Home/About

A small outfit for the AI-curious-but-skeptical.

Dev Cowboy is what we wished we could have hired when we were on the other side of the table. Senior practitioners, fixed scopes, code that lives in your repo, and the discipline to say when something isn’t worth doing.

[ founder photo ]
Jordan Hale
Founder · Dev Cowboy
How this started

The $400k quote that started it all.

In 2024, a friend who runs a 40-person logistics shop got quoted $400k and twelve months by a brand-name consultancy — to do what amounted to a few automations and a knowledge base. I built it for him in three weeks, on weekends, for the cost of API calls.

I spent the next year quietly doing the same for other operators in his network. Dev Cowboy is what that turned into: a small, opinionated practice for the 10–500 person band — too big to ignore AI, too lean for an enterprise engagement to make sense.

Before this I led ML platform work at a public fintech (2018–2022) and shipped applied agents at a foundation-model lab (2022–2024). Every engagement is staffed by a senior practitioner — no body shop, no rotating juniors. If you talk to us on the fit call, you’re talking to whoever will do the work.

Principles

Six things we believe enough to bet the practice on.

We’ve turned down engagements over each of these. They’re not posters on the wall — they’re how we decide what to take and what to ship.

01

Hours saved, not hype generated.

If we can’t put a number on the value of an automation by week two, we shouldn’t be building it.

02

The riskiest part first.

The dumb path is to build the easy parts and discover the hard part doesn’t work in week four. We don’t do that.

03

Humans in the loop, by default.

Replace-your-team automations are a fantasy. Multiplier automations are real. We build the second kind.

04

Your code, your repo, your call.

Everything we build is yours from day one, MIT-licensed back to you. No vendor lock-in, no proprietary platform fees.

05

If it can’t be evaluated, it can’t be trusted.

Every workflow ships with a test set. Every change runs against it. Regressions page someone awake.

06

We’d rather lose the deal than oversell.

About a third of fit calls end with us recommending no engagement. That’s the right number, not a failure.

The team

Three full-time, four on call.

Dev Cowboy is deliberately small. Three of us full time — one ML engineer, one full-stack engineer, one designer-who-codes. We bring in four trusted contractors when an engagement calls for it: a security specialist, two senior backend engineers, and an ops/infra hand.

Every engagement has a single named owner from day one. That person sits in your standups, signs the contract, and answers Slack pings. We’ve never rotated an owner mid-engagement and we don’t plan to.

We’re fully remote — Austin, Portland, and Toronto — but we get on a plane for kickoff sessions when it matters. Most clients see us in person at least once per quarter.

We hire slowly. The bar is: have you shipped something in production that real people use? Have you also been on the call when it broke? Both, or we won’t take the chance. If that’s you, we’d love to hear from you at hire@devcowboy.com.

Let’s ride

Twenty minutes. One call.
Walk away with a plan either way.

We’ll look at one workflow you hate, sketch what an automation would look like, and tell you straight if it’s worth doing. No deck. No follow-up sequence.