Three questions. Four stages.One person accountable throughout.
AI is the catalyst. The work is on the business: where the advantage is, what has to change to reach it, and how the change is carried through until your own team runs it.
Every engagement answers three questions.In this order.
- 01
Where is the advantage?
Which few changes could move revenue, cost, capacity or customer outcomes enough to matter, and what would each be worth?
- 02
What must change?
Which products, workflows, roles and decision rights have to be redesigned for that value to appear, and what stays with people?
- 03
How will we realise it?
Who owns each result, what gets tested first, and how will we know whether to proceed, adapt or stop?
Four stages, in this order.Stop after any of them and keep every output.
We can work with you from the first decision through to delivery. Each stage also ends with something you can use if you stop there.
Stage 01
Decide
Where is the advantage?
Choose the few opportunities worth funding, each with a value case in numbers and the measures it will be judged by.
Most AI lists are too long. We cut yours down to the handful that could move revenue, cost, capacity or customer outcomes, and record what is evidence and what is still assumption.
- Prioritised opportunities
- Value case and measures
- Readiness, risk and dependencies
Engagements: Workshop · Diagnostic · Deep Dive
Stage 02
Design
What must change?
Rework products, workflows, roles and decision rights (who decides what) around those opportunities, and agree which calls stay with people.
Automating a poor process preserves the wrong work. So the work is redesigned first: what stops, what stays with a person, what AI assists, and what AI takes on within agreed limits.
- Target workflows and roles
- Decision rights and accountability
- Data, technology and governance needs
Engagements: Operating Model & Transformation Design
Stage 03
Deliver
How will we realise it?
Test the riskiest assumptions in small, time-boxed pilots with agreed success measures, then build and scale what works.
Every pilot ends in one of three decisions: proceed, adapt or stop. We stay through the build: your teams or the partners we help you choose write the software, and we work alongside them until it is live.
- Pilot specifications and acceptance criteria
- Partner selection and briefing
- Proceed, adapt or stop decisions
Engagements: Implementation Support · Transformation Leadership
Stage 04
Embed
Who runs it after us?
Give every result a named owner, train people on the new work, and track against the agreed measures until your team runs it without us.
Training uses your real workflows, not generic demos. The methods and outputs are yours to keep, and the engagement ends when the operating rhythm no longer needs us in it.
- Named owners for each result
- Training on real work
- Value tracking your team runs
Engagements: AI Training for Leaders & Teams
How an opportunity is sized.So a hypothesis is never mistaken for a business case.
Revenue, cost savings, released capacity and customer outcomes are kept separate. Time saved creates capacity. It only becomes a cash saving when spending is reduced or avoided.
That distinction is where many AI business cases quietly fail, so it is made explicit from the first estimate.
- The outcome expected
- What should improve, by how much, and for whom.
- Where you are now
- Current performance, wherever evidence exists to measure it.
- The assumptions
- What has to be true for the estimate to hold, stated so it can be tested.
- The cost of getting there
- Costs, feasibility, readiness and dependencies, including data and technology.
- Whether people will use it
- Adoption, quality and risk, which decide whether a working pilot becomes a working business.
- How sure we are
- Evidence gaps and a confidence level recorded against every estimate.
Work is allocated on purpose.Workflow by workflow, with a named owner.
The design stage asks the same five questions of every workflow. The answers become the operating model: how the work gets done and who decides. They are also why people know where they stand when the change lands.
What is the work for?
What value is it meant to create, and for whom? Work that creates none should stop before anyone automates it.
What stays with a person?
Judgement calls, accountability, and the conversations where a customer or colleague needs a human.
What does AI assist?
Work a person still owns and signs off, done faster or better with AI alongside.
What does AI take on?
Work delegated to AI within defined controls, with a named owner and a way to review and reverse it.
What changes around it?
Roles, skills, systems, measures and governance, so the new allocation holds in practice.
Governance runs through every stage.It starts before anything is built.
It runs through every engagement, because the cheapest time to deal with a regulatory or accountability problem is before anything has been built.
Human and AI accountability
Every workflow redesign sets out who is accountable, what stays with a person, what is automated, and how a decision can be reviewed and reversed.
FCA Consumer Duty and conduct risk
For financial services and insurance, designs are tested against Consumer Duty outcomes and conduct risk before they reach a board.
UK GDPR and the ICO
Data use, retention and third-party processing are covered in scoping, before the build starts.
EU AI Act
Risk-category exposure is flagged early, while the design can still change.
Board oversight
Assumptions, risks and dependencies are made as visible as the opportunity.
Governance in the plan
Mobilisation plans include an AI governance and risk view, with owners and decision gates.
Five principles.Each one replaces a habit.
The habit: Start from the tools
The principle: Business value before technology
The question is what the business could do differently, and become, now that these capabilities exist.
The habit: Automate what exists
The principle: Redesign before automation
Automating a poor process preserves the wrong work.
The habit: Let allocation happen by default
The principle: Allocate the work deliberately
For every workflow: what stays with a person, what AI helps with, and what AI takes on within agreed limits.
The habit: Scale on enthusiasm
The principle: Evidence before scale
Small, time-boxed pilots with agreed measures. Each ends in proceed, adapt or stop.
The habit: Stay dependent on the adviser
The principle: Transfer the capability
The outputs and the methods belong to you. The aim is a team that runs it without us.
About the approach.Common questions.
What happens in each stage?
Decide: choose the few opportunities worth funding, with a value case and measures for each. Design: redesign products, workflows, roles and decision rights around them. Deliver: test the riskiest assumptions in small, time-boxed pilots, then build and scale what works. Embed: train your people, track results against the agreed measures and leave an operating rhythm your team runs without us.
Do we have to do all four stages?
No. You can stop after any stage and keep every output. Many clients start with a single fixed-fee step in the Decide stage and choose what follows once they have seen the result.
Who is accountable when AI is involved in a decision?
A named person. Every workflow redesign sets out who is accountable when AI is involved, what stays with a person, what is automated, and how a decision can be reviewed and reversed.
Do you cover the FCA Consumer Duty?
Yes. For financial-services and insurance clients, opportunities and designs are tested against Consumer Duty outcomes and conduct risk before they reach a board.
How do you handle the EU AI Act and UK GDPR?
Where a product or workflow could fall into an EU AI Act risk category, it is flagged early, while the design can still change. Data use, retention and third-party processing under UK GDPR are covered in scoping. We are not a law firm, so formal legal advice stays with your counsel.
Do you sell software or resell AI tools?
No. Training and implementation support are tied to the changes agreed with you. They are not off-the-shelf courses or standalone builds, and no tool vendor pays us.
Start with a conversation.Thirty minutes, free.
Tell us what you are trying to decide. If we can help, we will say how and what it would cost. If we can't, we will say that too.