Useful Agents Hackathon w/ NEAR AI, SF Compute & Electric Capital 💻
Welcome to the Useful Agent Hackathon
A 24-hour sprint dedicated to building agents that reliably execute tasks and meet concrete performance metrics! Win from a $20k USD prize pool while solving real problems for startup founders while receiving free inference and post-hackathon go-to-market support. Agents DO NOT need to interact with crypto or a blockchain.
Co-organized by NEAR AI, SF Compute, and Electric Capital 💻
Agenda
Friday, March 14, 2025
9:30 AM: Doors Open, Breakfast & Coffee
11:00 AM - 12:00 PM: Hackathon Kickoff and Panel - "What is a Useful Agent?" w/ Avichal Garg (Electric Capital), Illia Polosukin (NEAR), Adam Wolff (Anthropic), and Evan Conrad (SF Compute)
12:00 PM: Hacking Begins
1:00 PM: Lunch
2:00 PM - 5:00 PM: Workshops
2:00 - 2:30 PM: NEAR AI Agent Hosting (Jay Zalowitz, NEAR AI)
2:30 - 3:00 PM: Data Sourcing (Jordan Gray, Public AI)
3:00 - 3:30 PM: Edge AI Data Agent (Shu Dong & Su Dong, Mizu)
3:30 - 4:00 PM: Private Vector Database (Felix Meng, xTrace)
4:00 - 4:30 PM: NEAR AI Typescript SDK (Zahid Islam)
4:30 - 5:00 PM: Building Shade Agents (Miao Jiang/MJ, Proximity Labs)
5:00 PM - 6:00 PM: Boba & Engagement Activity
6:00 PM - 8:00 PM: Dinner
9:00 PM - 11:00 PM: Taiyaki Soft Serve Ice Cream Bar
8:00 PM - 12:00 AM: Evening Hacking Session with Snacks
Saturday, March 15, 2025
12:00 AM - 1:00 AM: Late Night Food & Secret Fun Activity
1:00 AM - 7:00 AM: Overnight Hacking Session with Snacks
7:30 AM - 9:00 AM: Breakfast & Espresso Bar
9:00 AM - 12:00 PM: Morning Hacking Session with Snacks
12:00 PM - 1:30 PM: Project Demos & Judging
12:00 PM - Project Submission
12:00 PM - 1:00 PM: Lunch
1:30 PM - 1:50 PM: Judging Deliberation
1:50 PM - 2:00 PM: Closing Remarks & Winners Announced
2:00 PM - 3:30 PM: Networking
Eligibility Requirements
Projects must be newly created or have new features added during the hackathon period
All code must be open source and publicly accessible
Teams can consist of 1-4 members
Each participant can only be part of one team
All team members must be registered for the hackathon
Technical Requirements
Agents must be hosted on the NEAR Agent Hub (https://app.near.ai/agents)
Projects must integrate at least one AI component (LLMs, ML models, autonomous agents)
Code must be hosted in a public GitHub repository
Projects must include quantitative benchmarks for agent performance
Required Repository Contents
Project name and one-line description
Problem statement and solution approach
GitHub repository URL
Detailed README.md with project description
Setup and installation instructions
Demo video or link to live demo
License file (MIT, Apache 2.0, or similar)
Benchmark suite and evaluation metrics
Technical architecture overview
Future development roadmap
Benchmarking Requirements
Defined success metrics for each task the agent performs
Quantitative performance measurements
Reproducible testing environment
Comparison against baseline approaches
Error rate analysis
Task completion time metrics
Resource utilization measurements
Demo Materials
3-5 minute demo video
Link to live demo (if applicable)
Screenshots/diagrams of key features
Test accounts or sample data (if needed)
Recorded benchmark runs
Team Information
Team member names and roles
Contact information
Project background/inspiration
Track selection and bounty targets
Judging Criteria
Technical Innovation (20%)
Novel problem-solving approaches
Implementation quality
System architecture
Task Completion & Benchmarks (30%)
Reliability of execution
Quality of benchmark methodology
Performance metrics
Reproducibility of results
Comparison to baselines
User Experience (30%)
Interface design
Ease of use
Documentation quality
Presentation (20%)
Demo quality
Documentation clarity
Team presentation
Disqualification Criteria
Submissions after the deadline
Incomplete documentation
Code plagiarism
Violation of open source licenses
Missing or incomplete demo materials
Lack of quantitative benchmarks
Non-reproducible results
Special thanks to Digital Garage - our incredible venue partner!
Bounty List
➡️ General Prize by Electric Capital and NEAR AI
💰Prizes: 1st Place - $5,000; 2nd Place - $3,000; 3rd Place - $2,000
🔗 Resources: Near Ai Docs
➡️ Coinbase CDP - Multi-Agent Ecosystem
💰 Prize: 2,000 for the Most Innovative Use of CDP AgentKit 🔗 Resources: Hackathon Guide
➡️ Nevermined - PayPal for AI
💡 Description: Build a multi-agent swarm where agents use Nevermined to pay and reward each other.
💰 Prize: Up to $1,000 per swarm deployment (based on complexity).
🔗 Resources: Nevermined Docs
➡️ MIZU - Decentralized Edge AI Network
💡 Description: Use MIZU’s workflow for hosting models and explore potential applications.
💰 Prize: $1,000
🔗 Resources:
➡️ Phala Network – Confidential Computing Cloud
💡 Description: Three winning projects will receive $500 + $2,400 in Phala Cloud credits (6 months) Additional: $400 in Phala Cloud credits for all projects that deploy to Phala Cloud
🔗 Resources: Phala Cloud | Documentation | Telegram Support | Discord Support
➡️ XTrace - Privacy-First Ai Infrastructure
💡 Description: Build privacy-preserving AI Personal Staff (e.g., tax accountants, legal assistants) using XTrace RAG. Explore:
Supporting diffusion models and GUI agent models.
Integrating Hugging Face as a model provider.
Building tools for easy node deployment via Python/JS. 💰 Prize: $1,000 🔗 Resources: XTrace Docs
➡️ Questflow - Multi-Agent Ecosystem
💡 Description: Integrate Questflow's multi-agent system to solve user problems or create new AI agents.
💰 Prize: $3,000 split across 5 prizes
➡️ Pond - Crypto-Native Ai Model Powerhouse
💡 Description: Build AI-powered solutions using Pond AI models.
💰 Prize: $1,000 USDT
🔗 Resources:
➡️ Silverstream AI - Browser-Based AI Agent Infrastructure
💡 Description: Develop the best-performing web agent using a full-stack open-source solution.
💰 Prize: $500
➡️ PublicAI - On-Demand AI Training Data
💡 Description: Break the AI hype with humor! Submit the most silly AI project. No limits—embrace vibe coding and chaos crafting.
💰 Prize: $5,000 in credits for running AI data collection campaigns (image, text, video, audio).
➡️ Exabits - Crypto's AI Infrastructure
💡 Description: Build AI-native applications, crypto AI products, or AI agents using Exabits infrastructure.
💰 Prize: Yes (Amount TBD)
🔗 Resources: Exabits API