The AI Adoption Journey Every Developer Goes Through

There’s a predictable path developers go through when adopting AI coding assistants — from “this is useless for programming” to “I can build agents that solve certain problem types.” This video maps that path on two axes: trust in AI output, and trust in AI autonomy.

What’s covered:

The two-axis model: trust in AI output vs. trust in AI autonomy
Common developer stances, from outright refusal to full delegation
The “good enough code” turning point — and the practices that get you there (spec-first, tests, mutation testing)
Why “I can build agents” is a different kind of trust than “I trust the output”
The overconfidence trap: agents burning tokens on the wrong thing
Two more axes that pull on this model: budget (chat → coding agents → custom agentic systems) and privacy (when local models make sense)

Wherever you are on this graph right now, it’s probably not a fixed point — it’s a phase.