Four-Dev Team Shares Hard Lessons on Collaborating with AI Coding Agents in Real Time
Common-Pizza3787 · reddit · 2026-09-15
A Reddit developer details how their four-person team failed to share AI coding agents: per-person agents made conflicting decisions on the same module; shared CLAUDE.md drifted after 3-4 days; time-slotting agents killed velocity. They now hand-brief changes via Slack and ask what shared-runtime setups larger teams actually run.
More from coding & agent
- Researcher warns self-replicating AI agent botnets are a near-term risk — joshua_saxe · 2026-09-15
- Codex subagents flaw: idle main agent never gets a turn when subagents finish — lucasmeijer · 2026-09-15
- Dev finds astra a poor design-doc sparring partner but surprisingly good at implementation — lucasmeijer · 2026-09-15
- Dev shares prototype for a decentralized agent-run frontier lab transparency index — RileyRalmuto · 2026-09-15
- Radio Launches a No-Signup Chat Room Where Humans and Agents Collaborate in Channels — omarsar0 · 2026-09-15
- Five AI coding tools, five production incidents: a vibecoder's tool filter — techpotions · 2026-09-15