Businesses should begin AI adoption by identifying and prioritising a real organisational problem—not by asking what tasks they can automate. In an opinion article published on 5 October 2026, Martin Paton, CTO at CreateFuture, argues that AI can first help uncover issues such as technical debt, continuity risks, duplicated technology, compliance concerns and unused cloud spend. Only after diagnosis should a business choose a focused, measurable use case.
Why the starting question matters
“What can AI automate?” makes the technology the starting point. That can produce experiments that generate activity without addressing a business priority. Paton’s alternative is to start with the organisation: which challenges matter, and in what order should they be solved?
As an Amazon Associate I earn from qualifying purchases.
That shift changes the sequence. A business first connects a proposed use of AI to an objective, then investigates whether AI is suitable for the problem. The answer might be automation, but it might instead be better visibility into operational or technical weaknesses.
Use AI to diagnose before deploying it to automate
Paton says AI can help identify problems that are difficult to see across a complex organisation. Examples in his article include technical debt, continuity risks, duplicated technology, architectural weaknesses, compliance issues and cloud spend that is not being used effectively.
He illustrates the argument with a client example involving unused cloud spend. This is an account reported by Paton, not an independently verified case study or evidence of a typical result. The broader point is that diagnosis may be a valuable outcome in its own right: a company can learn what needs attention before committing to an automation project.
Diagnosis does not mean deploying AI indiscriminately to inspect every system. The business still needs to define the question, determine what information the system may access, and decide who is responsible for acting on the findings.
Rank #2
Choose one high-impact use case and measure it
After prioritising the problem, Paton recommends a thin-slice implementation: one high-impact use case with executive sponsorship, clear measures and an iterative path toward a longer-term objective. This limits the initial scope while giving leaders a way to assess whether the work is producing business value.
Free tools Windows power users keep installed
One-click scans. No signup required.
- Define the business problem. State the operational or organisational challenge in terms that matter to the business, rather than naming a model or tool as the goal.
- Set a baseline and a success measure. Choose an outcome that can be assessed against the current situation. Faster output may matter, but it should connect to the intended business impact.
- Name an executive sponsor. Give a senior owner responsibility for the use case’s priority, decisions and progress.
- Run a focused implementation. Keep the first effort narrow enough to learn from and adjust, rather than making a blanket rollout.
- Review results and decide what comes next. Use the measures and lessons from the first use case to refine the approach or progress toward the longer-term objective.
The article does not provide a universal set of metrics or promise a particular return. The useful measure depends on the original problem: a business should decide what evidence would count as progress before judging the AI system.
Budget for usage and manage risk
Paton notes that enterprise AI can use consumption-based pricing, where usage and processing volume affect costs. That makes it important to understand how a proposed workflow will be used and to monitor its consumption. It does not mean every provider uses the same pricing model, and the article supplies no rates or savings estimates.
Risk work should accompany the implementation, not follow it as an afterthought. NIST’s voluntary AI Risk Management Framework organises risk management into four functions:
- Govern: establish responsibility, policies and oversight.
- Map: understand the system’s context, intended use and potential impacts.
- Measure: assess relevant risks and performance.
- Manage: prioritise and address risks over the system’s lifecycle.
NIST’s Generative AI Profile is a companion resource for applying the framework to generative AI risks, including in contexts such as large language models, cloud services and acquisition. It is a risk-management aid, not an endorsement of Paton’s strategy or evidence that any particular deployment will deliver a return.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Read the 95% claim in its stated scope
Paton’s article attributes a 95% figure to a 2025 MIT report and describes it as referring to pilots that failed to show direct P&L impact. That is not the same as showing that AI technology does not work, nor does it establish a failure rate for all AI projects. The underlying report’s denominator and methodology are not established by the cited article, so the figure should be treated as Paton’s account of the report rather than a fully independently verified statistic.
His central caution is about framing: a claim about pilots not demonstrating direct financial impact should not be repeated as a blanket verdict on AI. For an individual business, the more useful question remains whether a defined use case addresses a priority and produces evidence against a relevant measure.
The decision to make before choosing a tool
Paton puts the test this way: “Success, therefore, isn’t defined by the tech you buy, but by the questions you ask – and these should start with two fundamentals: which core organisational challenges do we need to solve, and in what order?”
Quick Recap
That is the practical distinction between a technology-led rollout and a business-led AI strategy. Diagnose and prioritise first; then test AI in a focused, sponsored use case whose costs, risks and outcomes can be evaluated.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchRead Martin Paton’s article in The AI Journal.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
