Case studies are reported as operational metrics: queue depth, average handle time, first-contact resolution and the share of minutes resolved without an agent. Names are used where the customer has agreed.
A voice agent handles dispute intake end to end, verifying identity, reading the transaction back and filing the case in the core system. Agents now receive only the escalations that need judgement.
Peak-season status calls, previously 61% of inbound volume, are answered by an agent that reads the tracking system directly. Average handle time on the remaining agent-served calls fell because the simple calls stopped competing for the queue.
Recognition and generation run on in-country infrastructure with no audio egress. Appointment booking and rescheduling complete in a single turn, with confirmations on WhatsApp.
An outbound campaign archetype places renewal reminders and hands interested customers to a licensed agent with the policy already open.
Every figure is taken from the platform reporting tables of the deployment named, comparing the 90 days before cutover with the most recent 90 days. Deflection counts only conversations that ended without an agent assignment and without a repeat contact inside 72 hours.
For serious evaluations we introduce you to an operations lead running Ittisal in a comparable queue.