It's Time: The Work Changed

​A year ago you could still debate whether AI belonged in serious software. That debate is over. Agents write real code, run real tasks, and break in real ways - and most of us are improvising our way through it alone.

​Four people who aren't improvising are comparing notes, in public, for 90 minutes. These are the sections each of them walks through on screen:

​

​Theo Browne (t3.gg) - Thinking bigger

• ​End-to-end prompting, not one-shot snippets

• ​Holding quality up when agents produce more than you can read

• ​Pushing your tools to the edge of what they can do

• ​A setup flexible enough to move with the field

​Angie Jones - Building for Autonomy

• ​Better models and wider adoption are not enough to make engineering autonomous.

• ​Agents need context, access, and feedback from the systems around them.

• ​Teams can give agents more independence without giving up control.

​Kent C. Dodds - Primitives that get better results from looser specs

• ​What a primitive actually is

• ​How to design ones worth reusing

• ​Getting your agents to actually use them

​John Lindquist (egghead.io) - Designing and iterating with AI

• ​Prompting for image batches, shared themes, locked-in storybooks

• ​Directing with video: dictating and annotating what you want

• ​Iteration loops so tight that compiles, tests, and APIs stop slowing you down

​Then all four together, live, for 30 minutes of Q&A. Bring the questions you've been solving alone at midnight.

​Who it's for: developers already working with AI agents who suspect they're using 10% of what their tools can do.

​The plan:
19:00 - Four sessions: Theo → Angie → Kent → John
20:00 - Live Q&A with all four
20:30 - Done. Back to shipping.

​Free. Recorded for registrants. Hosted by the MEGA team.