Retell AI
80.4% of support calls, resolved alone
Anker sells consumer electronics around the world and takes millions of support calls a year. Its old chatbots and phone menus failed its bar for natural talk and first-call fixes. It put voice agents on real calls in the US and UK. The agents handle product questions, warranty checks, order tracking, and fixes after hours. The vendor reports under 2 seconds of delay and full handling of interruptions.
How it wentThe pilot set three targets. Resolution rate: target 80%, result 80.4%. Net promoter score: target 60, result 63. Coverage of top call types: target 70%, result 70% or more. The figures come from the vendor's case study of its own product.
Try it · based on the source's workflow
Act as the after-hours voice agent for [company name]. Handle these call types only: [order status, warranty check, basic product fix]. Use only our help docs: [paste or link]. For a warranty check, ask for the order number and purchase date, then state the cover from the docs. If the fix is not in the docs, or the caller is upset, take a name and number and book a callback in [system]. Read back any booking or case number before you end the call. Send me a daily report: calls held, resolved, and handed off.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.