Every engineering team is bottlenecked on the same thing: too few engineers, too many requests. AI assistance was supposed to fix this. Instead, most teams use AI to make their existing engineers slightly faster, while non-engineers still file tickets and wait.
What if the actual unlock isn't AI for engineers — it's AI for everyone else? What happens when your designers ship their own components, your editorial team writes their own admin tools, your marketing lead wires up her own automations?
Over the past year, I've trained two non-engineers at a 130-year-old media company to ship production code with Claude. Not toy projects — real features running on real infrastructure used by real customers. The work is legitimately good, because the governance scaffolding around them does the heavy lifting that engineering experience used to provide.
In this talk, I'll show the exact training arc, the guardrails that made it safe, and why distributing AI building skills across non-technical teammates is the real productivity revolution most teams are sleeping on.