4-AI relay writing experiment: ChatGPT drafts, Muse, Grok and Claude review in the open
AggressiveGift1532 · reddit · 2026-09-17
A Redditor ran a writing experiment assigning four models distinct roles: ChatGPT drafted, Muse reviewed and requested one revision, Grok did a fresh-eyes pass, and Claude did final review. The author manually handled all 8 handoffs in about an hour, publishing the brief, drafts, reviews and a timestamped handoff log on Notion. Key method: separate roles per model, one shared brief to prevent drift, and a visible record of every change — with the honest caveat that four AIs don't automatically yield a better answer.
More from coding & agent
- Baseten Adds Server-Side Web Search for Open-Weight Models via One-Line Tool Config — PolarBearby · 2026-09-17
- A 30-day playbook: do everything through Claude Code to actually master AI — jacob_posel · 2026-09-17
- Jev Model Router Cuts Latency 95% vs GPT-5.6, Runs Inline in Agent Sessions — pwendell · 2026-09-17
- Tsinghua's CERA-MoA co-evolves routing and LLM agents via RL — Tsinghua · 2026-09-17
- AI functions thrive in databases because nobody wants the MLOps burden — sh_reya · 2026-09-17
- Inside a $1M MRR solo maker's paid AI stack for September — tibo_maker · 2026-09-17