An AI hackathon facilitator frames the problems before the event, coaches teams in the tools on the day, runs the demo-and-judging process, and packages what works afterwards so it can be reused. Day Seven facilitates internal AI hackathons for UK business teams of 10 to 40 people, in three formats from one day to the seven-day Day Seven Sprint, on-site anywhere in the UK or live online.
Four things, and the first and last are the ones companies most often miss when they run it themselves.
Before the event: works with your team leads to turn a list of "AI ideas" into five to eight problem statements that are specific, have data available, and can be judged by the person who does the job. Agrees the guardrails with IT and data protection. Forms teams and checks tool access.
On the day: briefs the room, sets the checkpoints, and coaches. Coaches walk the floor, get teams unstuck in minutes, and stop them building the wrong thing well. They push for tests on real examples early and for a demo a sceptic could watch.
Demo day: runs the demos to a fixed format, supplies unseen examples for live runs, and chairs the judging panel against published criteria.
Afterwards: writes up each build (problem, build, test results, guardrails, what it would take to run properly) and hands the sponsors an adopt, pilot or park recommendation for each.
Run it yourself if you have internal champions who are confident in Copilot, ChatGPT, Claude or Gemini, a problem list that already exists, someone with two or three weeks of preparation time, and a sponsor who will follow through. Our playbook for running an internal AI hackathon covers everything you need.
Hire a facilitator if any of those is missing. In practice that means: it is your first hackathon; nobody internally has run one; the problem list is really a wish list; you want teams coached by people who use these tools every day; or you want the seven-day format, where the aim is a tested build rather than a demo. A facilitator also removes the awkwardness of a colleague judging a colleague.
| One-day hack | Three-day build | Day Seven Sprint (seven days) | |
|---|---|---|---|
| Team size | 10 to 40 people, in teams of 3 to 5 | 10 to 40 people, in teams of 3 to 5 | 10 to 40 people, in teams of 3 to 5 |
| What you get at the end | Rough prototypes for each problem, a shortlist of what deserves a pilot, and a room of people who have built something with AI | Two or three prototypes worth piloting, each tested on real examples, with a written package and an adopt/pilot/park recommendation | One or two working builds, tested by the people who do the job, measured against the current process, documented and ready for sign-off |
| Who should attend | Any team, mixed functions; good as a follow-on to training | Teams with a defined process problem and a business owner who can attend demos | A team that owns a real, costly problem, plus the people who will run the build afterwards; IT and data protection involved before day one |
| Prep needed | Problem statements agreed two weeks ahead; tool access checked; guardrails circulated one week ahead | Same, plus a one-hour briefing call per team the week before | Same, plus a half-day problem-framing session with business owners two weeks ahead and access to relevant documents or data sources set up in advance |
Most companies start with the one-day format, then run a three-day build or a Sprint on the problems it surfaces.
Problem framing before the day. We run the problem-framing session with your team leads and write the problem statements with them. This is the work that decides whether the hackathon produces anything.
Trainers who coach in the tools you use. Our trainers work in Microsoft 365 Copilot, ChatGPT, Claude and Google Gemini, and in the prompt engineering and workflow automation layer around them. They coach in whichever your company licenses. Sessions are lab-style: minimal slides, people build with their own tools on their own examples.
Guardrails. We bring a standard set of hackathon guardrails covering data classification, approved tools, access and sign-off before anything goes live, and adapt them with your IT and data protection people. Where you have an AI policy or an ISO/IEC 42001 management system, the guardrails are drawn from it.
Demo-day structure. Fixed-length demos, live runs on examples the judges supply, published criteria, a chaired panel that includes the business owner of each problem.
A packaged pattern for reuse. Every build worth keeping is written up in a standard format so it can be handed to engineering, copied by another team, or put through a pilot. This is what turns a hackathon from an event into the Prove stage of an AI programme.
A sponsor who will attend demo day and make the adopt/pilot/park decisions. Team leads who can spend two hours in the problem-framing session. Participants who are released for the full event; part-time attendance breaks teams. Enterprise licences for the tools in play (consumer accounts are excluded). A contact in IT to set up access, and a contact in data protection to agree what data can be used. A room with power, wifi and wall space, or a video platform for online events.
On-site anywhere in the UK. We are based in Leeds and regularly work in Manchester, Sheffield, York, Newcastle, Liverpool, Birmingham, Nottingham, Bristol, Cardiff, London, Glasgow, Edinburgh and Belfast. All three formats also run live online, with teams in breakout rooms and coaches moving between them. For the seven-day Sprint, a mix works well: framing and demo day in person, build days online.
Price depends on format and team size. A one-day hack for twelve people and a seven-day Sprint for forty are different pieces of work. Tell us what you have in mind, including the format, rough numbers, and whether you want it on-site or online, and we will send a proposal.
Someone who prepares, runs and follows up an internal AI hackathon: framing the problems, coaching teams in the tools, chairing demos and judging, and packaging the results. The role exists so the company's own people can concentrate on building rather than organising.
Around four weeks for a one-day hack, to allow for problem framing, team formation and access setup. Six weeks is more comfortable for a three-day build or a Sprint, mainly because of the data and access preparation.
Yes. All three formats run live online with teams in breakout rooms and coaches moving between them. For a Sprint we often do framing and demo day in person and the build days online.
Microsoft 365 Copilot, ChatGPT, Claude and Google Gemini, plus the automation tools around them. We work with whatever your company has licensed on enterprise terms rather than bringing our own.
Not necessarily, but it helps. Teams that have had a hands-on workshop get further on the day. Many companies run a half-day or full-day workshop first and use the hackathon as the follow-on.
Ten to forty, in teams of three to five. For more than forty we run parallel tracks or two events, each with its own judging.
Every build worth keeping is written up in a standard package, and sponsors get an adopt, pilot or park recommendation on each. Anything adopted needs a named owner and a sign-off before it touches live work; we help set both up.
A workshop teaches people the tools on their own examples. A hackathon puts teams on a real problem for a fixed time and ends with a demo and a decision. The AI hackathons page covers how the two fit within the Day Seven method.
Tell us the format you are thinking of, the size of the group, and the problems you have in mind. We'll get back to you within one working day.
Get a hackathon proposal