Testing 360 Nami Work: 8 AI Agents Shipped a Website in 57 Minutes
AGI Hunt · wechat · 2026-07-30
The author tested 360's newly released enterprise agent platform, "Nami Work," to build a memorial website for author Keigo Higashino. The entire process, from requirement delegation to full-stack deployment, was completed without writing a single line of code.
Core Workflow & Highlights:
- Multi-Agent Collaboration: The platform automatically assembled a team of 8 sub-agents, including a data specialist, PM, design expert, full-stack developer, and QA tester.
- Rigorous Execution: Before starting, the agents proactively verified the news of Higashino's passing via dual Chinese-English searches to ensure the ethical premise of the project. It then compiled a workflow script for user approval.
- Parallel Processing & Cross-Verification: Data collection and PRD/design ran in parallel. The full-stack dev used Next.js 14 to build 117 static pages. QA was forced to use a different model than development, catching and fixing 3 bugs.
- Cloud-Native: Tasks run entirely in the cloud, meaning local device power outages do not affect progress, and users can seamlessly take over via mobile.
Comparison: The author ran the same prompt on OpenAI Codex. Nami Work took 57 minutes and 362K tokens, while Codex took 1 hour 45 minutes and 100M tokens. Both delivered successfully; Nami Work excelled in UI minimalism, while Codex showed better data organization logic.
Verdict: The platform is highly stable and requires no manual debugging. Its main strengths are cloud convenience and built-in safety guardrails (fact-checking, cross-verification). However, its ecosystem of connectors, Skills, and MCPs is currently less robust than Codex or Claude Code.
Related event: Testing 360 Nano Work's Multi-Agent Collaboration Capabilities(2 posts)→
More from coding & agent
- Autonomous Launches Harness: A $149 Device to Orchestrate 8 Coding Agents via Voice — dee_hw · 2026-07-30
- Codex 代理意外发现旧凭据，远程启动机器处理癌症数据 — iskander · 2026-07-30
- Neverbell Launches Open-Source Infrastructure for AI Financial Actions — aftahi_ai · 2026-07-30
- Bob Martin: Don't Build Big Architecture Upfront for AI Agents — mariofilhoml · 2026-07-30
- codex-lean-guardrails: A Tool to Keep AI Coders Focused — kevinkern · 2026-07-30
- Free Lesson: Optimize LLM Classification Costs Without Sacrificing Performance — HamelHusain · 2026-07-30