Jury member 1
Technology & product leadership
Focus on engineering quality, architecture, and applied AI in products.
CURSOR × DEEP AGENTS SPEEDRUN~150 ATTENDEES
CURSOR
DEEP AGENTS SPEEDRUN
Build a working AI agent in ~3 hours using Cursor and Deep Agents. Deploy it publicly. Score highest in the automated eval at 9pm.
Register
TEAMS OF 4–5 OR SOLO / CASH + CREDITS
Scroll
Partners
Our Sponsors
Leading Partner
Co-organizers
CursorSponsors
CursorCursorCursorCursorCursorCursorMedia & Social Partners
CursorCursorCursorCursorCursorCursorCursorPRIZES
Prizes
0
+ Bonus credits · Best Architecture · Best UI/UX · Best .cursorrules
2nd Place
1,500
+ Bonus credits · Best Architecture · Best UI/UX · Best .cursorrules
2
1st Place
2,500
+ Bonus credits · Best Architecture · Best UI/UX · Best .cursorrules
1
3rd Place
1,000
+ Bonus credits · Best Architecture · Best UI/UX · Best .cursorrules
3
SCHEDULE
Tonight's Schedule
Demo 4–6pm · Build 6–9pm · Finale 9–10pm
4:00 – 6:00
Demo & Presentation
Cursor tips, architecting from scratch, Deep Agents overview
* Deploy early — a working URL at 7:30pm beats a crashed one at 8:59pm.
THE JURY
Meet the Jury
Experts will evaluate your projects at the hackathon finale
Jury member 2
Product & growth
Building AI-powered products end-to-end: strategy, growth, adoption.
Jury member 3
Industry innovation lead
Evaluating business model, scalability, and commercial potential.
Jury member 4
Cloud & infrastructure
Evaluating technical solutions, architecture, and cloud usage.
Jury member 5
Financial services product
Evaluating products, services, and AI/ML usage.
Jury member 6
Entrepreneur & community builder
Experience across SaaS, integrations, and developer communities.
MENTORS
Meet the Mentors
Mentors help teams move faster and make their final pitch stronger
Mentor 1
Product management
Mentor 2
AI & data
Mentor 3
Product management
Mentor 4
Sales & AI
Mentor 5
Product & growth
Mentor 6
Web engineering
Mentor 7
Backend engineering
Mentor 8
Product management
Mentor 9
Software engineering
Mentor 10
Software engineering
JUDGING
How You're Scored
10-task automated eval at 9pm + UI/UX bonus from roving judges
Total
100%
Gauntlet Score25%
10 automated tasks fire at your agent URL at 9pm. Each passes or fails — no subjectivity.
Task Range25%
Tasks range from basic (respond coherently) to complex (complete a full 5-step chain in one session).
Deep Agents20%
Agent must use create_deep_agent or a LangGraph subgraph. Deploy via deepagents deploy to LangGraph Platform.
UI/UX Bonus15%
Roving judges 8–9pm award +0 / +1 / +2. Polished custom frontend with animations and streaming = +2.
Tie-Breaker15%
Equal scores? Earliest submission timestamp wins. Submit your URL before 9:00pm.
DELIVERABLES
What to Submit
4 required deliverables — submit your public URL before 9:00pm
Required
Filled PRD
system-design-prd-template.md — fill this out before writing code
Agent
Built with create_deep_agent + Cursor as your primary editor
Hosted Public URL
Anonymous auth only — anyone opens the URL in a browser, no login required
.cursorrules File
Shows how you used Cursor during the build
Bonus Points
Submit Early
Earliest submission wins ties — don't wait until 8:59pm
Custom Frontend
Polished UI earns up to +2 bonus points from roving judges
Tip: Architecture thinking first. Teams that skip the PRD tend to build the wrong thing.
TOOLS & RULES
Tools & Rules
Required tools and constraints for the build
Required Tool
Cursor
Primary IDE — must use Cursor throughout the entire build
Allowed to use
Deep Agents frameworkAny language or frameworkAny AI modelCustom frontends
Forbidden
Login walls or user-facing API keys
Pre-built agents started before kickoff
Logistics
Meals
Lunch, dinner, snacks. Coffee, tea, water — unlimited
Infrastructure
Wi-Fi, power outlets, extension cords, lounge area
Assistance
Help from organizers