How does AI handle currency, time zones, and jurisdictional rules?

When a customer service operation crosses borders, the AI has to handle currency conversion in its replies, time zones for hours of operation and SLA clocks, and jurisdictional policies that differ (refund windows, consumer rights, data-protection rules). AI handles each cleanly when it is configured for them; the failures are usually at the boundaries someone forgot to configure.

A customer in Germany contacts a US-headquartered company about a return. The return window is 14 days under the company's US policy and 30 days under EU consumer rights law. The AI quotes 14 days. The customer is correct that the EU rules apply. The mistake is technically a configuration miss; the consequence is a regulatory complaint.

What people in the field are saying

Service Matters covers the orchestration side of multi-jurisdiction support in "Demystifying orchestration: the key...": routing the customer to the right jurisdiction-aware path is part of the orchestration work AI inherits from the old contact centre.

What does AI handle well?

Currency conversion in replies (the customer's local currency, with the conversion source and timestamp). Local-time formatting (3pm in the customer's time zone, not 3pm at headquarters). Local language and date conventions. Time-zone-aware SLA clocks (the 24-hour window starts in the customer's day).

What needs explicit configuration?

Per-jurisdiction policies: return windows, refund rights, data-protection rules, disclosure requirements. Each has to be encoded as a rule the AI can evaluate against the customer's jurisdiction. The customer's jurisdiction has to be reliably identifiable (often it is, sometimes not).

Where does this go wrong?

The default policy is the company's home jurisdiction, applied everywhere. The AI cites the wrong return window, the wrong refund rights, the wrong disclosure. Regulators in the affected jurisdiction notice eventually. A customer in the affected jurisdiction notices first.

How do you identify the customer's jurisdiction?

From the account record (where they registered, where they ship to). From the channel (the phone number's country code, the IP address). From explicit selection (the customer picked their country at signup). Each can be wrong or stale. The AI should ask if unsure.

What is the practical first step?

Audit the AI's responses for a sample of customers in your top three non-home jurisdictions. Compare what the AI told them against the local policy and law. The gaps are the configuration work. Fix them before the regulatory complaint forces it.