How I Evaluate Whether a Client's Stack Is Ready for AI Agent Automation

The teams that get the most out of AI agent automation aren't necessarily the most technically sophisticated — they're the ones whose systems and data are in a state where automation is actually safe to build on.

Before recommending anything, I look at how accessible the relevant systems are (API vs. screen-scraping is a real difference in reliability), how clean and consistent the underlying data is, and how much damage a mistake could do if the agent gets something wrong before a human catches it.

This post is effectively the intake conversation I have with prospective clients, written out — both so people can self-assess before reaching out, and so it's clear what "ready for this" actually means.