Home/Approach
How we work
Most automation projects do not fail technically. They fail because the wrong problem was chosen, the tool was picked before the process was understood, or the team was handed something nobody could maintain. Our method is designed around avoiding exactly those three failures.
Diagnose · Design · Deploy · Embed
The same four phases apply whether the engagement is a two-week audit or a six-month programme. Only the depth changes.
Diagnose
We start where the work happens. That means sitting with the people who run the process, watching it end to end, and following the data as it moves between systems and spreadsheets and WhatsApp groups.
We are looking for three things: where the hours actually go, where errors enter, and which steps exist only because of a decision somebody made years ago and nobody has revisited since.
Output: process maps, effort and error baseline, ranked opportunities
Design
We choose the lightest solution that will still be standing in a year. Sometimes that is an integration between two systems you already own. Sometimes it is an AI model. Sometimes it is deleting a step entirely.
Before anything is built, we agree what success looks like in numbers — hours returned, cycle time reduced, error rate cut — and who owns the process once it is live.
Output: solution design, success measures, scope and estimate
Deploy
Delivery happens in short cycles with something demonstrable at the end of each. We pilot with one team or one location first, and run the new system alongside the old way until it has earned the team's trust.
Every build includes the unglamorous parts: error handling, logging, alerting, and a defined answer to the question "what happens when this breaks at 2am?"
Output: a working system, piloted, then rolled out
Embed
A system nobody understands is a liability, however clever it is. We document what was built and why, train the people who will own it, and set up the monitoring that tells you when something needs attention.
Then we measure against the baseline from phase one — honestly, including the places where the result fell short of the estimate.
Output: runbooks, trained owners, measured impact review
The rules we hold ourselves to
Fix the process before you automate it
Automating a broken process just produces the wrong answer faster. The first question is always whether the step should exist at all.
Start where the pain is loudest
Early wins buy the patience needed for harder work later. We sequence for visible relief, not for architectural elegance.
Boring technology, chosen on purpose
Your automation should be maintainable by an ordinary competent person a year from now — not only by the consultant who built it.
Humans stay in the loop where judgement matters
We automate the repetitive and the routable. Decisions with real consequences keep a person on them, with better information.
Adoption is part of the build
A system the team works around has failed. We design with the people who will use it, and treat their resistance as information.
Measure, then say what happened
Baselines before, numbers after, and a straight account of the gap. Including when the honest answer is "less than we hoped".
Your first thirty days
So you know exactly what you are signing up for before you commit anything more than half an hour.
Engagements move quickly at the start because the diagnostic work is deliberately time-boxed. The aim is to have something real in front of your team inside the first month.
- Day 0 — Discovery call. Thirty minutes. You describe the friction; we tell you whether we are the right help.
- Week 1 — Immersion. Interviews and process walkthroughs with the people doing the work.
- Week 2 — Baseline. Effort, error and cycle-time numbers documented, opportunities ranked.
- Week 3 — Roadmap review. A prioritised plan with costs, effort and expected returns, presented to your leadership.
- Week 4 — First build begins. Scope agreed and locked on the highest-value opportunity.
What we will not do
Being clear about this saves everybody time.
Charge you before it works
Fixed price, agreed upfront, invoiced once the system is live and doing what we said it would. Not before.
Sell you licences
We hold no reseller agreements and take no commissions. If your current stack is sufficient, that is the recommendation you will get.
Automate headcount out quietly
If a project has workforce implications, that conversation happens openly with leadership at the design stage — not as a surprise at the end.
Build something you cannot maintain
No undocumented systems, no dependencies only we understand, no engagement that survives on your inability to leave it.
The first step is a conversation, not a contract
Bring one process that frustrates you. We will tell you what it would take to fix it, and whether it is worth fixing at all.