# Notion’s Token Town: 5 Rebuilds, 100+ Tools, MCP vs CLIs and the Software Factory Future — Simon Last & Sarah Sachs of Notion

Canonical URL: https://www.traeai.com/articles/3a723267-37dc-4c11-a1b8-a2a36787c035
Original source: https://www.latent.space/p/notion
Source name: Latent Space
Content type: article
Language: 英文
Score: 8.2
Reading time: 95 分钟
Published: 2026-04-15T00:31:14+00:00
Tags: AI智能体, Notion, 软件工程, 产品架构, 大模型应用

## Summary

Notion团队揭秘自研AI智能体历经5次重构，强调产品需围绕人机协作设计，而非仅封装大模型。

## Key Takeaways

- Notion Custom Agents历经4-5次重建，因早期模型能力与工具调用标准不足
- 成功关键在于构建支持人机协同的系统，而非简单包装大模型
- Notion正探索由智能体组成的“软件工厂”，实现代码全生命周期自动化

## Citation Guidance

When citing this item, prefer the canonical traeai article URL for the AI-readable summary and include the original source URL when discussing the underlying source material.