Most of us don’t struggle with having ideas. We struggle with connecting them and executing on them when it matters.

Over the past year, I’ve been building a Second Brain: not a note-taking app, but an agentic system that captures, classifies, links, and executes on thinking at scale.

The result? A system where I just send a message to a Slack bot, and it figures out what to do with it. No apps to open, no categories to choose, no formatting decisions. Just capture and let the automation handle the rest.

What Makes This Different

Unlike traditional note-taking systems that require upfront categorization decisions, this system:

  • Separates capture from classification - AI handles the organization automatically
  • Creates intelligent connections - semantic linking across your entire knowledge base
  • Routes actions automatically - tasks go to OmniFocus, events to calendar, ideas to structured storage
  • Builds over time - becomes smarter as it learns your patterns and preferences

The Technical Foundation

Built on AWS serverless architecture using:

  • Lambda + API Gateway for Slack integration
  • AgentCore + Bedrock for AI classification with behavioral learning
  • CodeCommit for versioned knowledge storage as structured Markdown
  • SQS + DynamoDB for reliable, idempotent processing

What I found using it

This is a personal project, so these are observations from my own use rather than measured benchmarks:

  • Zero-friction capture noticeably increased how often I recorded notes
  • Processing typically completed in a second or two per message, measured informally
  • Removing the upfront categorization step took away the decision fatigue that stopped me capturing
  • Related thoughts were linked across time without manual tagging

Read the Full Story

I’ve written a comprehensive deep-dive on this system, covering the architecture, implementation details, real-world usage examples, and the future of multi-agent thinking systems.

Read the full article on Medium: “Building a Second Brain on AWS with Agentic AI”

The complete implementation is also available as an open source project on GitHub.