Traditional IVR (Interactive Voice Response) menusā€”'Press 1, Press 2'ā€”are being phased out at an accelerating pace. A 2024 Forrester survey shows that 62% of consumers feel 'frustrated' with fixed-menu IVR systems, while intelligent IVR powered by natural language understanding (NLU) enjoys a 37-percentage-point higher user satisfaction rate.

The latest NLU engines can now handle complex intents, such as 'My order was supposed to arrive last Friday but didnā€™t, and Iā€™ve changed my phone number.' The system automatically breaks this down into three sub-tasksā€”order inquiry, shipment tracking, and personal information updateā€”and processes them in parallel. After one U.S. telecom company deployed such a system, its IVR self-service resolution rate jumped from 29% to 67%.

Key technological advances include zero-shot learning (identifying new intents without requiring large amounts of labeled data) and multi-turn dialogue state tracking (DST). DeepSeekā€™s latest model scored 89.6 on a DST benchmark, approaching human-level performance.

GlobalConnectā€™s SmartIVR solution combines LLMs with NLU, allowing customers to interrupt and correct at any time, and seamlessly transfers to live agents. Customer data shows an average 35% reduction in call handling time, along with a 22% decrease in complaint rates. Industry consensus is that intelligent IVR is becoming the first line of defense for contact centers seeking to reduce costs and improve efficiency.