Submission deep-dive: exactly what to include (and how judges read it) - Build with Gemini XPRIZE

Don’t forget, you’re not done when you complete your project - completing the submission form can feel like another project in itself! 

                        Remember:
                         📋 You can review the questions ahead of time by starting your submission early.   
                         ✍️ You can always continue editing your project up until the deadline. Check out some tips here.  

                        There is 1 MONTH left to enter your submission to the Build with Gemini XPRIZE Hackathon. Submissions are due August 17, 2026! 

You've got the idea and you're building. Now let's make sure none of that effort gets lost in the submission.

This is a full walkthrough of what to submit, what each piece is for, and how to make it easy for judges to say yes.
🧰 A reminder on the core requirement: your business has to be operated by AI agents and must use at least one product from Google Cloud. Everything below is about proving that clearly.

What to submit

1. GitHub repo. Link your code repository in the submission form. It can be public or private — but if it's private, you must share it with both testing@devpost.com and judging@hacker.fund, or it can't be reviewed. Make sure the repo contains everything needed to understand and run your project.
2. A 3-minute video. Show your business operating with AI live in production — the agents actually executing key decisions, not just a slideshow about them. Keep it under three minutes; anything longer won't be viewed. Host it on YouTube or Vimeo, make sure it's public, and paste the link into the form. Upload early: large files can take hours to process.
3. A written narrative (500–1000 words). Tell the story of how you built this business and how it runs day to day. Cover what your AI does versus what humans do, and the jobs and economic opportunities your business creates or enables for people beyond the founding team — both actual and potential. This is where you give judges the context the numbers alone can't.
4. Revenue evidence. Real business, real revenue. Provide a Stripe dashboard export or a bank statement, and a simple P&L — we've got a P&L template here to make this painless. Include your corporate ID if you have one. You'll also report total revenue and a month-by-month breakdown (May, June, July, August), so keep clean records as you go.
5. Expenses. Disclose your expenses incurred during the hackathon period with your P&L (template here). This includes marketing and customer acquisition spend. This is required even if it's $0 — it's how judges sanity-check your business viability.
6. Product evidence. Strengthen your case that AI is genuinely running your business in production: agent execution logs, API usage records, dashboard screenshots — anything that shows it operating continuously, not just once for the demo.
7. Customer evidence. Real customers make a real difference. Share contact info for real customers (name, email, phone) and any testimonials or feedback they've given you.

A few form questions worth prepping for

The submission form asks you to explain your build in your own words. Don't wait until deadline day to think through these:

• Which category you're competing in, and how your project creates impact within it
• Your business model — and why it's sustainable and viable (five-year goal, path to profitability, and the traction you've seen so far)
• The extent AI is live in production and executes key decisions
• Which Google Cloud product you used and how
• If you use an LLM, how you use the Gemini API for at least one LLM call
• Revenue, expenses, users acquired, and paying users during the hackathon
• Confirmation that no single customer represents more than 40% of your revenue, plus disclosure of any related-party (friends/family/pre-existing) revenue
• How your submission gets read

Judging comes down to three things, so make sure each is easy to find in your materials:

1. Business Viability — real customers, real revenue, and a model that lasts beyond 90 days
2. AI-Native Operations — how deeply AI is live in production running key decisions
3. Category Impact — does this meaningfully move the needle for real people?

The teams that score well don't just build something great — they make it obvious. Clear video, clean financials, a narrative that connects the dots.

Read the full rules and check the FAQ. Stuck on anything? Join us in Discord — we answer questions there daily.

Keep building. We can't wait to see what you build.

Michelle from Devpost