AI Staff security and privacy
Each AI Staff member has a profile with tool switches and an autonomy level, and staff and their conversations stay inside your organization. Some of these settings are saved but not yet applied everywhere, as explained below.
Tool switches
A staff member’s profile lists these tools: Web Search, Knowledge Base, Memory, Calculator, Code Execution, Browser / Computer Use, Email, SMS, Phone, External APIs and Spend.
- The switches are saved on the profile, but ordinary staff chats don’t apply them yet. For example, a staff chat can still search the web when web search is allowed for you in chat, even if the staff member’s Web Search switch is off.
- Several tools depend on products that are not open yet, such as the knowledge base, the Call Center and agent deployment. Staff can’t use those today, whatever the switch says.
Autonomy
Each staff member has one of three autonomy levels, saved on the profile:
| Level | Meaning |
|---|---|
| Suggest | Proposes actions; you run them |
| Approve | Acts, but asks before anything risky. The default |
| Autonomous | Acts on its own within its guardrails |
The level matters for outside-world actions such as email, SMS, phone, code, browser, external API and spending. Those actions are not open for staff yet, so the level has no practical effect today.
Memory and data
- Organization admins can turn memory off with Disable AI Memory in Settings → Admin Settings.
- Details you put in Settings → Account, such as your job title, can be used by your AI Staff.
- Dismissed staff can be restored for 30 days. See Data retention.
Model training
The Help improve Syntic switch in Settings → Data Controls applies to staff conversations too. See Model training and your chats.