Inside the world's most elite student hackathon – Full Documentary on Stanford Tree Hacks 2026

Extremely competitive admission process

Stanford received 15,000 applications for only 1,000 spots, requiring even Stanford students to apply, with successful candidates highlighting research experience and prior hackathon wins.

Massive institutional backing

Stanford covered travel costs for all 1,000 participants to fly to Palo Alto, with major sponsors including Anthropic and OpenAI contributing to a $500,000 prize pool.

Multi-track targeting maximizes winning odds

Winners recommend designing projects that intersect multiple prize categories simultaneously rather than focusing on a single track, allowing one demo to compete for grand prizes and sponsor-specific awards.

Missionary versus mercenary mindset

Veteran mentors advise participants to decide early whether they are 'missionaries' building for passion or 'mercenaries' targeting specific internships or prizes, as this determines networking and project scope strategies.

Speed over perfection

Experienced hackers emphasize not overthinking ideas and prioritizing functional prototypes within the 36-hour window over architectural perfection.

Multi-agent drug repurposing platform

One team built a system using seven specialized AI agents—including a 'skeptic agent'—that debate to identify shelved drugs that could treat new diseases, addressing the 70% of drug failures due to logistical rather than efficacy issues.

Cross-device Apple ecosystem AI

Team 'Iris' developed a unified AI assistant enabling seamless interaction between iPad and Mac, such as drawing math problems on an iPad to receive hints on a MacBook screen, surpassing Siri's siloed limitations.

Local-first AI video editing

The 'Flow Cut' platform processes video generation and editing using local compute resources rather than cloud APIs, significantly reducing latency and cloud costs while maintaining API configurability.

Automated research benchmarking

Participants created an agentic platform that automatically implements algorithms from research papers, runs comparative analyses on custom hardware, and researches additional algorithms to benchmark against new code submissions.