Aaron Leary has been building with AI for longer than most people have been thinking about it. As founder and CEO of Bedrock Learning - a UK EdTech business that provides schools with a scalable way to massively increase reading and writing ability in curriculum subjects - he is not the typical ground-zero buyer. He does not need a map to the basics. The value of the assessment for him was something different: a structured prompt to re-examine the parts of his business he is not in day-to-day, the sales team and the marketing function, through an AI-efficiency lens.
For Aaron, the session landed in a specific place. Moving a business forward with AI takes consistent, deliberate effort - only a couple of people will ever actually drive it, week in, week out. As CEO, that responsibility is his: knowing what is possible is not enough if he is not actively leading the people around him through it.
The diagnosis
The assessment looked across the business and surfaced three custom agent opportunities. Aaron already understood the technical landscape; what the session gave him was a structured view of inefficiency in areas he was not running himself, and the push to act on it.
It worked as a catalyst in a more direct way too. Within weeks, Aaron's team had built their own internal agent: Bedrock Smarty Pants, a sales-prep tool that takes the roughly 1.5 hours of government-data research an account manager would spend preparing for a school visit and turns it into a one-click report. Bedrock built that themselves, inspired by the session. We did not build Bedrock Smarty Pants.
Of the three opportunities the session surfaced, Aaron commissioned one immediately. The other two remain on the table. He expects to revisit the report before the academic year is out to check coverage and look for further gains.
For business leaders at ground zero who need a starting point, the price point is justified and valuable.
The recommendations
The one opportunity Aaron commissioned straight away was an impact report generator. Bedrock's account managers are responsible for producing impact reports for each of their partner schools - documents drawing on reading-test data to show how the platform was performing. The reports were being produced manually, quality was variable, and the deadline was hard: hundreds had to go out by the end of May. Manual production, at 2-3 hours per report for one person back-to-back, would have missed it.
Aaron trusted HoursBack with the build because he had already seen the depth of our report-generation work first-hand. That, alongside a clear, structured, quick way of working through the scoping stage, gave him the confidence to commission it.
The build started from scratch and went through multiple rounds of QA - careful, collaborative iteration against real data and edge cases. What we contributed to those rounds went beyond the original brief: we flagged issues with the clarity of chart legends ourselves, made improvement suggestions that had not been asked for, and delivered on time.
You weren't just going through the scope blindly. You actively helped to improve the outcome.
The outcome so far
The generator ran hundreds of school impact reports in approximately 60 seconds. Manual production, at 2-3 hours per report for a single person, would have been the best part of 10 weeks of work - and Aaron confirmed the 10-week figure is conservative. One person, back-to-back, before any leave or sickness, could not have done it in time.
Every school got a consistent, polished report that met the hard end-of-May deadline manual work would have missed. Schools that had not been using the standardised reading test showed up as empty charts in the data - which is exactly what those conversations were designed to surface. Renewal-season meetings got booked. Account managers were delighted.
Two of the three opportunities from the original session are still on the table. Aaron expects to revisit the report before the academic year ends. The work continues.
The takeaway
The pattern here is different from the assessment-only engagements. Aaron's engagement ran the full ladder: an assessment that reframed where the gains were, a bespoke build that turned the biggest of those gains into a working asset Bedrock owns and can re-run, and two more opportunities still in play.
The build is not a one-off service. The generator is Bedrock's to keep. Each time it runs, it produces hundreds of reports in 60 seconds. That is the value of a bespoke build over a tool recommendation: when the report flags an opportunity that is genuinely worth building, the build becomes a permanent part of how the business works.
A good price for a piece of functionality I can repeat and use myself. A worthwhile investment, without doubt.
Case study subject: Aaron Leary.