Plan requirement
| Subscription | Any plan |
| Access | Admin |
What the AI features see, where that is decided, and the three questions worth answering before you switch them on.
What they work from
Your own content and your own tickets. An AI agent answers from your published articles; copilot works from the ticket in front of the agent; triage reads incoming messages to label them.
None of it invents an information source. What it can see is what you have put in Zendesk.
Where the answers live
Data processing, retention and location are contractual matters, set out in your agreement and the account's own settings rather than in a feature screen. Some accounts also hold a data centre location arrangement that fixes where data is stored.
If you need the exact position for your account, ask pluscloud support and we will confirm it against your contract rather than from a general description.
Three questions to settle first
- What is in your tickets? Health details, identity documents and payment data are different from order numbers, and they change the conversation.
- What may an AI agent tell a customer? It answers from published articles, so anything an article contains is something it can say.
- Who needs to agree? In many organisations a data protection officer or a works council has a legitimate view, and involving them early is faster than involving them late.
Restriction is a design tool
Keep articles an AI agent should not use out of its reach, scope triage to the brands you mean, and give copilot to the teams whose work suits it. Narrow beats blanket, and it is easier to widen later.
Tell customers they are talking to AI
Not only because rules in some markets require it. A customer who realises halfway through trusts the whole exchange less, including the parts that were correct.
Comments
0 comments
Article is closed for comments.