Services
Product, culture, and code. In your organization, not in a workshop.
Rileytech is a product, culture, and code consultancy. We work embedded: next to your product people on your real customers, with your leaders on the culture you actually have, and in your codebase with your developers.
Every seat in your building now has a capable model in it, and what separates the teams getting leverage from the teams saying "AI sucks" is whether the people can say what they do, plainly, and defend it.
One rule for this page: everything below is something we have actually done, with a receipt where we have one. If it's not here, we haven't run it.
How we work
In your organization. No workshops. We haven't seen them stick. We work on your product, with your culture, in your actual codebase, so what's left behind is yours and you have examples to point back to.
Short and intense. We're not looking for a long contract. Think of us as your own special ops team: in, on the real problem, out. We expect discipline, participation, and candor from everyone in the room.
Action over discussion. Months of discovery and a stack of reports is waste. If a thing might work, we try it now, at the smallest size that can teach us something.
We'll tell you. If it's working, we'll say so. If it's a fabrication, we'll say that too, and early.
What we do
Three lanes. Most engagements touch all three.
Product
What to build, for whom, and how you'll know.
- Customer interviews with the people who'd pay, run by your team with us next to them, not by us on your behalf.
- "What problem are we solving?" asked until the answer is a customer's problem and not the team's belief.
- Pricing and demand tested with real money before real code. Call the last ten people and take their money.
Receipts: How do you price your product? A real example. Output vs Outcomes, a real example.
Culture
What's actually going on under the posters, and doing something about it.
- Retrospectives that draw out what the team already knows and isn't saying, then turn it into an experiment with a name on it.
- The oracle: the tenured expert whose "no" blocks every change. Getting what's in their head into the build and the team.
- Hiring for the record, not for sound. What broke, what it cost, what they'd never do again.
Receipts: Retro cheat sheet: collective consciousness. Managers, Leaders: why a team has not changed.
Code
Next to your developers, in your codebase, making it cheaper to change.
- Test-driven work on your legacy code. Finding the seams, strangling what has none, writing the tests at the moment they'll pay off.
- Single responsibility plus YAGNI and DRY as the load-bearing wall, and the rest of the letters showing up on their own. We shaved SOLID down in public.
- Mobbing, pairing, and diverge-converge, so a whole team learns a technique at once instead of one person carrying it.
Receipts: The Only Load-Bearing Letter. Diverge-Converge.
Where AI fits
It's not a fourth lane. It runs through the other three.
We use these tools every day and write about how, including what went wrong. The skill isn't prompting. It's legibility: saying what you do specifically enough that a stranger with zero context can act on it, and knowing a bad answer when it comes back polished. The model just stopped letting people fake it.
What we bring to the AI part of an engagement:
- Five people, not five thousand seats. Find the ones already getting real work out of the tools, put them on one specific problem with a number attached, and let the wins recruit the next cohort. Buying licenses first and hoping is spending as strategy.
- Get it out of your head. Make the model ask your people the questions, so the constraints they never wrote down land where the tool can use them. Three questions for a low-stakes call, ten for an architecture decision.
- Vendor and model skepticism. Which "AI-powered" tool is a thin wrapper with a markup. What happens when the price quadruples. Which model is actually better for your workflow, tested, not assumed.
The argument in full is Skills Over Seats. The operator's manual, four parts, starts at The Blank Slate.
Who this is for
The owner of a number. A founder, a VP, a team lead, an operator measured on whether something moved, with a team that's stuck, not broken: good people, real product, and nothing changes.
And we've heard why it can't happen here. Every version of it, from teams all over the world:
- That can't happen here
- I love what you're saying but
- In reality...
- How it works here
- But we are following another company's model
We think it can happen here, and that "in reality" the organization is stuck and needs a shove. We dare you to call our bluff. What's the catch? We will expect a lot more out of you and your teams afterward.
Who it's not for. Anyone shopping for a workshop, a vendor selection, a strategy deck, or long-term staffing. Strongly put, we're not for everyone, and we'd rather say so in the first conversation than the sixth month.
What happens first
One conversation. Thirty minutes, one question: is there a specific problem here, with a person who owns it and a number attached? Most of the time the honest answer is not yet, and we'll say so.
Then a shape that fits the problem. A few days in one team's codebase. A few weeks across product and delivery. A week here and there over a quarter. We size it to the problem, not to a menu.
Then we leave. Your people can run it, explain it, and run the next one without us.
On the Front Range? We do this in person in Denver and Boulder. See Colorado.