The emergence of Conversational AI agents is poised to bring profound changes to the contact centre industry.
With most CX platforms now offering autonomous agent capabilities across their offerings, it’s clear that the Agentic Age has finally dawned. Within the contact centre industry, the implications of this are profound.
Given that much of the work of human agents involves following specific scripts and performing specific actions, the concern is that this work will increasingly be performed by AI agents.
The Agentic Age refers to the period where most digital tasks are performed by AI agents that can reason and act on their own. But how do traditional contact centres even begin to adopt these agents? And how do they find effective ways to leverage both AI and human agents to the advantage of the customer and the company?
Splitting the work load
AI agents are not simple chatbots. They can now have conversations at the level of human experts. They are designed to automate rule-bound engagements with more consistency and at a higher level of compliance than human agents.
AI agents are especially effective when it comes to dealing with the kind of rule-driven, repetitive calls that take up most of a contact centre’s time. In a bank setting, for example, the most frequent query is usually, “What’s my bank balance?”. In a telco, it often relates to data or billing queries and in online retail, it increasingly relates to logistics and deliveries.
Those calls can quickly stack up, particularly at certain times of the year (such as the festive season), resulting in long waiting times at best and unanswered queries at worst. That, in turn, equates to a poor customer experience, which is ultimately damaging to the company’s image.
Assigning these high-volume calls to a AI agent can help. Think about how many human agent hours could be saved if a virtual agent could answer the most common, simple, and repetitive queries. Then think about how much customer frustration and brand damage could be avoided. Suddenly, having more virtual agents in your CX team becomes a no-brainer.
The danger of pilot purgatory
Automating these calls is often made to sound easy. Most AI Agent software providers make it sound incredibly simple to set up and get working. It feels you could have an AI agent answering your calls and working seamlessly with your various operating systems within a matter of minutes. Yet, like everything that sounds too good to be true, there is a catch. While these AI-first agents can handle unstructured conversations well, they really struggle with conversations that require the application of complex rules to many contexts across multiple channels and systems.
That is when things get tricky. After FAQs and basic requests are automated, the rest have to be transferred to live human agents. The effort to script these complex, dynamically shaping conversations is too great and the cost too high. The result is that full automation is never realized and high volumes of digital conversations still have to be handled by live agents.
AI Agents for CX
To get Conversational AI Agents working at scale in production requires a rethink in the logic models. Most solutions have opted for a Probabilistic First – Deterministic Second logic approach. And while this holds great promise in non-regulated contexts, it fundamentally limits the range of conversations that can be fully automated in a regulated company.
To make automation viable within these organisations, a Deterministic First – Probabilistic Second model needs to be adopted. Conversational guardrails must be prescribed based on rules, yet dynamically shaped by context (data). Conversational fluency must be retained, yet it must be layered over a structured approach to conversational logic.
Done right, Conversational AI Agents can be trained to have simple to highly complex conversations with customers, totally unassisted. This transforms the self-service space, making it viable to get most conversations fully automated safety.