An NHS Copilot 'Prompt-A-Thon' : From Fuzzy Ideas to a Working Prototype
During 2025, one of the most rewarding parts of my role as Digital Navigator has been taking part in the NHS‑wide trial of enterprise-level Microsoft Copilot.
It’s been fascinating to watch a large‑scale, enterprise‑level innovation programme unfold from the inside—seeing how different organisations approach experimentation, as well as supporting colleagues at the Jean Bishop Centre, in Hull, as they begin to explore what’s possible with AI.
As part of the country-wide NHS 'Copilot Crew', I joined a virtual 'Prompt‑a‑Thon' in November. The format was simple: groups of four strangers, three hours, and one shared challenge. Identify a real NHS problem and explore how Copilot could help transform it. Simple in theory… but in practice, it required a lot of listening, patience, and gentle steering.
But before I go any further... What an earth is a Prompt-A-Thon?
.A Prompt‑A‑Thon is a guided innovation session where small groups use AI to unpack a real operational challenge and shape early prototypes. It's a term inspired from the coding world - a hack-a-thon- where groups of coders meet to solve problems with code! However a prompt-a-thon is designed for non‑coders and is focused on ideas and insight — not technical skills. It's a great way to brainstorm and wonder how practical problems could be innovated through the use of Artificial Intelligence.
My three group members were all Copilot novices, with no experience in developing digital solutions or shaping early‑stage ideas.
By contrast, I’ve spent years in IT development and just as many years training others, so my natural instinct is to sift through fuzzy requirements, spot the patterns, and carve out a clear path forward. That skillset came in handy.
We used our enterprise version of MS365 Copilot to transcribe our two‑hour, often meandering conversation thru Teams.
From there, Copilot helped us highlight the themes and pain points that were most suitable for an AI‑supported solution. One stood out clearly—a real, everyday problem faced by one of the team members in her ICB role.
We agreed to explore an agent‑based solution. Over the next couple of weeks, I spent time with my prompt-a-thon team member to understand the real‑world workflow, then used Copilot to map how an agent could streamline it.
Asking AI how to best use AI always feels a bit weird but is a very effective way to develop strong and detailed prompts.
Together, we shaped a reasonable working prototype that is practical, testable, and genuinely useful. It’s now in her hands to refine further, and it has the potential to save significant time for her team.
Sadly, we didn't win 🙁 BUT...
What struck me most was how quickly confidence grew in the group once they could see a tangible outcome.
Very often the biggest barrier to making progress with practical AI use isn’t the technology—it’s helping people to 'see' the possibilities and to move from abstract possibility to a specific practical first step.
I hope our team member can take what we prototyped and find support to get it turned into a 'real' scale-able solution.
And I'm excited to continue enabling colleagues to navigate the space between curiosity and capability. It's in small steps like this where digital innovation becomes real.
Cheers, Sue x