AI startup CEO vs UK AI Action Opportunities Plan: in detail

It’s not every day that the UK government pins its plan for economic growth and national revival on the technology you work with and build every day.

With that in mind I thought I’d take a closer look at the AI Action Opportunities Plan and highlight what’s here, what matters and what’s missing.

This plan accepts all 50 of Matt Clifford’s recommendations, and sets out to use them to achieve three things:

1. Lay the foundations to enable AI
2. Change lives by embracing AI
3. Secure our future with homegrown AI

Taking concrete steps here is very welcome. We’re all about opportunities here, but Peter Kyle MP lays out the biggest risk right at the start: “the UK risks falling behind the advances in Artificial Intelligence made in the USA and China.”

That risk has arguably already come to pass with LLMs, where Open AI went from challenger to incumbent in a matter of months. Before looking at the detail, I must pause and stress:

Delivery is what matters.

Not the recommendations, not the speeches, not the press coverage. Everything - everything - will come down to how boldly and quickly our policymakers are prepared to implement this plan.

Let’s dive in!

1. Breadth is the enemy of focus

Looking at the plan as a whole, boy is there a lot here. Matt Clifford has done an excellent job and is making the right calls throughout. But when we look at Labour’s industrial strategy, it’s broad - delivering clean power by 2030, harnessing data for the public good, caring for the future, building a resilient economy. They’re all incredibly important goals.

I have a few startup folks in my subscribers and you will all know that split focus is the death of quick results. To ship something big, complex and impactful, shutting out other priorities to an uncomfortable degree is what will get it done. If this plan is happening it needs to be the #1 focus. Will it be? What will be traded off - clean energy? Care reform?

2. We can win in the application layer

The UK missed the LLM wave. We’re polling our community members on LLM usage at the moment and ChatGPT, followed distantly by Claude, Gemini and Llama, are way out in front (despite DeepMind being a UK company originally).

But the application layer is all to play for. So many exciting UK companies alongside us on the Sunday Times Tech 100 list are finding innovative ways to harness AI to make life better for people and businesses. Are there multiple billion-dollar businesses waiting to emerge? Definitely. We just have to take the opportunity.

3. Infrastructure is important …

… but it’s not everything.

This is obviously true. But is the UK going to outcompete the existing infrastructure providers powered by China and the US? I suppose it’s possible, but it’s unlikely, at least in the short term. This feels like the least we should expect for any serious economy.

4. … data might be more important

Datasets in the public and private sectors could unlock huge value, as long as the concerns flagged above are addressed. It’s not hard to think where this might happen in legal. Land registry data? Court reports? These datasets touch legal processes that have massive impacts on our lives - the potential is huge.

5. Training, attracting and retaining top talent

The most important people in AI used to be, exclusively, academics. They worked at universities and research institutions and were probably chronically underpaid. Outliers like Daniel Hulme were making an impact commercially, but they were the exception, not the rule.

Now, to be top-tier talent in AI, you don’t need to be physically innovating in the hardware layer. You could just be a great AI developer. Their salaries are marching upwards inexorably, and we need to be able to attract them to the UK with the right immigration setup and tax incentives.

6. Diversity in the talent pool is more important than ever

Businesses in certain corners are rolling back DEI initiatives and policies.

Quite apart from anything else, this is counter to mountains of evidence about how diverse teams outperform those that aren’t. The AI revolution will be no different and I fear that businesses will be in too much of a rush to care. Improving diversity is right but it’s not fast. We must be mindful of this.

7. Getting the growth environment right

The UK’s approach to innovation is better than some and worse than others. Much of the debate here has been focused on listing rules to encourage IPOs here rather than in the US, and that’s valid.

But the truth is that the groundswell of innovation is not happening at the public company level. It’s happening at startups and scaleups who want to move incredibly quickly. The government tries to encourage this through R&D tax credits for businesses pursuing real innovation.

The problem is that R&D tax credits are only awarded for furthering the state of the art. While some of the work that Juro does qualifies, much of our AI development does not. That R&D tax credits are  narrower and narrower in application is a problem where there’s no reward for most AI development.

Ultimately, as a CEO, I’m spending lots of money on AI R&D, and applying it practically to business problems; if AI is the number 1 priority and we are an AI software company, why isn’t all of that R&D recognised in the tax system?

8. Changing lives? Absolutely

The ‘scan-pilot-scale’ approach will be familiar to anyone in a large business who ever procured some software. What’s good about that is it’s a proven approach; what’s bad about it is it implies really long sales and implementation cycles.

If the UK wants to leap on this AI opportunity and have faster, more productive public and private sectors not next year, but next month, it’s just not going to happen through a glacial procurement process.

Mistakes have of course been made very recently in that area by the previous government during the pandemic. But if we take the lessons and find a better middle ground, there must be a faster way than what’s mooted here of getting the right British companies in front of the right people.

Is there a central place where opportunities are shared? How do we get it moving faster? If you’re sitting on technology like AI Extract, which can read and then onwardly automate documents in seconds, at scale, how do we point it at the mountains of PDFs sitting around in local and national governmental departments?

9. The key takeaways

It’s exciting. If you spend a big chunk of your time working to convince people to adopt AI, and then turn on the TV to find the Prime Minister saying it’s a national imperative - well, that’s good.

But to bring it full circle, delivery is everything.

• The industrial strategy barely mentions AI: change that drastically
• The UK can excel at the application layer
• The right environment for talent is critical
• Bold regulatory and tax initiatives could give UK startups an edge
• The prize on offer is changing people’s lives - if we move quickly enough

The fact that Matt Clifford is being co-opted not only into the policy but into the delivery is a good thing. He’s actually built before, unlike almost anyone in Government. That’s a huge asset in a world where Elon Musk’s X.ai can open a data centre in 19 days. That’s the bar.

AI will continue to have a dramatic impact on business - everything from writing emails to legal team metrics will be affected - but to make that impact positive for the UK, we must back up these fairly bold words with decisively bold actions.