Skip to content

Blog

Field notes

Notes on building products end to end, the AI workflows that speed the work up, and the tooling in between. Written from production, not from a tutorial.

Latest

How Manager and Worker AI Agents Work Together

A deep dive into the manager/worker pattern in multi-agent AI systems: how a manager agent decomposes a goal, delegates subtasks to worker agents, coordinates sequential vs parallel execution, handles failures, and integrates results.

12 min read
Read the note

All notes

4 notes

  • 12 min read

    How Manager and Worker AI Agents Work Together

    A deep dive into the manager/worker pattern in multi-agent AI systems: how a manager agent decomposes a goal, delegates subtasks to worker agents, coordinates sequential vs parallel execution, handles failures, and integrates results.

  • 4 min read

    GridFS from Bun, without the ceremony

    Why I wrote bun-gridfs-storage, what the old Multer engine got wrong under Bun, and the four decisions that made file uploads boring again.

  • 3 min read

    Test data that tells the truth

    Fixtures lie by omission. mongoose-test-factory reads your schema and generates data that respects it. How it decides what to generate, and the three modes I reach for every day.

  • 4 min read

    What an agent actually needs from a tool

    After a year of building MCP servers for real workflows, the tool surface matters more than the model behind it. Seven rules I now apply before a tool ships.

The next note, when it is ready.

New notes land here first. Follow the feed if you would rather not check back.

Subscribe with RSS