AI Lab: Seed Agents Challenge with Beta Fund: $70K Cash Prizes
🏆 AI Lab: Seed Agents Challenge
The "Agentic" Frontier: Experience the Future of AI Video
The current AI video is a passive loop. The Seed Agents Challenge is changing that. Beta University invites the world's elite engineers to move beyond "wrappers" and build autonomous agents that see, reason, and generate high-fidelity video in real-time.
Official Portal: Betahacks.org
💰 Prize Pool & Incentives
1st Place: $8,000
2nd Place: $4,000
3rd Place: $2,000
Beta Fund 2 Investment Checks up to $50,000
Track Winners: $1,000 (6 Winners)
Travel Sponsorship: Top 20 teams receive covered flight and hotel for the May 2nd finale in Silicon Valley.
Pitch to investors: Live feedback from partners at a16z, NEA, Mayfield, OpenAI, and Anthropic.
🟢 Competition Tracks
Track 1: Vertical AI Video Agents Automate end-to-end video for real estate, product demos, or compliance training. Build domain-specific agents where video is the core software layer.
Track 2: AI-Powered Content Automation Build "always-on" content engines. Create automated pipelines (Input → Generation → Distribution) for marketing and e-commerce at scale.
Track 3: AI DevTools for Video Build the infrastructure. Create SDKs, APIs, or evaluation systems that allow other developers to use video as a core primitive.
Track 4: Physical AI + Simulation Bridge the digital and physical. Use video generation to create synthetic training environments for robotics and autonomous systems.
Track 5: Most Creative (Multimodal Focus) Push the boundaries of Seedance 2.0. We are looking for the most "out of the box" applications combining multimodal inputs with generative video.
🛠 Technical Requirements & Video Structure
The Core Rule: All video generation must run on BytePlus Seed models. Third-party LLMs are allowed for orchestration, but external video tools (Sora, Runway, Kling) are strictly prohibited.
Required Toolset (Use at least one):
Seed 2.0: Analysis, prompt enhancement, and script generation.
Seedance 2.0: AI video generation (T2V, I2V).
Seedream 5.0: Image generation (Keyframes, storyboards).
Seed Speech & OmniHuman: Voice synthesis and digital human generation.
Submission Video Format (2 Minutes Total):
Live Demo (0:00-1:00): Show the prototype in action. Lead with the experience; let judges see the output before you explain the problem.
Technical Architecture (1:00-1:30): System design overview. Show how Seed models connect in your pipeline.
Vision (1:30-2:00): Future evolution beyond the hackathon.
📅 Timeline
April 19: Early Submission Deadline (Priority for flight/hotel sponsorship)
April 27: Hard Deadline for all submissions via Betahacks.org
May 2: In-Person Finale (Silicon Valley) — 20 selected team demos and investor networking.
Join our Discord for updates, API keys, and credit release