Agent dev tip: kill single chats — agents messaging each other is the real unlock
xeophon · x · 2026-09-12
Dev xeophon builds on the take that OpenAI/Anthropic models should be treated as a 'blob' you throw tasks into for finished deliverables — not code or prose. His addition: design clients the same way. Single chats are dead; letting agents communicate with and about each other, passing messages back and forth, is a huge unlock once you get used to it.
More from coding & agent
- Kitaru Auto-Ingests Traces From Langfuse, Braintrust to Surface Agent Cost and Accuracy Wins — strickvl · 2026-09-12
- Steve Yegge asks what IDE people use to multiplex 10-20+ agents; wezterm gets the nod — peterjliu · 2026-09-12
- LukeW's Intent Adds Agent Evidence Capture and Multi-Device Coordination — LukeW · 2026-09-12
- Agent Coordination App Intent Adds Screenshot Evidence Capture and Multi-Device Sync — LukeW · 2026-09-12
- Cognition's Devin CLI Fusion Cuts Coding Benchmark Costs 39% by Splitting Planner and Executor — DeryaTR_ · 2026-09-12
- Andrew Ng: AI engineering skills are blurring developer, PM and designer roles — AndrewYNg · 2026-09-12