There is a Spanish proverb I grew up with: en casa del herrero, cuchillo de palo. In the blacksmith’s house, a wooden knife. The village blacksmith spends his days forging fine steel for everyone else, and eats his own dinner with a crude wooden blade. As a teacher, I often felt I was letting my own children down because the expertise I poured into other people’s children too rarely made it home for my own. I remember my mum tutting and saying, more than once: en casa del herrero, cuchillo de palo.
The proverb has stayed with me, because I now see a version of it in almost every school I visit. Leaders who know very well what good teaching looks like, and who nonetheless preside over digital strategies that have little to do with learning. Most of the documents I am asked to review are not digital strategies at all. They are infrastructure plans cosplaying as a strategy. They tell me what platforms the school has bought, what devices it intends to roll out, what cyber-security certifications it is pursuing. These are necessary things. But they are not, in any meaningful sense, a strategy for learning.
A genuine digital education strategy answers a different question. Not “what will we buy?” but “what will our students become as a result of being taught here for seven years in a digital age?” The first question is operational. The second is philosophical, pedagogical, and deeply uncomfortable, because most schools have never been asked it.
For a long time my contribution to that uncomfortable question was a framework refined over almost two decades in senior roles in schools, and the slow, one-school-at-a-time work of helping leaders think it through. Useful, and certainly worthwhile. But slow. Very much a wooden knife of my own.
So this year I decided to put my money where my mouth is and do it differently, blowing some new life into the old framework by learning the rudiments of coding and building something with the trusty help of Claude. The result is a free digital strategy diagnostic for schools, powered by the Anthropic API but grounded in the framework I have spent years refining. It is now live, and already being used by schools and trusts I have never met.
This piece is about why it exists, what I learned from building it, and what I think it tells us about how schools might think about AI itself.
The framework, and why a static version was never enough
The framework rests on six areas: vision and strategy, leadership and culture, infrastructure and access, teaching and learning, professional learning, and evaluation and impact. None of these are original on their own. What I think is novel is the way the diagnostic forces them into conversation with each other. A school can score well on infrastructure and abjectly on evaluation, and the report will say so plainly. The point is not to produce a number. The point is to hold up a mirror.
I could have built this as another static self-evaluation. A spreadsheet, a PDF, a printable framework. I have built such things before, and people download them. Some might even have used them. But the reason I chose a different route this time is simple: a static framework asks the user to do all the interpretive work. It says, in effect, “here are the questions; you figure out what your answers mean.” That is a high cost for a busy head or senior leader to pay.
”The framework and the judgement remain mine, encoded in the prompts. But the reader gets something that addresses their school, not a generic archetype.
What AI changes, and this is the part I think the sector has not yet fully metabolised, is the cost of bespoke. A well-prompted language model, working from a carefully designed framework, can read a school’s responses and write a report tailored to them in a way that would have been the work of a few days’ consultancy just a couple of years ago. The framework and the judgement remain mine, encoded in the prompts. But the reader gets something that addresses their school, not a generic archetype.
This matters, I think, beyond the tool itself. It is a small, concrete demonstration of a much larger principle: that AI’s most useful role in education is rarely to replace human expertise, and almost always to make that expertise scalable in ways it could not be before. The diagnostic is, in that sense, the same argument I make to schools about AI, not delivered by me in person, but carrying my voice and my judgement to schools I may never set foot in.
What I learned from building it
Three things, quickly, because the learning has been the point for me.
First, the framework had to get sharper with every build. When you ask an AI to make sense of a school’s responses, every vague or poorly worded question reveals itself instantly. Building the tool forced me to rewrite questions and rethink mental models that I had been using comfortably for years.
Second, the aggregate picture is humbling. I will share the full findings properly when the anonymised dataset is large enough to matter, but the early pattern is unambiguous: the area schools score lowest on, by some distance, is evaluation and impact. It turns out we adopt, but we rarely measure.
Third, building this thing is the most honest form of consultancy I have ever done. It has tested and sharpened my practical grasp of the digital bit in digital strategy. There is nowhere to hide when the tool either works or it doesn’t, and when its output is read by people I may never meet.
The invitation
The tool is free. It takes around fifteen minutes. Probably less if this is something that has already been on your mind. Responses are consented, and the data is only used for benchmarking. There is neither a sales funnel at the end nor a gated PDF. An email address is the only thing asked for, and only so that the report itself can be sent to you.
If it is useful, use it. If it is not, tell me why, because the next iteration is already taking shape in my head.
The blacksmith, it seems, eventually made himself a proper knife.
How digitally mature is your school?
Get a free, personalised report with priorities that are actually relevant to you.
