How we work
A method built to survive contact with a real business.
We quantify before we design, and pilot before we commit.
- 01
Discover
How work arrives, who touches it, where it stalls — and where demand is being left on the table.
- 02
Quantify
Volumes, handling times and upside become numbers, with assumptions written down.
- 03
Design
Process designed to the outcome first. Technology is selected to fit it, never the reverse.
- 04
Deploy
Pilot on real volume, then integrate, train and hand over clear ownership.
- 05
Measure
Performance compared to the baseline, then the next opportunity in sequence.
The stages
What happens, and when.
- 01
Discovery
Sessions with ownership, management and the people doing the work. Nothing is recommended yet.
Output
Objectives, constraints, functions to examine.
- 02
Process mapping & data
How work actually flows: volumes, handling times, handoffs, rework and the data at each step.
Output
Current-state maps with evidence.
- 03
Business case & priority
Each opportunity sized in hours and pounds, then ranked by value, effort, risk and time-to-impact.
Output
Costed register and prioritisation matrix.
- 04
Pilot
One workflow, real volume, measured against the baseline. Thresholds agreed before it starts.
Output
Measured result and a go / no-go.
- 05
Implementation & integration
Built into your existing stack with exception handling, alerting and documented ownership.
Output
Production systems and runbooks.
- 06
Training & change
Trained by role, with what the system does and does not decide made explicit before go-live.
Output
Role-based training and internal ownership.
- 07
Measurement & optimisation
Results compared to baseline on an agreed cadence, then the next opportunity in sequence.
Output
Reporting against baseline and a live roadmap.
Principles
Non-negotiables.
Measurable ROI or no recommendation
Every initiative carries a stated value, stated assumptions and a baseline.
Human oversight where it matters
Automation handles the repeatable; judgement and sensitive cases route to a person.
Security and data discipline
Least-privilege access, agreed systems and jurisdictions, documented read and write scope.
Proportionate governance
Enough control to be defensible, sized for an SME rather than a multinational.
Economics before software
We arrive with no preferred platform. The workflow is designed to the outcome first.
Ownership stays with you
Documentation, access and configuration are yours. Continuing with us is a choice.
Questions
Timelines, measurement and governance.
How long does an AI opportunity audit take?
Typically two to four weeks, depending on the number of systems, sites and functions in scope. You receive current-state maps, a costed opportunity register and a prioritised roadmap. Implementation timelines are then set per workflow rather than as one programme.
How do you measure the ROI of AI projects?
Against a baseline agreed before any build: cost, cycle time, conversion, capacity released or revenue captured. Results are reported against that baseline on an agreed cadence. Where figures are scenario-based we say so plainly.
What happens if an AI pilot does not work?
We stop and say so. Each stage carries a commercial gate with thresholds agreed before the pilot starts, so nothing scales on assumption. A failed pilot still leaves you with evidence about where the value is not.
How is company data handled in an AI project?
Access is least-privilege, scoped to what the work requires and held under agreement. Read and write scope is documented, and systems and jurisdictions are agreed in advance. Your data is used only for your engagement.
Do AI systems replace staff or support them?
Automation handles the repeatable; judgement and sensitive cases route to a person. The usual outcome is absorbing growth without proportional headcount and returning hours you already pay for. What the system does and does not decide is made explicit before go-live.