What did Boring Magic do for Sport England?

We ran discovery and rapid prototyping inside Sport England’s Data and AI Lab, working with Register Dynamics and Oxford Insights. The focus was on operational bottlenecks first, then practical prototypes for grant application review and partner report analysis.

Why was Copilot used in the Sport England project?

Copilot was chosen deliberately because it fit Sport England’s policy and workflow constraints. That made the prototypes feasible inside the organisation’s existing governance, rather than requiring a new system or a long approval cycle.

What was the main challenge in the project?

The brief started broad and the delivery conditions were messy. Staff time was limited, access to data and systems took time to come through, and the work had to stay within clear policy boundaries. The project succeeded by narrowing the problem quickly and building in public with staff.

What prototypes were produced?

The team built a prototype for reviewing grant applications and a set of approaches for analysing partner reports and setting objectives with Active Partnerships. That included prompt-based analysis, workflow automation to reduce Copilot friction, and a collaborative objectives-setting interface.

What makes this case study relevant to public-sector teams?

It shows how to make practical progress on AI without hype or risky overreach. The approach was human-first, constraint-aware, and focused on usable prototypes that fit public-sector governance, which is exactly the kind of delivery many government-adjacent organisations need.

Rapid AI prototyping at Sport England

In March 2026, Boring Magic joined Register Dynamics and Oxford Insights inside Sport England’s Data & AI Lab to explore what AI could actually do for the organisation. The brief was vague. The constraints were real. Five weeks later, there were working prototypes in staff hands.

We used discovery and rapid prototyping to produce credible, testable prototypes, not abstract recommendations.

“The initial brief was quite techy in nature: do something with data and AI. Steve’s thinking helped focus our work on solving an actual problem that the business faced.”

Simon Worthington, Register Dynamics

The context

Sport England is not a startup. It’s a public body with a mission to help people across England be active, and with that comes real governance. Sport England’s AI position statement, published in February 2026, is clear: AI can support tasks like processing applications, analysing monitoring data, and summarising reconciliations. But it cannot make decisions.

That’s the right guardrail, one we used to define the design space.

The organisation was dealing with slow, duplicative reporting, clunky legacy systems, and staff who were understandably sceptical about whether AI would actually help. The Data & AI Lab existed to test what was genuinely achievable, and to build trust with staff through working prototypes rather than promises.

Our job was to find the operational friction worth solving, and show what ‘useful’ could look like within those constraints.

What we did

We used Lean UX and rapid prototyping to move through the problem space and solution space at the same time. The approach was structured around a few clear principles.

Design with staff, not just for them. We used participatory methods so that Sport England staff shaped the prototypes, not just reviewed them. The objectives-setting prototype, for example, was heavily influenced by the Partnerships team’s own ideas.
Map constraints before building. We reviewed Sport England’s AI policy and the government’s AI Playbook before writing a line of code or designing a single screen. That’s how we ended up working inside Copilot rather than proposing a new system.
Augment, don’t replace. In two prototypes, human review was the primary pattern. AI-assisted analysis was a secondary option users could choose. Staff stayed in control of review stages and the outputs.
Solve for now, point to the near future. Quick wins grounded in real operational problems, not multi-year transformation programmes.

We ran regular show-and-tells to keep feedback loops open even when staff availability was limited.

What got shipped in five weeks

Two areas of operational friction became the focus. Both produced working prototypes.

Reviewing grant applications. The Movement Fund receives around 300 grant applications per week. Decisions are made manually, and the 12-week target is regularly stretched to 10 weeks in practice. We spoke with the team responsible for investment management and mapped the assessment pipeline: submission, assessment, recommendation, peer review, decision. The prototype used AI to pull key information out of applications to support a recommendation, with the goal of making peer review meetings more focused and reducing the number of contentious cases that needed extended discussion. Human review remained the default throughout.
Analysing partner reports and setting objectives with Active Partnerships. This became three distinct approaches: a set of prompts to help staff analyse reports inside Copilot without hitting context window limits; an automated workflow to reduce the friction of Copilot’s chat-only interface; and a prototype UI for setting objectives with partners collaboratively, based on the analysis.

Copilot was the deliberate choice throughout. It was the approved, accessible tool already inside Sport England’s ecosystem. Using it meant the prototypes were immediately feasible within existing governance, rather than dependent on new procurement or IT approval.

That’s not a compromise. Working within existing constraints helped us to move fast and deliver value early.

What made the work credible

Like many public sector organisations, the constraints were real – and worth being honest about.

With only 1 to 2 hours with staff for context-gathering, we had to be disciplined about what we explored and how we used their time.
We lost one week of prototyping time to delays getting access to accounts. A slower start, but two Lab accounts now exist inside Sport England for future work.
One prototype was paused after speaking with the relevant team. But that was a positive: the work was governed by fit and usefulness, not by a need to ship everything.

None of that derailed the project. We were still able to keep moving forward.

The show-and-tell rhythm meant feedback kept flowing even when staff couldn’t respond to messages. The scrappy-first approach meant we weren’t precious about discarding ideas that didn’t hold up. And being honest with Sport England teams about what we were doing, and what we weren’t, built the kind of trust that makes future work easier.

The teams we worked with gave us some greaet feedback:

“If you’d have asked us to spend 4 hours in a workshop at the start of this, we’d have said we didn’t have the time. But seeing what you’ve done now, we’d find 2–3 days at least to spend with you.”

That’s the model for rapid discovery and prototyping: practical progress under real conditions, not a polished case study built under ideal circumstances.

Why this matters

Most public-sector and government-adjacent organisations aren’t short of AI ambition. They’re short of working methods that suit an ambiguous brief, or work within policy and procurement constraints, keep staff in control, and still get to something useful and testable quickly.

That’s the gap we filled. Not a big, multi-year transformation, and not a strategy document for someone else to act on. A four-to-eight week discovery and prototyping project that finds what’s valuable, viable, feasible and usable, and leaves behind something the team can actually build on.

If you’re working through a similar brief, or trying to figure out where AI can genuinely help your organisation without the hype, get in touch.

21 July 2026 · artificial intelligence, AI, innovation, product, third sector, prototyping, discovery