
NEW!Latest episode 2026-08-17

How much autonomy is too much?
Agent intrusions and new security holes, what “review this PR” really means, the shifting Overton window on autonomy, not knowing what you’re buying, free tokens as money, and the revenge of the terminal.
Episode archive 9 episodes on file
| # | Episode | Date | Length |
|---|---|---|---|
| 009 | How much autonomy is too much?Agent intrusions, free tokens & the terminal | 01:24:44 | |
| 008 | Fable, RL, loops, and the cost of computeWith Mario Zechner · Vienna | 01:20:53 | |
| 007 | The one-year anniversaryTokens, slop forks & open weights | 01:34:31 | |
| 006 | The beginning of the end of subsidiesSecurity, token spend & data moats | 01:38:22 | |
| 005 | Time to slow down?Quality, security & token substance abuse | 01:38:48 | |
| 004 | Are you going exponential?New models, vibe projects & slop forks | 40:33 | |
| 003 | Have we become AI-pilled?Personal agents, OpenClaw & gates | 59:01 | |
| 002 | The new models are good, actuallyPricing, context handoffs & meta-programming | 51:44 | |
| 001 | The December editionWhere the monthly conversation began | 49:05 |
About the show AI talk without the psychosis
State of Agentic Coding is a recurring conversation about how coding agents are changing the practice – and economics – of making software.
Sponsor webring
This broadcast is brought to you by the soac sponsor webring:
