Industries rarely fix a structural problem because someone asked them to. What moves them is often competitive survival.
The financial services sector in the UK just made the biggest commitment to skills we have ever seen. There is a very good reason why they are doing it.
Why this sector had to move first
Financial and insurance services contributed £223.6 billion in 2025, eight per cent of UK economic output, and 1.09 million jobs. It is the fourth largest industry in the country.
If you look at the application of AI across the financial services sector, you realise how broad and widespread the usage is.
The work is fraud detection, financial crime screening, credit and risk analysis, regulatory reporting, and an enormous volume of drafting, summarising, reviewing and checking. It's knowledge work end to end.
The industry is already taking a lead on AI
NatWest saved more than 70,000 hours in its retail business through automated AI call summaries, part of a £1.2 billion technology transformation in 2025. Writing up a customer service call seems like a minor task, but at scale, it was costing tens of thousands of hours.
Lloyds built Athena, a tool that searches 13,000 internal articles so customer service colleagues can find an answer while the customer is still on the line. It cut the average search from 59 seconds to 20. For telephone banking teams alone, that is an estimated 4,000 hours a year of colleagues hunting for information and customers waiting.
These efficiencies are already being found across the industry, and there will be many more like them. Wins like that are sitting in every financial services business in the country, and unlocking them is dependent on building a business's AI capability at scale.
What the sector signed
The Financial Services Skills Compact launched in July, supported by HM Treasury and the Financial Services Skills Commission, and led by Claire Tunley.
Signatory firms commit to upskilling their whole UK workforce in up to five critical skills, at least one of which has to be AI, over a rolling three-year period. They name a senior executive who carries accountability, and they publish their progress every year.
Twenty-four regulated firms have signed, employing almost 270,000 people between them, with another 21 organisations backing the agenda as Supporters. HSBC, Barclays, Lloyds, NatWest, Nationwide, Standard Chartered and Zopa are among the signatories.
Tunley wrote that she believes it is one of the most significant sector-wide skills initiatives we have seen in a generation. On the evidence, that is not overselling it.
What happened when construction tried this
Back in 2014, I was building a startup in construction technology, trying to improve how payment moved through a supply chain. Construction is another significant contributor to the UK economy, worth around 6.2 per cent of output, and it had an endemic problem that many legislations sought to solve. The subcontractors that are actually doing the work sit at the end of the chain, get paid last, and are the ones who go under when cash stops moving.
Everybody knew about it, and everybody agreed it was bad. Which was why the Prompt Payment Code was born.
The industry made supportive noises and organisations signed up, including Carillion. Then, in January 2018, Carillion collapsed, holding an estimated £800 million from its supply chain and leaving around 30,000 small firms unpaid.
Lots of good intention, very few outcomes.
This policy was destined to fail. The construction industry runs on working capital. The earlier you pay your subcontractors, the less working capital you have. The less working capital you have, the less capacity you have to take on new projects.
An aside I keep enjoying. Rachel Reeves announced this financial services Compact, who back in 2018, then chairing the Business, Energy and Industrial Strategy Committee, called it "a bitter irony" that Carillion had signed the prompt payment code while making suppliers wait 120 days or more. She has watched a voluntary code fail from close range.
Financial services is a different case, and this is the bit that gives me much more optimism. There are no losers here. When everyone in an industry gets better at this, everyone in that industry competes better internationally, which is the whole point.
The transparency and the self-interest point the same way. That is why I think this one holds.
What being good with AI looks like
The Commission has a Future Skills Framework, first launched in 2021 and updated last year, naming 13 skills the sector needs. Machine learning and AI are among them. So are data analytics, cybersecurity and digital literacy. But what I love in particular is that creative thinking, adaptability, coaching and empathy are also mentioned, which tells you the Commission already understands that technical skill alone does not produce capability. Firms pick up to five to track, and one has to be AI.
That framework answers which skills matter. It does not answer what the AI one looks like.
That is a question we spent the early part of this year on, going deep into the Turing Institute's AI Skills for Business competency framework and into Skills England's research, which drew on more than 150 employers.
We came out with six capabilities.
- Literacy. Knowing the foundations of what we're working with
- Safe and responsible use. Applying AI safely and ethically across privacy, security and bias.
- Framing. Spotting opportunities for where AI can help
- Specification. Knowing how to create clear objectives and structure the ask well.
- Application. Putting AI to work in real tasks, along with the right tool or model for the job.
- Evaluation and reflection. Judging what comes out, and being honest about how you are using it.
The Skills Compact lays out the importance of AI and the need to report on how AI skills are developing. Our approach is about what being good with AI looks like.
The part that needs a version two
The Compact measures upskilling as the number of unique learners engaged, rather than hours logged or courses completed. That is a real improvement. Hours and completions are the two worst metrics in corporate learning, and they have survived for decades because we never had a better way of tracking it.
This is where learning and skills meet business transformation, and it is also where our education system tends to break down. We teach, we certify, and we hope people can apply what they have learned to real work. Very rarely do we track how that learning landed or what impact it made.
That is the problem we are solving with AcademyAI. Build the foundational AI skills the industry needs, help people apply them to real work, then track and evidence how an increase in capability leads to an increase in outcomes.
The first annual reports will do the arguing for me. Some firms will publish a learner percentage and stop. Others will publish what changed. The gap between those two is where the opportunity sits.
My question to you: if your board or your biggest client asked you next month to evidence that your people are more AI-capable than they were in January, what would you send them? If the honest answer is a training completion report, better to know that now. Deciding what evidence looks like is easier before you are required to produce it.
Every industry ends up here
I am biased, and I still think this is true.
We cannot buy the tools and hope. That is the default almost everywhere right now.
We cannot leave AI capability with a small group of enthusiasts while everyone else waits to be told how their job is changing, or spends energy resisting it. That resistance is usually a rational response to being handed a tool and no explanation.
Everyone needs to understand what they are working with, to see the opportunities in their own role and the risks in their own hands, and to share an idea of what good looks like, so that a workflow one person improves becomes something the whole organisation does.
Financial services has decided to take capability seriously and make progress measurable in public. The fourth-largest industry in our economy is going first, which makes it worth watching closely.
Thanks, as always, Andy
PS. Eight per cent of UK economic output, and 1.09 million jobs (Source: House of Commons Library, 18 August 2026). The Compact and its signatory list are at financialservicesskills.org/skills-compact, and the Future Skills Framework is here. Economic figures come from the House of Commons Library briefings on financial services and construction. NatWest's figures are reported by Computer Weekly, and Lloyds' Athena figures are in their own announcement. The Turing Institute framework is at turing.ac.uk/skills and Skills England's research at gov.uk.
