Grok Bot deep dive: role-isolated memory, X search, main-Bot orchestration
A SSPAI Matrix author examines Grok Bot through two recent updates: the piece says Grok Bot can search X activity in real time without consuming quota, and relays Elon Musk's statement that Grok Bot will automatically route agent tasks to the best model, naming Opus 5.5, also without extra quota. The author argues its main difference from agent harnesses such as Codex, Claude Code and WorkBuddy is per-role isolation of instructions and long-term memory, a main Bot that dispatches work to specialist Bots (including group chats), and its own cloud computer that keeps logins and files. The article also says product lead Roman framed Grok Bot as a deliberately minimal tool for knowledge workers; most of this is the author's hands-on experience or relayed official and Musk claims.
The article's central claim is that agents can move from handling a single task to owning a long-running role, accumulating role-specific memory and skills — a framing the author derives from his own use.
前段时间，随着 Grok Bot 两项大更新，Grok Bot 的热度又在 X 上重新火了起来。
什么更新呢？第一是 Grok Bot 可以在不消耗额度的情况下，直接检索 X 的实时动态。这意味着 Grok Bot 背后，将直接连接 X 的信息源，对追踪 AI 前沿、做内容或找选题的人来说，是杀手锏级别的存在。
第二项就更加重磅了，马斯克宣布 Grok Bot 将自动路由完成 Agent 任务的最佳模型，还特意点名了最近火出圈的 Opus 5.5，同样不会消耗额外额度。
#Grok Bot#xAI#X#Elon Musk#Claude Opus 5.5#WorkBuddy#Claude Code#Codex#Notion