Inspiration

I'm a 4th-year PhD student in organic chemistry. Every time I need to run a reaction, I have to search through 8 physical notebooks from 4 years of work to plan my next steps. This manual process is tiring and tedious, so I decided to build an AI solution that could learn from my past work and guide my future research.

What it does

An AI-powered digitized lab notebook that integrates Claude AI with Notion and Perplexity MCPs. It reads my reaction data from Notion, recognizes patterns across multiple experiments, and can either analyze reaction outcomes or predict outcomes for reactions I haven't performed yet. Perplexity provides real-time literature analysis so Claude can give accurate, research-backed suggestions for optimization and new experiments.

How I built it

  • Entered sample lab notebook data into structured Notion databases
  • Connected Claude AI to Notion MCP using GitHub instructions to read and analyze the reaction patterns
  • Integrated Perplexity MCP for deep literature research and current scientific analysis
  • Created classification systems to help Claude identify substrate types, reaction success patterns, and optimization opportunities
  • Built query systems that combine my historical data with real-time literature intelligence

Challenges I ran into

I attempted to integrate Google Calendar MCP to automatically schedule Claude's suggested reactions into my weekly lab planning. Despite multiple attempts, I couldn't get this integration working properly and had to abandon this feature for the hackathon timeline.

Accomplishments that I'm proud of

I'm very proud of what I've accomplished today as I am a completely non-technical person. I know very little about AI, product design, and overall use of vibe-coding. I specialize only in organic chemistry, and no one that I know that practices chemistry knows how to do any of this! I learned so much and am very grateful that AI Tinkerers gave me my first hackathon!

What I Learned

  • First-time experience with command prompt and technical development
  • How to integrate multiple MCP tools (Claude, Notion, Perplexity)
  • The power of combining domain-specific knowledge with AI capabilities
  • How AI can transform traditional scientific workflows
  • The importance of structured data for pattern recognition

What's next for AI-Powered Digitized Lab Notebook:

Successfully integrating Google Calendar MCP for automated experiment scheduling - this would provide tremendous value for planning weeks and months of lab work based on AI recommendations and resource availability. Also running this locally instead of on the server to prevent scientific data being publicized.

Built With

  • claude
  • github
  • mcp
  • notion
  • perplexity
Share this project:

Updates