For companies with no engineering function
AI put custom software in reach. It didn’t supply the engineer.
Some clients come here to learn the tools and build their own site alongside someone experienced. Others arrive with something they already built with AI — it works, people started using it, and now real work runs through it. The missing piece is the same in both cases: somebody who has run production systems against real data and can tell you what will hold and what won't. That judgment is the expensive thing to get wrong twice.
What changed
The software got cheap. The judgment didn’t.
For decades custom software was expensive because translation was expensive. You understood your business perfectly but couldn’t code, so you paid people who could code but didn’t understand your business. That translation layer has collapsed — which is why capable software is now appearing inside companies that never had a development budget.
What has not changed is everything that happens after it works. Whether the data model survives contact with real volume. Whether an AI-written query left your customer records readable to anyone who asks. Where it runs, who gets paged when it stops, and what happens the day you need to change it and can’t remember how it fits together.
Most companies in this position don’t need a development shop. They need an engineering function — and not a full-time one.
How we work together
Three ways in, in ascending order of ambition.
Most clients start small, because the fastest way to find out whether you want to work with an engineer is to give them something bounded and see what comes back.
The first rung carries a price because it is a fixed thing — same shape every time. The other two are scoped to whatever the work turns out to be, and both cost multiples of it. The first one is deliberately the smallest engagement we take.
- 01
Ship your own site
$1,500 · fixed · three sessions
Three sessions, about six hours. You stand up GitHub, Claude Code, and Azure in your own accounts and get your site live on them — the one you build during the sessions, or the one AI already made you and you had nowhere to put. You do the typing, so afterwards you can change it yourself.
For you if: You want a site you can edit yourself, and you would rather own the tools than rent a platform.
How it works → - 02
Demo to production
Fixed-fee assessment · then weeks of work, not sessions
You built something with AI, it works, and now people depend on it. We review the architecture, data model, and security, then rebuild what needs rebuilding and put it somewhere it can be trusted — without throwing away what already works.
For you if: Something you or your team built is now load-bearing and nobody has checked it.
How it works → - 03
AI agents
Scoped per engagement · the largest of the three
Agents that do work, not chatbots that answer questions — quoting, intake, document handling, the routine judgment your team burns hours on. Built into the systems you already run, with a human holding the decisions that matter.
For you if: A process that eats hours a day and is mostly rules plus a little judgment.
Talk through the process →
Client
HealthCare Comp walked the first two rungs.
HealthCare Compis a nationwide single-source provider of workers’ compensation ancillary services — durable medical equipment, home health, home modifications, complex and catastrophic care — coordinating referrals for adjusters, case managers, and TPAs in all 50 states.
Owner Scott Guimond didn’t start by hiring a development firm. He used AI to build his own site — and then hit the wall that stops almost everyone at that point: he had a website and nowhere to put it. Three sessions later, about six hours, he had GitHub, Claude Code, and Azure standing in his own accounts, the site live on his own domain, and the ability to change it himself. Scott did the typing throughout.
Then he built a working internal application with AI. It outgrew the demo it started as, real work began depending on it, and the hosting and architecture it had been prototyped on were not built to carry that. We are rebuilding it onto Azure now.
That order is the point. A $1,500 engagement is not a small version of a big engagement — it is how somebody finds out whether you are worth trusting with the big one. And a client who has built software himself is a better client on the second job, because he knows what he is asking for.
Once I had this new site, I didn’t know what to do with it or where it should live. That’s when I met with Greg from Datos. He walked me through the process of making the site real and getting it out into the world.
Could I have figured it all out on my own? Sure — but it would have taken far longer and delayed the launch. I likely would have missed things, leaving gaps in the site that could have cost me significant time and money down the road.
With Datos, my site is live and running smoothly, and I can make changes myself through the AI system, from a single word to an entire page. Since launching, I’ve seen a major increase in people reaching out to work with us, and the site is ranking high in most states.
Why us
Anyone can prompt. Fewer have kept it running.
We have done this the expensive way
For eight years Datos was the costly thing AI just made cheap — the engineer who understood both a specialist industry and the code. We are not commenting on this shift from the outside. We ran the version of it that cost six figures.
Production, not prototypes
Since 2018 we have run systems against live insurance data where an outage is somebody’s medical bill not getting paid. That is a different discipline from making a demo work, and it is the discipline your AI-built app is missing.
Security is not a phase
Claims data and PHI have been the normal case here for eight years. When we review what you built, we are not consulting a checklist — we are applying the standard we already hold ourselves to.
We run our own business on it
Datos operates on AI agents we built and maintain, handling real internal work every day. Everything we recommend to you, we already depend on ourselves.
Tell us what you built, or what you keep doing by hand.
A scoping conversation is usually enough to tell whether this is worth doing and what it should cost. If the answer is that you don't need us yet, we'll say that.