Ditto v2 发布：跨工具统一记忆层

Kalista.hl (τ τ) ⟠ @0x_Kalista原推发布于 2026.10.09

在 X 上打开原推

I was looking through @heydittoai's own weekly numbers and the timing lines up with the Ditto v2 launch at Exploit Summit. Before that, the product was still a small memory app. In the week ending October 4, memories stored were 500,411, up 54%, with 51 new sign-ups and 136 weekly active users. About 91 of the people active in the last 30 days were using it in the app and through MCP. V2 put Ditto in front of a new pool: one memory under Claude, Cursor, Codex and Grok, Telegram and Slack in the same place as the docs, and workflows that act with an approval on anything risky. The Bittensor crowd around SN118 was already watching. But now the product gets to test something different. Can Ditto hold weekly actives once the summit posts stop? because those are two different types of growth. → launch-led growth comes from a keynote, a subnet narrative and people trying the new inbox → product-led growth shows up when those people come back because the memory is actually there in the next tool The week is still well above a dead start, which matters. A slowdown from here would not automatically mean v2 missed. It could just mean the first curiosity cycle is normalizing. What I'd watch from here is whether weekly actives settle above the pre-summit base, and whether MCP use stays a large share of that base. If it does, part of the launch wave may have turned into people who stopped re-explaining themselves. That makes the next few months a useful test of whether Exploit Summit gave Ditto temporary distribution, or helped it become the memory layer people still open after the thread fades. #Bittensor $TAO #Ditto