The Groundwork Matters More Than the Hype

Adam Sturdee
Apr 1
4 min read

There is a version of this blog post that opens with breathless predictions about AI transforming education. This is not that post.

If you are a school leader reading this and your instinct is that AI is overhyped, I understand that reaction. Much of what has been said about artificial intelligence in schools over the past two years has been vague, speculative, and disconnected from the reality of running a school. The gap between the conference keynote and the Monday morning briefing is vast.

But I want to make a different argument. Not that you should be excited about AI. Not that you should rush to adopt it. But that there is quiet, unglamorous work you can be doing right now — work that will serve your school regardless of what AI does or does not become — and that the schools who do this work will be in a fundamentally stronger position than those who do not.

That work is building your digital core.

What I mean by digital core

Most schools have data. The problem is where it lives and how it connects. Assessment data sits in one system. Behaviour logs in another. Attendance in a third. CPD records might be in a spreadsheet on someone's desktop. Curriculum planning lives in department shared drives that nobody else can access. Timetabling is its own island.

The result is that answering even basic cross-cutting questions — which students are underperforming across multiple indicators? which departments have the highest staff turnover and lowest CPD engagement? — requires someone to manually pull data from several places, reconcile it, and hope the formats are compatible.

A digital core means centralising and structuring this information so that it is accessible, interoperable, and clean. Not because AI demands it, but because good decision-making demands it. AI simply makes the cost of not doing this much higher.

Why this matters now

The honest position on AI in education is that nobody knows exactly what it will look like in three to five years. Policy is lagging. The DfE is cautious. Schools are right to be wary of vendors making promises they cannot keep.

But here is what we do know: every meaningful application of AI depends on structured, accessible data. Whether that is generating personalised revision materials from assessment records, identifying patterns in behaviour data, or — as we have found at St Augustine's with Starlight — generating coaching insight from lesson transcripts, the quality of the output is determined by the quality and accessibility of the input.

Schools that have their data scattered across incompatible systems, locked in formats that only one person understands, or buried in filing cabinets (physical or digital) will not be able to benefit from these tools even when they arrive in mature, trustworthy forms. The bottleneck will not be the technology. It will be the data.

What this looks like in practice

At St Augustine's, part of my work as Assistant Headteacher has been thinking about where our data lives and how it flows. Some of this is straightforward: making sure assessment data is held centrally and consistently across departments. Some of it is more deliberate: when we introduced Starlight for coaching, one of the things that made it work was that the data it generates — anonymised, aggregated coaching trends — feeds directly into leadership decision-making about CPD priorities. It does not sit in a separate silo. It connects to the questions we are already asking about teaching quality.

That integration did not happen by accident. It happened because we thought about data architecture before we thought about the tool.

This is the principle I would encourage every school leader to consider: before you evaluate any AI product, ask yourself whether your school's data is in a state where that product could actually be useful. If the answer is no, that is where your energy should go first.

Addressing the scepticism honestly

I think some of the resistance to AI in schools is well founded. The sector has been burned before by edtech promises that did not deliver. Teachers are exhausted and do not want another initiative.

But I also think some of the resistance is about something harder to name. When a technology threatens to automate or augment parts of what professionals do, the natural response is to question its value. That is human and understandable. Teaching is deeply personal work, and the suggestion that a machine might contribute to professional development can feel like an affront to the craft.

I would gently push back on this. The best AI tools in education are not replacing professional judgement. They are giving teachers and leaders better information, faster. The question is not whether you trust AI. The question is whether you trust yourself to make better decisions with better data. I think most school leaders would say yes.

The cost of waiting

Building a digital core is not an AI project. It is a school improvement project. Clean, accessible, well-structured data helps you make better decisions today, with or without artificial intelligence. It reduces the time senior leaders spend chasing information. It makes reporting more efficient. It supports accountability without adding workload.

The fact that it also positions your school to benefit from whatever AI tools do prove their worth is a bonus, not the primary justification.

The schools that will struggle are not the ones that chose the wrong AI product. They are the ones that, when a genuinely useful tool arrived, discovered they had no foundation to build on.

You do not need to believe the hype. But you do need to lay the groundwork.

Interested in this kind of work?

I work with school leaders on digital strategy, inspection readiness, and AI-augmented leadership practice. If you're thinking about how to build your school's evidential infrastructure — or how to use it more effectively — I'd be glad to talk.

Adam Sturdee is a senior leader and co-founder of Starlight, the UK’s teacher-first AI-powered transcript-based coaching platform for educators. His work sits at the intersection of dialogic practice, instructional leadership and responsible AI strategy for schools and trusts.

He will be presenting his research on AI-supported coaching at the BERA TEAN Conference 2026: https://www.bera.ac.uk/conference/bera-tean-conference-2026

The Groundwork Matters More Than the Hype

Recent Posts

Comments