← All industries
03 · Commerce support · Returns and compensation
Collect evidence without approving money the assistant cannot spend.
A photo, order number, or complaint can start a review. It cannot authorize a refund or store credit. Tonetify separates evidence intake from the decision owned by support and payment systems.
Reference rollout scenario, not a deployed customer case study.
For commerce operators, support leaders, and marketplace teams
The costly misunderstanding
A bot can promise a full refund or compensation before anyone verifies the order, damage, or policy scope.
What the host knows
A customer uploads a photo of a damaged coffee machine. Support has not reviewed the evidence or approved compensation.
Unsafe reply
“We will issue a full refund and add TWD 1,000 in store credit.”
Bounded reply
“We received the damage photo and will send the request to support for review. Nothing has been approved yet. Please provide the order number.”
Division of responsibility
The product keeps authority. AI stays useful.
The host retains
- 01Order and payment truth
- 02Return eligibility
- 03Refund and compensation approval
Tonetify handles
- 01Evidence-aware reply boundaries
- 02Required questions before review
- 03Escalation without invented approval
Rollout targets to measure
- 01Measure useful evidence completeness on first contact
- 02Measure unauthorized financial promises
- 03Verify that support receives a reviewable conversation record