How AI use is judged
Judges reward thoughtful use of AI: clear prompting, iteration, sensible tool choices and the human judgement you applied. Using more tools earns nothing by itself.
Judging criteria
100 POINTS.
ONE RUBRIC.
Every entry is scored by people, against the same published rubric, before judging begins. A simple product that works beats a flashy mockup.
| Criterion | Points |
|---|---|
| Problem and impactIs the problem clear, and is the solution genuinely useful to the people who have it? | 10 |
| Creativity and originalityIs there independent thinking or a distinctive approach? | 10 |
| Effective use of AI and promptingWas AI used thoughtfully? Can the entrant explain prompting, iterations, tool choice and human judgement? Tool count alone earns nothing. | 20 |
| Product functionality and executionDoes the product actually work? Are the core features implemented and usable? | 20 |
| Reliability and interaction qualityDo buttons, forms, links, navigation and validation work without major errors? | 10 |
| UI/UX design and accessibilityIs it understandable, responsive, usable and accessible? Simple, useful designs can score highly. | 10 |
| Potential for further developmentIs there a credible path to more utility and reliability? | 10 |
| Clarity of video pitchDoes the pitch clearly explain the problem, solution and main features within two minutes? Editing quality earns no points. | 5 |
| Documentation and demo readinessCan judges access and evaluate the project easily? | 5 |
| Total | 100 |
Each criterion is scored from 0 to its maximum points; the nine maxima add up to 100. Judges score with evidence-based comments. Video editing quality is not judged.
The video pitch
Up to two minutes, screen-recorded — no need to show your face. Explain the architecture, show the code if you can, and demonstrate the main features. Editing quality earns no points.
Fair, consistent scoring
Judges score only the entries assigned to them, give evidence-based comments for each criterion, and declare conflicts of interest. The rubric is locked once scoring starts.
Ties
Ties are broken by the higher score for functionality and execution, then effective use of AI, then problem and impact. See section 9 of the rules.
Read the full judging rules, or get practical tips in How to Build.