Running more than 2 sub-agents at once burns tokens for zero quality gain, dev argues
kevinnbass · x · 2026-09-16
A developer argues that running more than 2 sub-agents in parallel is almost certainly burning tokens with no quality gain, because agents don't trust each other enough and end up double-checking each other's work. He pushes back on the multi-agent scaling hype, calling it a 2025-era take.
Related event: Developer: running more than 2 parallel sub-agents just burns tokens(3 posts)→
More from coding & agent
- Pawel Huryn ships open-source AI review kit: document intent, then audit code against it — PawelHuryn · 2026-09-16
- OSINT MCP Server integrates Shodan, VirusTotal and HaveIBeenPwned for AI agents — modelcontextprotocol · 2026-09-16
- funn.to launches free AI-powered MCP connector for Amazon affiliate funnels — modelcontextprotocol · 2026-09-16
- MIT CSAIL spinout G5 Labs emerges with $14M seed for AI-native software development — tim_kraska · 2026-09-16
- 7 hard-won truths about working with coding agents — PawelHuryn · 2026-09-16
- After a year of agentic coding, dev concedes traditional artists were right: no soul to be found — ryunuck · 2026-09-16