Grok Bot Adds a Vault and Refines Approval Steps
In its October 23 update, xAI added a settings Vault, email forwarding, clearer Slack attribution, and changes to approval handling in Grok Bot, its persistent-computer agent product.
- Announcement
- Oct 23, 2026
- Reported here
- Oct 26, 2026
A product update, not a new Grok model
Grok Bot is xAI’s agent product: persistent bots work from their own computer across apps and return for human decisions. It is distinct from xAI’s Grok language models. The October 23, 2026 changelog labels its latest release v0.75.5 and groups the update around a settings Vault, email forwarding, and approval cards.
What changed on October 23
The update lets a Bot save logins, payment cards, and contact details in a Vault in Settings and fill them into sign-in or checkout flows. Bots can forward messages from their own inbox while retaining formatting, images, and attachments, and Slack messages now record who requested them. The release also adds pumpkin-shaped avatars. Auto-review asks less often when a user has already requested an action in their own words; after a user denies an approval card, the Bot stops and does not retry that action until given another message. The changelog says messages requested by the user can send after Auto-review allows them, while routine and other sends remain drafts.
Who it is for and access context
Grok Bot is aimed at people and teams that want a persistent agent to work in connected apps rather than return only a chat response. xAI’s August 11 launch described beta access for selected SuperGrok and Cursor plan subscribers on desktop and iOS, with an enterprise waitlist. An August 26 announcement also listed access through SuperGrok and Cursor plans. These are dated plan descriptions; check xAI’s current product page for present eligibility.
Permissions still matter
The Vault can hold sensitive account and payment details, so its presence does not remove the need to review permissions and approvals. This changelog documents product behavior, not an independent security assessment or a guarantee that every action will be correct.
Go to the source.
Official announcements are the basis for this report. They describe the providers’ claims; they aren’t independent performance tests.