Blog

gws is a new CLI that turns every Google Workspace API into shell commands with 89 agent skills for Claude Code. I set it up, tested it against real GTM work, picked 27 skills worth having, and learned why you should not load them all at once.

The SDR role is evolving into something more technical. The ones learning now will become the go-to-market engineers companies need.

The role sits between SDR and infrastructure engineer. Here's what it looks like in practice - the tools, the evaluations, and why independence matters.

I run 4-6 Claude Code terminals at once. The single handoff file worked until it didn't. Here's the parallel-safe architecture I replaced it with.

I tried separate AI platforms, burned through API credits, and landed on the simplest architecture possible. Here's the full setup — CLAUDE.md, soul files, SQLite persistence, a DNA evolution system, 50+ skills, and 9 MCP servers. Updated weekly.

Why importing someone else's skill pack into your system is the fastest way to destroy what you've built. And why everything lives on three open websites instead.

Three sites, one parent keyword, and a deliberate topology that tells search engines and AI platforms exactly how everything connects.

No GPU, no After Effects. Just React components, Remotion, and the same monorepo that runs three websites.

100+ content files across 6 platforms became opaque. So I built a SQLite index that makes the entire repo queryable from the command line.

I was spending hundreds per month on API calls to chat with my AI agent. Then I realized I was already paying for unlimited access. Here's the CLI-to-chat bridge pattern that changed everything.

The first post from ShawnOS.ai. Where GTM engineering meets building in public.