How to Keep Shipping When You Walk Away from Your Desk — Zack Proser, WorkOS
Summary
AI agents can now scale infinitely, but human attention is the real bottleneck—developers are burning out by 11 AM despite unprecedented productivity. The solution: give agents verification criteria and tool access to self-validate work, then implement signal layers and voice-first flows to keep yourself from collapsing.
Key Takeaways
- Agents with verification criteria and tool access can self-validate work autonomously. Give Claude access to Slack and Linear, tell it to fix bugs AND verify the fix is complete—it will loop until the task is truly done without requiring your attention.
- The agent isn't the constraint anymore—human attention and judgment are. Simon Willison fires up 4 parallel agents and is wiped out by 11 AM. You must find your personal cognitive load limit before stacking too many concurrent tasks.
- Don't just scale output linearly with new tools. The temptation to fix 150 bugs daily with agents will destroy you. Instead, architect workflows where agents handle infinite looping while you provide taste, judgment, and business context.
- Implement 'signal layers' (framework mentioned but underdeveloped in transcript) and voice-first coding flows to bring human developers into balance with hyper-charged agent capabilities. This prevents cognitive overload from context switching.
- Give agents MCPs (Model Context Protocol connections) to external systems like Slack and ticketing tools. This creates closed-loop task completion without requiring you to context-switch between windows or manually verify work.
Related topics
Transcript Excerpt
Hey everyone. Uh I'm Zach. I work at WorkOS. Thanks for coming. Uh WorkOS is um provides drop-in APIs that allow you to take your software and go upmarket and sell larger deals to enterprises. Uh but what I'm going to talk about now is um sort of the way that I'm finding to try and maintain balance with all the insane new tools that we're getting every day. So, show of hands if anyone is uh AI coding with agents lately and feels a little bit like this and yeah, despite, you know, getting more done than ever before, like you're completely fried at the end of the day, right? And like adrenaline dumping constantly. So, this has been my experience and I've noticed that some of the worst of it is like the context switching was always super expensive for me and now it's it's worse than ever befo…
More from ai.engineer
- Why Eval++ Is the Next Great Compute Primitive — Sunil Pai & Matt Carey, Cloudflare
- Why More Context Makes Your Agent Dumber and What to Do About It — Nupur Sharma, Qodo
- RAG is dead, right?? — Kuba Rogut, Turbopuffer
- Text Diffusion — Brendon Dillon, Google DeepMind
- The maturity phases of running evals — Phil Hetzel, Braintrust