How do you audit what an AI support agent stored about a customer?
To audit what an AI support agent stored about a customer, look up that customer in the agent's memory or knowledge store, review the conversation transcripts and action logs linked to them, check which source systems the agent read from or wrote to, confirm retention settings and who can access the data, and verify that deletion requests remove it everywhere.
What to review
Vendor questions
Can we export all data for one customer? Is memory permission-aware? Is customer data used to train models? Are deletions propagated to memory and logs?
