Back to the learning library

Set a clear boundary between AI and your team

A practical framework based on information, permissions, judgment, and customer preference.

2 min read4 practical steps
WHAT YOU’LL TAKE AWAY

Escalate when the answer is uncertain, the action needs authority, or the customer asks for a person.

WRITTEN FORSupport managers and CX leads
THE IDEA TO TAKE WITH YOU

Escalate when the answer is uncertain, the action needs authority, or the customer asks for a person.

01

Check the information

Can the question be answered from a reliable source available to the assistant? If the source is incomplete, decide whether one clarification can help or whether the team must investigate. Do not fill the gap with a plausible guess.

02

Check the authority

A policy explanation and a policy exception are different tasks. Identify actions that need approval or verification, and set their permissions explicitly. Include refunds, cancellations, and changes to personal details in the review.

03

Check the customer’s preference

A direct request for a person should have a direct route. Also plan for frustration, repeated failures, and sensitive issues. Make the handover context useful and the expected response time honest.

04

Review the boundary over time

Sample both escalated and un-escalated conversations. Look for unnecessary handovers and situations where the assistant should have stepped aside. Update the rules with your team and retest the affected scenarios.

READY FOR THE NEXT STEP?

See the experience for yourself.

Bring a shopper question to the interactive demo.

Try a conversation