Handbook - AI Tinkerers - New York City Hackathon
AI Tinkerers - New York City
Sign in

AI Tinkerers - New York City Hackathon

Handbook for the AI Tinkerers - New York City Hackathon hackathon.

Sign in
DevRel Lead - Restate
Build durable AI agent infrastructure. Drive US remote adoption today.
Founding Applied AI Lead - Paxos Health
Build healthcare AI agents in NYC. Apply now.
All Jobs
Final round winners have been announced. View Results

Handbook

Handbook

🚀 Quick‑Start Resources

  • Redis (State & Pub/Sub) - https://redis.io/docs/latest/
    • Node client: npm i redis
    • Python client: pip install redis
  • Tavily (Live Web Retrieval) - Docs: https://docs.tavily.com/welcome
    • Get an API key from the Tavily dashboard.
    • Example request:
    curl -X POST https://api.tavily.com/search \  
  -H "Authorization: Bearer YOUR_TAVILY_KEY" \  
  -H "Content-Type: application/json" \  
  -d '{"query":"latest AI research 2025","max_results":5}'
  • CopilotKit (Agent Orchestration) - Docs: https://docs.copilotkit.ai/

đź“š Sample Project Ideas

Track Idea Core Tools
Web‑navigation & automation Auto‑fill job‑application forms across LinkedIn & company sites Tavily (search), Redis (session), CopilotKit (decision flow)
Tool‑chaining agents “Research‑to‑Report” bot: fetch latest papers, summarize, push to Google Docs Tavily, CopilotKit, Redis (cache), Google Docs API
Context‑aware agents Stock‑watcher: monitor live price tables, trigger buy/sell alerts on a brokerage API Tavily (live feeds), Redis (price history), CopilotKit
Personal assistants Daily agenda builder that scrapes calendar invites, weather, and news Tavily, Redis (user prefs), CopilotKit
Augmented web‑apps CRM enhancer that auto‑logs meeting notes from Zoom transcripts into Salesforce Tavily, CopilotKit, Redis (state), Salesforce API

Feel free to remix or combine ideas - the best projects often blend tracks!

🛠️ Development Tips

  1. Persist State Early - Store each step’s output in Redis (e.g., agent:session:{user_id}) so the agent can recover after a crash or rate‑limit pause.
  2. Rate‑Limit & Retry - Wrap Tavily and any external API calls in exponential‑backoff; Redis can hold a retry counter.
  3. Validate Web Content - Use a simple HTML parser (Cheerio / BeautifulSoup) to guard against malformed pages before feeding results to the LLM.
  4. Secure Secrets - Never hard‑code API keys; load them from environment variables or a Redis‑backed secret store.
  5. Prompt Engineering - Keep prompts short, include explicit “Tool‑use” instructions (e.g., “If you need up‑to‑date stats, call the Tavily tool.”).
  6. Testing Locally - Mock Tavily responses with a static JSON file; run the agent in a Docker container to mirror the hackathon environment.

📦 Starter Boilerplate

# clone the starter repo  
git clone https://github.com/CopilotKit/coagents-starter-langgraph  
cd agent-starter  
  
# install dependencies  
npm install   # or pip install -r requirements.txt  
  
# set env vars  
export REDIS_URL=redis://localhost:6379  
export TAVILY_API_KEY=your_key  
export COPILOTKIT_API_KEY=your_key  
  
# run the dev server  
npm run dev   # or python main.py

đź“ž Help & Community

Good luck, and have fun building the next generation of autonomous web agents!