Find the AI work worth backing before you build.
The six-step method behind every AI Opportunity Roadmap.
Most firms have more AI ideas than evidence. The Sprint Method turns real workflows into a costed order: what to do now, what to test next and what to leave alone.
Across 3 to 4 weeks, I apply the method to your people, your work and your numbers. You finish with a sequenced 90-day plan and the evidence behind every recommendation.
Direct reply from me within two business days. No sales sequence.
A good method earns its value in the ideas it stops.
The expensive AI decision is not always the one you make. It can be the process you automate before understanding it, the pilot you scale from a polished demonstration, or the tool you buy without knowing which result it must improve.
The Sprint Method tests the case before the commitment grows.
An opportunity can fail for useful reasons:
Finding that early is not a failed engagement. It is budget, attention and implementation effort protected.
The purpose is not to produce more AI projects. It is to find the few that deserve to exist.
No AI initiative earns the next pound until it proves a result.
The Edge Loop is the principle behind the brand: pick, test, measure, keep, stack. For one person, it builds an advantage around the way they work. Inside a firm, the same discipline turns scattered use into shared evidence and repeatable systems.
The Edge Loop shapes the judgement. The Sprint Method structures the engagement.
No recommendation earns more budget because another firm is doing it. It has to make sense for your work, your people and your numbers.
Six steps. Six decisions. One defensible order.
Each step answers a different question. Each one leaves evidence behind. Nothing is prescribed until the work, baseline and value case are understood.
I work with leadership to define the business question, the workflows in scope and the measures that matter. That might be recovered capacity, a shorter lead time, better quality, lower cost or less operational risk. The outcome and boundary are written down before the diagnostic begins.
I interview leadership and the people closest to the work, then map the workflow as it happens. We capture the steps, systems, decisions, exceptions and handoffs. We do not rely on the idealised process. The workarounds and hidden coordination often contain the real cost.
I trace each workflow until the delays, rework, manual loops and decision points are exposed. Not every bottleneck needs AI. Some need clearer ownership, a simpler process or better information. The diagnosis separates credible AI opportunities from ordinary operational problems.
Each opportunity is quantified using your own frequency, time, loaded cost, quality and risk data. The assumptions stay visible. Finance and operations can challenge the number rather than accept a promise. Where the evidence is incomplete, the Roadmap gives a range and names what must be measured next.
I compare each opportunity by recoverable value, effort, evidence, risk and readiness. Quick wins are separated from bigger bets. The process, data, knowledge and governance dependencies are made explicit. An attractive opportunity does not become a priority until the firm is ready to act on it.
The final Roadmap sets out the first moves, the smallest useful tests, the suggested owners, the measure of success and the decision each test should unlock. It can recommend an existing tool, a process change, a custom workflow, foundation work or no build. The recommendation follows the diagnosis.
The six steps are repeatable. The answer cannot be templated.
A worksheet can ask how long a task takes. It cannot tell whether the estimate is credible, whether the delay is a genuine cost, or whether automation will remove the bottleneck or push it somewhere else.
A scoring model can compare opportunities. It cannot resolve conflicting accounts from leadership and operators, expose a missing dependency or tell a board when the responsible recommendation is to do nothing yet.
That is where the consultancy earns its fee.
I start with the work, not a preferred product or predetermined answer.
I test the assumptions behind the brief, the baseline and the projected value.
I connect operational friction to capacity, cost, quality and risk in terms leadership can examine.
I connect the reality experienced by operators to the decision the board needs to make.
I do not receive commission from software vendors. A process change, an existing feature or no build can be the right recommendation.
Every engagement is led by me. I run the interviews, map the work, build the value case and present the final recommendation. The method creates consistency. Senior judgement makes it specific to your firm.
Every step becomes part of the Roadmap.
The work does not disappear into workshop notes. Each stage creates evidence that supports the next decision.
Five expensive mistakes the method is built to catch.
A fashionable product is not a business case. The workflow and result come first.
The process on paper often misses the exceptions, retyping and coordination that consume the time.
AI can make the first draft faster and the completed task slower. The whole job must improve.
A successful example does not prove that the approach will work across different people, volumes and edge cases.
Higher-value AI work often depends on documented processes, usable knowledge, clean data or clearer governance. The dependency must be visible before implementation starts.
A smaller, evidence-led first move is cheaper than unwinding the wrong system later.
The method removed 93% of the time and cost from one workflow.
At NatWest, I redesigned a structured component review workflow around one engineer working with AI. The original process took five hours across a developer, test analyst and business analyst. The bottleneck was not the depth of the review. It was the coordination between the review, the evidence and the development backlog.
The redesigned five-stage workflow produced evidence-linked findings and backlog-ready recommendations in 20 minutes with one engineer.
This was a workflow redesign, not a completed client Roadmap engagement. It demonstrates the evidence discipline behind the method: establish the baseline, find the real constraint, redesign the work, measure the whole result and let the evidence decide what happens next.
AI is a new context. The discipline comes from more than a decade of experiments for organisations including NatWest, Vitality, RNLI and Farrow & Ball.
Useful when the business has activity, but no defensible order.
The Sprint Method is built for established UK firms where work crosses enough people, systems and decisions to make prioritisation valuable.
Sector matters less than the shape of the problem.
From the first interview to the final recommendation.
I am Steve Quinlan, an AI consultant and product leader who has spent more than a decade finding what changes behaviour, removes friction and moves a measurable number.
My work includes product, conversion and AI experimentation for NatWest, Vitality, RNLI and Farrow & Ball. I have been building production AI workflows inside NatWest since 2023.
Every Roadmap is led by me. I run the interviews, map the work, build the value case and present the final recommendation. I do not sell the engagement and pass the analysis to a junior team.