No Assets, Just Agents: Questioning the Value of Critics Loops
Silver-Chipmunk7744 · reddit · 2026-08-30
The author showcased a 3D scene built purely in Three.js (0 external assets), generated primarily by Claude 5.0 Opus using 18 agents working on specific tasks, with only one minor correction prompt.
- Workflow Critique: The author questions the industry-standard "Critics Loops", suggesting they waste time and money. The demo shows that using many specialized agents (18) for specific tasks yields better results than fewer agents relying on a QA agent for feedback.
- Cost vs. Quality: A QA agent cannot match a human tester. Furthermore, running 18 agents for 24 hours would burn through token budgets rapidly. The author argues it's more optimal to have humans perform the critic loop.
More from coding & agent
- Vercel Goes Full Backend: Python GA, Containers, and Daemons — cramforce · 2026-10-16
- WorldMemArena Accepted at NeurIPS: Memory Storage ≠ Memory Use — xwang_lk · 2026-10-16
- Vercel Ship Demos Container Deploys for Go, Ruby, Java and More — edwin · 2026-10-16
- TRL adds day-0 support for fine-tuning Muse Spark Open's 816B MoE with LoRA on 64 H100s — QGallouedec · 2026-10-16
- Databricks engineers land impactful PRs on day 2 with AI dev tools, vs week 3 before — Yuchenj_UW · 2026-10-16
- Perplexity open-sources SPACE, the sandbox platform powering its Computer agents — AravSrinivas · 2026-10-16