AI Briefing
KO

PMs Live in the Terminal Too

·2026.04.17 04:05

Key point

A real-world account of how Claude Code and MCP shifted half a PM's workday into the terminal and VS Code.

Details

Even as a Product Manager, I spend more than half my day in the terminal and VS Code.

After using Claude Code for a few months, the way I work has changed dramatically, going beyond a simple data analysis tool to reshape my entire working environment.

The key wasn't the AI itself but the workflow built around it.

  • Connected databases, Notion, Slack, Gmail, Granola, and Metabase as MCP servers
  • Built a custom memory system that stores what went well, what went wrong, and what decisions were made, per session
  • Used /session-learnings to accumulate conversation content in the project folder so the next session can pick up where it left off
  • Added hooks that get applied when feedback or error corrections come in, so corrections keep accumulating
  • Tracked every change with Git

Thanks to this environment, I've been able to directly carry out simulations, analysis pipelines, and large-scale operational data analysis that I couldn't have done before.

I never formally learned to code, but because I understand the problem and know what needs to be done, Claude Code has become the bridge between understanding and execution.

I'm also running a skill library for things like data extraction, dashboard creation, and writing analysis documents for leadership reports, and I'm expanding some tasks toward running multiple agents in parallel.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.