Use AI assistance responsibly

Protodesk AI can help summarize context, find relevant knowledge, and prepare suggested replies. Treat its output as a draft that requires judgment, not as an independent source of truth.

Start with trusted knowledge

AI assistance is most useful when your published Knowledge is accurate, current, and specific. Test important questions in Knowledge before relying on suggestions in live conversations.

Review every customer-facing suggestion

Before sending AI-assisted text:

1. Confirm that it answers the customer's actual question.
2. Verify names, amounts, dates, URLs, and policy details.
3. Check the supporting source when the answer depends on Knowledge.
4. Remove assumptions the customer did not provide.
5. Match the tone to the conversation.
6. Make sure internal notes or private context are not included.

The teammate who sends the reply remains responsible for it.

Use the right action

• Use a suggested reply when the next customer-facing response is clear.
• Use rewriting tools to improve an answer you already understand.
• Add an internal note when the result is context for teammates rather than the customer.
• Escalate to a person when the evidence is missing, conflicting, or high risk.

Protect sensitive information

Do not place passwords, authentication tokens, payment details, or unnecessary personal information into AI instructions. Follow your organization's access and retention requirements.

If a suggestion looks confident but lacks reliable evidence, do not send it. Improve the Knowledge source or answer manually, then use the gap to make future support safer.