Imagine opening a project and realizing: “I did not know AI could do that.”
That is the kind of discovery AI OpenBuild is built around. For a few weeks, take a capability that interests you—language, vision, sound, prediction, generation—and see where you can push it. Give it an interface. Connect it to a workflow. Put it in a game. Make it useful, strange, expressive, or surprisingly simple.
The direction is yours. A tiny tool for a very specific frustration belongs alongside an ambitious experiment. So does an interactive story, a learning companion, a new way to explore data, or a physical prototype that responds to the world. The common thread is that AI does something meaningful in what you make.
Pick a possibility. Make it tangible. Let someone try it.
You can build with a model API, run an open-source model, train something small, or combine approaches. Use the stack you know or learn one as you go. There is no required vendor, no required industry, and no paid service you need to buy. Solo builders and student teams of up to five are welcome, subject to the event eligibility requirements.
When it is time to share, show us the actual experience: what a person does, how the AI responds, and why that response matters. Tell us about the decisions you made and the edge cases you found. An honest prototype with a clear purpose gives us much more to explore than a long list of promised features.
Bring the idea you keep putting off—or discover one by building. We are curious to see what you make.
Requirements
What to build
Create a project that uses AI as a meaningful part of its functionality or experience. Any topic, model, language, or platform is welcome. Show a working prototype or a clearly labeled interactive proof of concept.
Submission checklist
- Explain the idea: who it is for, the problem or creative goal, and the main features.
- Show the AI: identify the model or AI method, describe its role, and provide representative inputs and outputs.
- Demonstrate your build: include screenshots and a short video or visual walkthrough. A public demo, viewable prototype, or repository is strongly encouraged.
- Document your work: list your stack, model providers, datasets, open-source components, and AI assistance. Distinguish pre-existing components from work completed during this event.
- Include limitations: describe at least one failure case or constraint and how you tested the main feature.
- List your team: one to five people, with each member's role.
Make review links accessible without requesting permission where possible. Never include API keys, private credentials, or private personal data. Submit on Devpost by December 10 at the time shown in the event timeline.
Prizes
1st Place
Runner-Up Recognition
Eight standout eligible projects receive runner-up recognition. This is non-cash recognition; no monetary prize is attached.
Devpost Achievements
Submitting to this hackathon could earn you:
Judges
AI OpenBuild Team
Organizing Team
Judging Criteria
-
AI Integration & Technical Execution (30%)
How meaningfully is AI integrated, and how well does the prototype work? Consider the model or method choices, implementation, and the team’s contribution. -
Originality & Problem Framing (25%)
How thoughtful or original is the approach? Does the team clearly frame its intended problem or creative goal? -
Usefulness & User Experience (25%)
How well does the project serve its intended users or deliver its intended experience? Consider usability, accessibility, and practical value. -
Testing & Communication (20%)
How clearly are the build, AI behavior, limitations, credits, and testing demonstrated? Consider representative examples and failure cases.
Questions? Email the hackathon manager
Invite others to compete
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.