describes what we'll build, how it'll work, and what it costs. Then we build it.
The people doing the scoping are the people doing the building. There's no handoff from "strategy" to "delivery." That matters more than it might seem — a lot of implementation failures happen at that handoff, when the people who understood the problem aren't the ones writing the code.
What I've learned
Running a small consulting firm is different from what I expected in a few ways.
Clients don't always know what they're asking for. "We want to use AI" often means "we want to save time on X" or "we're worried a competitor will do this before us." Figuring out what's actually being asked — and being honest when AI isn't the right answer — is most of the job.
Scope is everything. The engagements that go badly almost always involve scope that wasn't clearly defined. We've gotten strict about fixed-scope proposals. If the client wants to add things, we talk about a new engagement. This feels rigid but it's actually what lets you ship reliably.
The best marketing is a portfolio. Every client we have came from a referral or saw something we built. Writing posts like this one is part of the marketing, but nothing beats being able to show a working system and say "we built that in eight weeks."
Small is a feature, not a constraint. We've been offered larger engagements that would require hiring and expanding. We've turned them down. The quality we can deliver at our current size is higher than what we could deliver at 3× the headcount. That's the constraint we've chosen to optimize for.
Is it working?
Two years in: yes. We've shipped production systems, we have clients who've come back for second engagements, and I sleep fine at night knowing that the things we've built are running in the real world and doing what they're supposed to do.
The market for this kind of firm is big. Most companies need help building AI systems, and most of the firms that offer to help are either too big and process-heavy or too small and junior. There's a lot of room in between.
If you're thinking about a project, email me. I still respond to every inquiry personally.