Descriptions:
At the AI Engineer conference, developer and YouTube creator Theo Browne (@t3dotgg) delivered a candid, opinionated talk about how rapidly advancing AI models are forcing software engineers to fundamentally rethink the way they work. Framing model progress through distinct eras — reliable tool calls with Sonnet 3.5, long-horizon autonomous tasks with Opus 4.5, and full multi-agent orchestration with Mythos and Fable — Browne argues that developers who keep prompting models the same way they did a year ago are missing the point entirely.
The talk takes direct aim at deeply ingrained developer habits: language tribalism, terminal attachment, fear of deleting code, and reflexive sunk-cost thinking. Browne’s core argument is that these aren’t rational choices — they’re accumulated identity markers from an era when skilled engineering judgment was the scarcest resource. That era, he contends, is ending faster than most engineers want to admit.
Browne’s central call to action is to “go bigger” — not by becoming a better individual programmer, but by delegating more ambitiously to models now capable of spawning sub-agents, verifying their own work, and completing tasks that previously took engineering teams days. The talk is aimed squarely at experienced developers with years of strong opinions, making the case that letting go of those opinions is the professional skill that matters most right now.
📺 Source: AI Engineer · Published July 08, 2026
🏷️ Format: Keynote Launch
