How to Win a Hackathon with AI: Use TinyFish to Supercharge Your Project (2026)

Most hackathon projects look the same: a chatbot, a nice UI, and a demo that works on the data the team typed in that morning. Judges see dozens of them.

The projects that win do something different. They work on the real, live web: they check today's prices, read real pages, click through real sites and catch changes as they happen. That's what TinyFish is for. It's one API that lets your AI agent search, read and act on any website, so your hackathon project can do things judges can actually watch happen.

We've seen it up close. At UCL Build Night, students shipped 30 working web agents in 90 minutes, judged live. At Columbia, a student built an internship tracker in an afternoon and walked away as our first campus winner. Here's how they did it, and how you can too.

Last checked: October 2026.

TL;DR

• Use live web data. A project that pulls real information from real websites beats one running on fake data, every time.
• Add TinyFish to whatever you're building: search the web, read a page, send an agent to click through a site, run a browser session, or monitor a page for changes. It's all one API.
• Plan your time: problem in 10 minutes, a working core in 45, polish and demo for the rest.
• Judges want four things: a real problem, live data, a 30-second demo and a clear "here's how it works".
• Build for free: students get $35 in their TinyFish wallet at tinyfish.ai/students, and you can submit your project as a paid bounty after the hackathon.

Why live web data wins hackathons

Judges are looking for one moment: the demo where something real happens.

• A price drops on a real store and your app catches it.
• An agent fills in a real form, or finds a real internship that went live this morning.
• Your tool proves a "was £200" sale is fake by checking the store's own price history.

That moment is hard to fake, and it's what separates "cool idea" from "this actually works". Most teams never get there because getting reliable data off real websites is the hard part: pages change, sites block bots, and there's no API for half of what you want. TinyFish takes care of that part, so you can spend the hackathon on your idea.

6 ways to add TinyFish to any hackathon project

TinyFish is one API with everything behind it, so you can mix and match these in the same project.

1. Search the web for fresh results

Get ranked, structured search results for any query, so your app starts from what's true today, not what a model remembers.

Example: a scholarship finder that searches for this month's open scholarships in your field.

2. Fetch any page as clean content

Turn any URL into clean text your AI can read. No scraping code, no fighting with HTML.

Example: a lecture-notes helper that pulls the latest course page and summarises what changed.

3. Send an agent to do a task on a site

Describe the goal in plain English, and an agent clicks, scrolls and fills things in on a real website, then returns the result.

Example: a "cheapest train ticket" agent that checks the booking site for your route and dates.

4. Run a real browser session

For anything that needs a logged-in or multi-step flow, control a remote browser yourself, step by step.

Example: a tool that walks through a university portal to collect all your deadlines in one place.

5. Monitor a page for changes

Watch a page and get told when something changes. This is perfect for a "wow" moment in the demo.

Example: an internship tracker that alerts you the minute a company's careers page adds a new role.

6. Plug TinyFish into Claude or ChatGPT

No time to build a full app? Connect TinyFish to Claude or ChatGPT and build your project as a conversation with a real tool behind it. See 10 student use cases for TinyFish in Claude and in ChatGPT.

Real hackathon wins with TinyFish

UCL Build Night: 30 agents in 90 minutes

At UCL Build Night with the UCL AI Society, every team built a web agent and the TinyFish team judged them live.

• 🥇 Fake Sale Detector checks a product's price history on the Wayback Machine to prove whether a "sale" is real. It caught a fake "was £200" price live during judging.
• 🥈 Pub Crawl Planner plans a walking route and proves each pub has what you want, straight from the pub's own website.
• 🥉 Fishies is a pixel-art sea-life guide that finds where to see, eat and cook every creature.
• One honourable mention was built entirely on a phone.

Columbia: a winner in one afternoon

At the Columbia AI Club workshop, Elad Hirsh built Internship Signal Tracker. It uses TinyFish in ChatGPT and TinyFish Monitor to find, verify and track real Summer 2027 internships, and became our first campus winner.

What they had in common: a problem people actually have, real data, and a demo that worked live.

Your 90-minute hackathon game plan

1. Minutes 0–10: pick one real problem. Something you or your friends deal with: deals, deadlines, internships, housing, travel. Write it in one sentence.
2. Minutes 10–20: find the data. Which websites have the answer? That's where TinyFish goes.
3. Minutes 20–65: build the core. Get one end-to-end flow working: input in, TinyFish does the web part, result out. Ugly is fine.
4. Minutes 65–80: make it demo-able. One clear screen, one example that always works, one "watch this" moment.
5. Minutes 80–90: rehearse the pitch. Problem, live demo, how it works. Under 2 minutes.

New to agents? Do our 15-minute first web agent guide before the hackathon so you're not learning the basics on the clock.

How hackathon judges score AI projects

Every event is different, but in our judging (and most AI hackathons) these four things decide it:

• A real problem. Would someone actually use this next week?
• Live data. Does it work on the real web, right now, or on a fake demo file?
• A 30-second demo. Can a judge see it work without a long setup?
• You can explain it. What does the AI do, what does TinyFish do, and why does it work?

Bonus points: it's useful beyond the hackathon, and you shipped it. A live link beats a slide every time.

15 AI hackathon ideas you can build with TinyFish

Beginner (one person, a few hours)

1. Fake sale detector: prove whether a discount is real using price history.
2. Scholarship finder: search and list scholarships that are open right now.
3. Event finder: pull this week's free events near campus into one list.
4. Restock alert: monitor a sold-out product and alert when it's back.
5. Menu checker: check nearby restaurants' menus for your dietary needs.

Intermediate (a team, a full hackathon)

1. Internship tracker: find, verify and monitor new roles on careers pages.
2. Cheapest flight or train finder across booking sites for your dates.
3. Student housing scout: compare listings and flag fair prices.
4. Research paper alert: watch journals for new papers on your topic.
5. Course deadline sync: pull deadlines from your university portal into your calendar.

Advanced (to really stand out)

1. Policy change monitor: catch changes to terms, prices or rules on any site.
2. Competitor price tracker for a small business.
3. Government tender or grant tracker by sector.
4. Accessibility checker: test real pages and suggest fixes.
5. Fact-checker: check a claim against official sources and cite them.

Want code to start from? Many of these have open-source versions in our 25 AI project ideas with source code.

After the hackathon: get paid for your build

Don't let your project die in a GitHub repo. Submit it as a bounty on TinyBounties: every approved build earns a gift card ($50 for an agentic app, $20 for a Skill, $10 for a prompt).

To submit you need:

• a link to your working build and how it uses TinyFish
• a post about your build on LinkedIn and X, tagging TinyFish (both required)
• a share in #showcase on the TinyFish Discord

Want TinyFish at your next campus hackathon? Request an event on TinyBounties. And if you love running events, the TinyFish Ambassador Program is for you.

Start building for free

Students get $35 in their TinyFish wallet. Sign up at tinyfish.ai/students with your school email and a password (not "Continue with Google"), and the credits are added automatically. More on the program: TinyFish Student Program.

Keep reading

• Web Scraping Project Ideas with Source Code (Python and No-Code, 2026)
• What Is an AI Agent? Explained for Students (and How to Build One)
• How to Make Money Building AI Products as a Student (2026)
• How to Host a Tech Meetup: Step-by-Step Guide + 20 Event Ideas (2026)
• How to Organize a Hackathon: Step-by-Step Guide + Free Checklist

Follow TinyFish for new guides and student builds: X (@Tiny_Fish) · LinkedIn (TinyFish) · Instagram (@tinyfish_ai)