NexoraDocs Start ₹1 trialLog in
Danny AI (AI agent) › AI best practices and safety

AI best practices and safety

#Get great answers

  • Feed it facts. Prices, timings, locations, policies. The AI answers only from what you give it.
  • Give it a goal. "Book a site visit" or "collect the order details" gives every conversation a direction.
  • Teach it from gaps. Check Identified Gaps every week and answer the questions it couldn't.
  • Test after every change in the Playground.

#Keep customers' trust

  • Be honest that it's an assistant. If a customer asks whether they're talking to a bot, the AI should say it's the business's virtual assistant. Write this in the Style section.
  • Never let it promise what you can't keep: discounts, delivery dates, guarantees. Write the limits down.
  • Make it easy to reach a person. Customers who ask for a human should get one.

#Stay in control

Control Where
Check replies before they go out Operating mode Draft only
Let the team jump in Human takeover window
Hand over uncertain answers Confidence Threshold
Stop the AI at once Operating mode Paused
See every decision Audit Logs

#Sensitive topics

Don't use the AI to give medical, legal or financial advice, or to collect card numbers, passwords or OTPs. Tell it in the Style section to hand these over to your team.

#Your customers' data

Customer messages are used to reply in that conversation. See Data and privacy.

Was this page helpful? Still stuck? Contact support or ask Nexora Help inside the app.