AI adoption outruns readiness as contact centers chase end-to-end resolution
SiliconANGLE Cheryl Knight ● Covered by 2 sources
AI is everywhere in contact centers, but most shops still use it in tiny bits. The real payoff comes when the whole process is wired together, not just one chatbot.
Based on reporting by SiliconANGLE, Cheryl Knight — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Artificial intelligence has spread through customer service fast enough to make “adoption” sound like the easy part. Talkdesk’s Pedro Andrade says the harder question is whether a company can actually use AI to finish a request, not just answer a message or sit a bot beside a human agent.
That split matters because most deployments still stop at one touchpoint. Talkdesk says 98% of companies have put AI somewhere along the customer journey, but only 15% combine agentic AI with orchestration across departments. Andrade’s view is blunt: the model is rarely the problem. Compliance, security and disconnected systems are.
The results back him up. Talkdesk classifies CXA leaders as seeing about four times the net promoter score improvement of less mature adopters, at 22% versus 5%. Cost-per-contact improvements are less dramatic, which is a useful reminder that shaving costs is not the same thing as fixing service. Churn prediction and personalized recommendations show the same pattern of stronger results among the more mature group.
CXA, short for customer experience automation, is Talkdesk’s answer to that gap. Andrade describes it as an operating model that coordinates a hybrid workforce of AI and humans by connecting systems, knowledge and workflows. That is also why Talkdesk built a CXA Operations Center: someone has to supervise AI agents, monitor their behavior and course-correct them, just as with human staff.
The metrics change too. Instead of average handle time, Andrade says companies should track how long it takes from the opening of a problem to the moment it closes. And the ambition goes beyond inbound calls. Service reminders, renewals and collections can all become part of the same journey, because the customer’s phone call is often just the last step after something else already went wrong.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI in customer service that nobody likes because it’s boring: plumbing, process and governance. The industry keeps selling clever demos, but the money shows up when the whole mess is stitched together. Fancy bots are nice; a system that actually closes the ticket is better.
Read more about this at: SiliconANGLE