Artificial intelligence may offer great potential, but early adopters are running into some of the same problems that have bedevilled human reps in managing customers’ issues, according to research published by Talkdesk.
Based in Palo Alto, Calif., where it provides customer experience automation solutions, Talkdesk commissioned a survey of more than 250 customer experience (CX) decision makers to produce its report, The State of Agentic Automation in CX.
Talkdesk found that while 81 per cent of brands have deployed autonomous AI agents, just 19 per cent have scaled them across multiple use cases. Despite vendor hype, CX leaders suggested they’re taking a cautious approach: 80 per cent said they’ve launched 10 or fewer AI automations so far.
There are some hopeful signs, however: organizations using agentic AI (which can perform tasks rather than simply produce or summarize information) are more than 50 per cent more likely to be very satisfied with its performance compared with using generic chatbots.
“Leading organizations are not just deploying more AI. They are changing how work gets done across three operating-model pillars: coordinated work, connected foundation, and closed-loop performance,” the report’s authors wrote. “When knowledge is not embedded where the work happens, the agent resolving the request—human or AI—has to stop mid-task to find it, and every pause becomes a delay the customer feels.”
360 Magazine Insight
Talkdesk’s report uses data from the survey to build an argument for better AI orchestration. In other words, the company is suggesting that brands shouldn’t simply be assigning AI agents to oversee areas such as billing and technical support, but having them to work together like human team members, sharing information about customers that improves overall results.
This is an entirely valid take, but it also requires more upfront investment and work from brands who may have inflated expectations based on what other AI vendors have overpromised about agents.
Just five per cent of those surveyed can prove return on investment (ROI) for the AI they’ve deployed to date. That means the vast majority are still very much in the learning stages. Telling them to go back and coordinate AI agents could be a hard sell if they’ve yet to see positive bottom-line results.
Talkdesk’s report stands out in how it breaks down some of the challenge by company size. Being a large enterprise doesn’t necessary make this journey any easier, with more than half (56 per cent) citing legacy tech issues and 46 per cent admitting AI is stoking employee fears about their job security.
AI orchestration therefore may also require additional modernization across a brand’s IT infrastructure, as well as more training to reskill, upskill and shift human workers into new roles where they work confidently alongside AI agents.
Talkdesk’s gated 36-page report goes into a lot more depth about CX automation and what AI usage looks like in more advanced, mature organizations. The study also includes a five-stage suggested framework to streamline adoption, with breakouts along common industry verticals such as retail, healthcare and financial services.