Automation you can see working.
We map the work your team repeats, build the automation in a controlled rollout, and hand it over running in production. Three phases. Nothing goes live before you understand it.
We're selective about where AI belongs. Language models where a judgment call is genuinely needed, ordinary software everywhere else. Much of what slows a team down doesn't need AI at all.
Someone on your team spends hours a week moving data between systems that should talk to each other.
You've bought AI tools and nobody can say what they changed.
The process everyone complains about is the one nobody has time to fix.
You probably already know what it is…
Three phases. Each one finishes before the next begins.
Most automation fails for the same reason: it goes live before anyone understands what it will do. We work in three phases and you commit to one at a time. Nothing is irreversible, and nothing is hidden.
Discovery
We sit with the people doing the work and map what actually happens, not what the process document says. You get a written picture of every step, how long it takes, and what it costs. Then we tell you which parts are worth automating and which aren’t.
You have a costed shortlist and a decision on what gets built.
Discovery is a fixed-price engagement in its own right. Plenty of clients stop here, take the map, and act on it themselves. That’s a good outcome.
Rollout
We build the automation and run it alongside your team, not instead of them. Every output is reviewed before it counts for anything. We find where it gets things wrong, fix it, and run it again. This phase takes as long as it takes to stop being surprised.
The automation has run clean for an agreed period, measured on your numbers.
Production
The automation goes live and your team stops double-checking it. You keep a dashboard showing what it’s doing, how often, and where it escalates to a person. From here most clients keep us on monthly, because a business changes and an automation that isn’t watched drifts away from it. We monitor, tune, and take on the next process.
If you’d rather own it outright: we hand it over fully documented and step back. The dashboard and the runbook are yours either way.
No long contracts
Two to three weeks, one process. You get the map and the shortlist whether or not you build anything with us.
We price it once we both know exactly what we’re building. No estimate before then would be honest.
Monitoring, tuning, and the next process. Optional, and most clients take it.
What you keep
Every phase produces something that stays yours, whether or not you continue with us.
Process map
Every step of the workflow as it runs today, with time and cost attached.
Automation spec
What we’re building, what it will and won’t do, and where a person stays in the loop.
Working automation
Built on your systems, in your accounts. Nothing runs on infrastructure you can’t see.
Monitoring dashboard
A live view of what ran, what it decided, and what it passed to a human.
Runbook
How it works, how to change it, how to switch it off. Written for your team, not for us.
Handover
A working session with the people who’ll own it, plus everything above in your hands.
Written case studies are in preparation. Until they're published, we'll walk you through the work in detail on a call.
How we build
We have firm opinions about how automation should be built. They're the reason our rollouts are slower than they could be, and the reason they hold.
Not on everything, which would defeat the point. On the decisions with consequences: money leaving, a customer being told something, a record being deleted. We agree those boundaries with you in writing before anything is built.
Every automated action is logged in plain language: what came in, what it decided, what it did. If an automation can’t be explained to someone who wasn’t in the room, it isn’t finished.
Automation earns its place by taking repetitive work off a desk, not by replacing judgment. The parts of a job that need context, taste, or a relationship stay with the people who have them.
Some processes are too rare, too variable, or too consequential to be worth it. Saying so costs us the work and saves you a system nobody trusts.
Who you'll work with
We've spent our careers inside fast-growing companies, on both sides of this problem. One of us builds the systems. The others have run the teams and the partnerships that outgrow them.
10+ years working with companies from Fortune 500 to startups, from founding member through Series C. Built the AI agents now handling customer support at a [Series B software company], from prototype through to production. On your engagement he does the architecture, the build, and the monitoring that comes after.
[X] years running sales teams through scale-ups, which means years spent inside the operations that break when a company grows faster than its processes. He runs discovery and scoping, and he’s your contact throughout.
[X] years driving strategic partnerships for startups, bringing in the systems and vendors a growing team has to work with. [Further detail to come.]
A founder is on every engagement. That isn't a growth stage we're in, it's how we intend to work.
Start with one process.
Thirty minutes, no preparation needed. Bring the workflow that annoys your team the most and we'll tell you whether it's worth automating and what it would take. If the answer is no, you'll get that too. If it's yes, the next step is a $4,000 Discovery, and nothing beyond that is committed.