The method in brief
- Six phases, from framing the challenge to an implementation roadmap
- AI plays an active part in all six phases, from analysing the problem to building the pitch deck
- 87% of teams have a working prototype within 48 hours
- 3 out of 5 prototypes are in use within 90 days
- Refined over more than 120 hackathons across the Nordics
From problem to prototype in 48 hours
We have run more than 120 hackathons and adjusted the method every single time. This is what we have learned: six phases, AI at every step, and follow-up after the event that actually keeps prototypes alive.
Six phases. AI at every step.
In practice the method is not linear. Teams move back and forth between phases, and AI speeds up the transitions. But the six phases give a clear structure that the facilitators keep the teams within.
Before the event: framing the challenge
2 to 3 weeks beforeGood preparation is what most often decides whether a hackathon succeeds. We interview stakeholders, look at internal data and use AI to identify the business challenges where AI has the most to offer. Teams receive a challenge brief that gives them the best possible starting point.
How AI is used
Claude analyses the stakeholder interviews and produces a prioritised shortlist of candidate challenges based on impact and feasibility.
What the phase delivers
A challenge brief with context, data and a clear problem statement
Ideation: from problem to possible solutions
Start of the eventTeams begin with a structured ideation session where AI acts as a brainstorming partner. Reverse brainstorming, SCAMPER and “how might we” questions produce a long list of ideas. AI then helps structure and prioritise them with an impact and effort analysis. Teams end up with a few promising concepts to build on.
How AI is used
ChatGPT and Claude are used to generate ideas broadly. Gemini brings in examples from other industries.
What the phase delivers
Prioritised concepts with a first assessment of impact and effort
Prototyping: build with AI tools
Most of the eventThe longest and most intensive phase. Teams build working prototypes with no-code and low-code AI tools. Our AI coaches are on hand to help with choosing tools, debugging and architecture. The aim is a prototype that can be demonstrated and that generates useful feedback, not production-ready code.
How AI is used
n8n for workflows and agents, Claude Artifacts and v0.dev for user interfaces, Cursor and Bolt.new for more complex functionality.
What the phase delivers
A working prototype that can be demonstrated live
Testing: validation with real users
Before the pitchThe prototypes are tested on potential users or internal stakeholders. Test scenarios are prepared in advance so the feedback is useful within a short time frame. AI helps analyse the feedback quickly and produce a prioritised list of the most important improvements before the pitch.
How AI is used
Claude analyses the raw feedback and sorts it into must-have and nice-to-have improvements. Maze is used for structured user testing.
What the phase delivers
A prioritised feedback report and an improved prototype
Pitch: presenting to decision makers
End of the eventEach team pitches its prototype, business case and implementation plan to a panel with the authority to make decisions. AI is used to create pitch decks and visual mock-ups. The pitch covers the problem, the solution, a demo and what it will take to bring the prototype into everyday use.
How AI is used
Gamma for pitch decks, Midjourney for interface mock-ups, ChatGPT for structuring the story and the business case.
What the phase delivers
Pitch deck, live demo and a draft implementation plan
After the event: from prototype to everyday use
After the hackathonWe deliver a detailed implementation roadmap for every prototype. It covers technical architecture, resources, who is responsible for what, success criteria and a timeline. We also offer follow-up sparring and workshops that hand over the skills, so your organisation can keep building on its own.
How AI is used
Notion AI and Claude are used to structure the implementation roadmap. HubSpot for tracking return on investment and follow-up.
What the phase delivers
Implementation roadmap, a plan for handing over skills and a framework for measuring return
The principles behind the method
The problem first. Always.
We do not care which AI tools are fashionable. We start with your business problem and then choose the tools that solve it best. That is a big part of why 3 out of 5 of our prototypes are in use within 90 days.
Mixed teams beat expert teams
The best hackathon prototypes we have seen did not come from purely technical teams. They came from teams with one person who knew the business, one with deep subject knowledge and one with a healthy curiosity about AI.
A hackathon is the cheapest place to be wrong
We actively encourage teams to try ideas that might not work. A prototype that fails fast is worth far more than a project that drifts along for three quarters and then fails.
Document while you think
We use AI to capture decisions, the thinking behind ideas and architecture choices as the work happens. It sounds dull. It is also what makes the work after the event go much faster.
How an AI hackathon differs from a classic hackathon
Classic hackathon
- Focus on coding and technical implementation
- You need a technical profile to take a meaningful part
- Prototypes are rarely ready for everyday use
- Success is measured in creativity and wow factor
- Teams are left to handle implementation themselves
AI hackathon (our method)
- Focus on business value: we solve problems that actually cost you something
- Every employee can build prototypes with AI tools
- 87% of teams have a working prototype within 48 hours
- Success is measured in potential return and readiness for implementation
- Every prototype gets a structured implementation roadmap
See the method in practice
The best proof that the method works is what our clients have built. See the formats we offer, and meet the facilitators who run them.