How we work
Seven steps. The same ones, every time.
Most automation starts with the tool and works backwards. We start by finding out what’s actually costing you, and we only build where the numbers say it’s worth it. This is how that goes, in order.
Map it
We draw the job as it actually happens — who starts it, who touches it, which systems it passes through, where it waits. If it can't be drawn, it can't be fixed on purpose.
Measure it
A before-number with a unit, a period and a source. Estimates are fine; they're labelled as estimates, so nobody mistakes a guess for a fact later.
Find the cause
Symptoms and causes get separated. A slow handoff because someone's off sick is not the same problem as a slow handoff because two systems don't talk.
Rank it
Each opportunity scored on what it's worth, how hard it is, whether you're ready, and what could go wrong. The first thing we fix is the one worth doing now — not the most interesting build.
Design it
The target job, the human checkpoints, what happens when something's unusual, and exactly what we'll measure afterwards. Sometimes the design is: don't automate this step.
Build it
Inside the tools you already use. Rules configured, people trained, a fallback in place. We own the change through go-live, not a slide recommending a tool.
Prove it
The same measurement, taken again over an agreed period. What improved, what didn't, and the caveats — written down. Success is observed, not declared.
Step 4, in detail
How we decide what’s worth fixing
Every opportunity gets the same five questions. The scores are shared with you, in plain English, in the operations map.
01
Worth
What changes for the business if this works — time, quality, cash, risk, or capacity.
02
Feasibility
Whether the systems, access and skills exist to build it without a science project.
03
Readiness
Whether there's an owner, a stable process, and data good enough to start.
04
Risk
What happens if it fails — for customers, compliance, or the day-to-day.
05
Confidence
How strong the evidence is for the before-number and the expected change.
The honest answer
When we’ll tell you not to automate
Some of the most useful things we say are “not this” and “not yet”. Any of these, and we’ll say so on the call.
- The job rarely repeats — one-off work isn't a pipeline.
- The process still changes every week; you'd be automating a moving target.
- Nobody is accountable for the outcome after launch.
- The data you'd build on isn't reliable yet.
- The decision needs unmanaged human judgement — safety, legal, or commitments you can't undo.
- A tool you already pay for already does it, if it were set up properly.
- What it's worth is less than what it would cost to build and get adopted.
- The business won't change how it works — the inbox would stay the system of record.
Start with the free review.
A 20-minute call to understand the workflow and decide whether there’s a sensible next step.
Prefer to start in writing? Tell us what’s getting stuck. We’ll read it and reply within one working day.