重點導讀
Guillermo Rauch 回憶,他最早試用 @eve 的內部測試版時,做出了一個個人購物 AI 助理,但當時還缺少不少基礎設施。
- 現在有了 @agentcardhq,模型也變得更好,iMessage 和 WhatsApp 的 eve 聊天管道更成熟,執行速度大幅提升,同時保留資料耐久性與加密保護。
- 他表示,現在能用開放式 AI 技術堆疊,打造像 Muse/Instinct 這類可自行掌控的平臺,並鼓勵大家從範本開始修改成自己的版本。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
Guillermo Rauch@rauchg
·2 天前
When I first tried an internal pre-release version of @eve, a personal shopper agent was my "hello world". It worked, but a lot of the infrastructure was missing.
Now, @agentcardhq exists, models are better, iMessage and WhatsApp eve channels are mature, and performance is vastly improved (without trading off durability and encryption.)
The result is you can ship a platform of the caliber of Muse / Instinct, that you fully control, on our open intelligence stack. The experience is pure delight. I highly recommend you try this. Run with this template. Make it your own.
推文來源 · X (Twitter)在 X 參與討論
來源: Guillermo Rauch (x.com)
在 X 查看推文