Dev Interrupted The agent wasteland, federated workflows, and a computer for computers
47 snips
Mar 6, 2026 Listeners hear a deep dive into viral OpenClaw adoption and the security risks that follow. They explore Steve Yegge’s idea of federated “wastelands” for orchestrators, reputation ledgers, and wanted boards. The Perplexity Computer as a persistent digital coworker and the implications of AI making basic development extremely cheap are also discussed.
AI Snips
Chapters
Transcript
Episode notes
Hosts Swap Agents And Hit Integration Chaos
- Ben and Andrew describe swapping agent outputs and using sub-agents to hand off tests and combine front-end/back-end code.
- They recount a two-person Code TV Goose challenge where combining outputs became a hilarious saga.
OpenClaw Shows Agents Scale By Solving Real Work
- OpenClaw's viral GitHub growth shows agent tools win by fitting everyday workflows, not just by improving foundation models.
- 250,000+ stars in months and broad adoption outside engineers reveal hunger for multipurpose agentic assistance and attendant security risks.
Prioritize Security For Viral Agent Projects
- Patch and monitor widely adopted agent projects aggressively because mass adoption exposes security and abuse vectors quickly.
- Andrew warns that as OpenClaw spreads to non-technical users, prompt upgrades and fixes become a safety imperative.