TinyFish $2M Pre-Accelerator Hackathon

TinyFish $2M Pre-Accelerator Hackathon. Build Applications That Do Real Work on the Web. Every AI tool being built today needs live web data. LLM agents, copilots, RAG pipelines, autonomous workflows all of them hit the same barrier: the web was designed for human eyes, not autonomous agents. It's locked behind dynamic rendering, multi-step workflows, and anti-bot systems. And it's also become far too vast, too complex, and too fast-moving for even humans to keep up with billions of pages, constantly shifting data, endless forms, and workflows that eat hours of skilled labor every day. AI agents are the next trillion users of the web autonomous software that does real work on behalf of humans. And TinyFish is the infrastructure that makes it possible. TinyFish is launching a global hackathon in partnership with HackerEarth to find the boldest builders ready to turn the live web into an executable runtime for AI agents. This isn't a toy hackathon. This is your gateway to The TinyFish Accelerator, a high-intensity, YC-style virtual accelerator program, backed by a $2M seed pool from Mango Capital. Here's the deal: Build an application powered by the TinyFish Web Agent API that does real work on the web Record a raw 2–3 minute demo of your agent doing real work on the live web Post it publicly on X (Twitter) tagging @Tiny_fish Submit your project on HackerEarth 🏆 Top 3 winners take home cash prizes AND a Golden Ticket — direct entry into Phase 2 (The Sprint) of the TinyFish Accelerator, skipping the application queue entirely and landing in a 2-week intensive sprint alongside the best agent builders in the world. Why This Hackathon Is Different 🎫 The Golden Ticket Winners don't just walk away with cash. They receive a Golden Ticket — direct entry into Phase 2: The Sprint of the TinyFish Accelerator, bypassing the entire application and review queue. What Phase 2 (The Sprint) looks like: 2-week intensive with the full accelerator cohort Full Partner Stack unlocked — direct engineering support and API access from 15+ partner companies: Google for Startups Cloud Program, v0 by Vercel, ElevenLabs, Fireworks.ai, MongoDB, Daytona, Composio, Dify, Axiom, Alguna, AgentOps, AgentMail, AG2, Superdesign, Creators Corner, Parasail, Yotta Labs, and more. Weekly Business Mentorship sessions on challenges of an early stage business with world-class mentors Free API keys from TinyFish Private engineering Discord with direct access to TinyFish core team 🎯 And Then: Demo Day Sprint participants advance to Demo Day in April, a studio-grade virtual livestream where you pitch your product directly to Robin Vasan of Mango Capital (lead investor behind HashiCorp, Netlify, and InfluxData) and the TinyFish executive team. Funding decisions from the $2M seed pool are made immediately. No committees. No months of follow-ups. You deal directly with the decision-maker. Build an autonomous web agent using the TinyFish API The Problem While LLMs can generate text and code with remarkable fluency, getting AI to actually do work on the live web navigating complex UIs, managing sessions, executing multi-step transactions remains one of the hardest unsolved problems in AI infrastructure. Most 'AI agent' projects today are wrappers a chat layer over a database, a summarization tool with a nice UI. Nothing that actually goes out into the world and performs labor. The Challenge We are challenging you to build something that does real work. Your mission: build a real-world application powered by the TinyFish Web Agent API that performs high-value, multi-step operations on the live web. Your application should use TinyFish's agentic infrastructure to navigate real websites, manage state, and execute meaningful workflows that solve a genuine business problem. What we're looking for: Applications that embrace the complexity of the web, multi-step workflows, dynamic UIs, session management, pagination, pop-ups, and form fills Products that solve a real pain point with clear business value think workflows that currently cost companies hours of manual labor Software that could become a real revenue-making business, not just a demo What we will reject: ❌ Simple text summarizers or chatbots sitting on top of a database ❌ Basic RAG applications that don't interact with the live web ❌ Thin wrappers that add a UI layer over an existing API ❌ Anything that doesn't actually need browser infrastructure to function The test is simple: If your application can be built without a web agent navigating real websites, it's not a fit. We invest in the Agentic Web software that goes out into the wild internet and gets things done.