Avi Kothari教你用Notion管上下文，模型随便换，不被一家锁死。比所有东西塞进一个模型灵活多了。
Avi Kothari建议将知识、记忆和工作流存储在Notion等工具中，不同模型可以竞争来访问该上下文，从而避免被单一模型锁定。他指出，一旦上下文锁定在某个模型内，就等于交出了自己的护城河。Jens Honack则补充“主权栈”与“模型可替代性”存在张力：团队希望跨模型可选性，但深度集成会积累切换成本。这种张力可能催生下一代基础设施。推文获得的4次转发和23个赞反映了业界对这一理念的关注。
The answer is to separate your context from your intelligence. Keep your knowledge, memory, and work...
The answer is to separate your context from your intelligence. Keep your knowledge, memory, and workflows in @NotionHQ (or similar), and let models compete to serve that context. The moment your context lives inside one model, you’ve handed over your moat. Jens Honack @JensHonack the hard part is that "sovereign stack" and "model fungibility" pull in opposite directions. teams want optionality across models, but the deeper agents get into memory, tools, and workflows, the more each model stack starts to accumulate its own switching costs. that tension is probably where a lot of the next infra gets built. 🔗 View Quoted Tweet 💬 4 🔄 4 ❤️ 23 👀 3752 📊 8 ⚡
- Notion07-06 21:45原文