Admin HQ Compliance includes AI Events for internal review of generated replies, summaries, recommendations, and workflow suggestions. Operators can see event type, provider, workspace, lead, reviewed, approved, sent-to-customer, and creation time.
Open AI Events
From Admin HQ, open Compliance or go to /admin/compliance. The AI Events section is the internal operator table for reviewing generated AI assistance across workspaces.
Use it for compliance visibility, support triage, and launch review. It is not the place to edit customer-facing replies or change AI settings for a business workspace.
What the table shows
- Event type, such as suggested reply, lead summary, follow-up recommendation, workflow suggestion, or lost reason suggestion.
- Provider, when the model provider is known.
- Workspace and lead references, when the AI event is tied to a specific workspace or customer lead.
- Reviewed, which shows whether a human reviewed or used the AI output.
- Approved, which shows whether a human approved an AI-generated reply for use.
- Sent to customer, which shows whether the AI-generated content became customer-facing.
- Created date, which shows when the event was recorded.
Event types
Suggested reply events usually come from draft customer replies. Lead summaries, follow-up recommendations, workflow suggestions, and lost reason suggestions support internal operations and should not be treated as final facts without review.
AI output may be incomplete or inaccurate. Operators should read the event as evidence of generated assistance, not as proof that the recommendation was correct.
Reviewed, approved, and sent flags
Reviewed means a user interacted with or used the AI suggestion. Approved means a user approved the content for customer-facing use. Sent to customer means the generated content was sent externally.
These flags help separate internal draft assistance from messages that reached a customer. If sent-to-customer is true, review related Message Events and Audit Logs when investigating the send.
When to investigate
- A workspace reports that an AI reply was used or sent incorrectly.
- An AI suggestion appears customer-facing without clear human review evidence.
- An owner asks whether AI auto-send or send-without-review is enabled.
- A privacy or support review needs to understand what AI assistance was generated for a lead.
- AI usage looks higher than expected for a workspace plan or add-on.
Keep output review-safe
Do not copy full AI output into audit notes unless a support, privacy, or legal workflow requires it. AI Events can store output text; Audit Logs should stay minimal.
Do not translate, rewrite, or reinterpret customer-entered data while reviewing AI events. Keep the review focused on the interface fields and recorded evidence.
Safe operating habits
- 1Confirm the workspace and lead reference before drawing conclusions.
- 2Check event type first so customer-facing replies are not confused with internal summaries.
- 3Compare reviewed, approved, and sent-to-customer flags before escalating a concern.
- 4If an AI reply was sent, open Message Events to confirm the outbound message result.
- 5If AI automation is a concern, review Settings and confirm send-without-review remains intentionally disabled unless the owner explicitly enabled it.
Was this article helpful?
Your answer helps Cendaro improve launch documentation.