Keith Dawson's Analyst Perspectives

Artificial intelligence (AI) has become ubiquitous in discussions of contact center technology. Vendors are rushing to incorporate it into platforms and applications. And end users have understandably mixed feelings about where it makes sense to use and what its impacts will be. No one should be surprised that AI has arrived, especially for customer support: Software companies have been working on customer experience (CX)- -related AI applications for many years, and the fruits of those efforts have been gradually working their way into real-world tools, particularly chatbots. Solutions that help automate the self-service gateway to customer support are able to boost speed and efficiency in customer interactions, and this can lead to faster response times, reduced wait times, and improved first-contact resolution rates, all of which can have a positive impact on customer satisfaction and loyalty.

The spotlight has been on self-service because the latest AI developments have come through large language models (LLMs) that facilitate exactly what people want to do when they encounter a chatbot: have a natural conversation enroute to retrieving information and resolving an issue. One of the great questions arising from this capability has been what it means for agents. Will increased automation mean fewer agents are needed overall? Or does it help agents do more, and do it better? The answer is yes, and yes.

There are contact center use cases for AI that have the potential to reshape how agents spend their time, which ultimately may be what finally turns customer support from a cost center to a profit center. Those use cases sit behind the scenes, in managing the information inside an organization, and in smoothing agents' access to all sorts of resources. Buyers are both tantalized by the potential cost-saving aspects of automating more agent work and genuinely fearful of a potential disruption in operations. The industry needs more vendor guidance in finding the specific use cases that can be deployed quickly, using existing platforms. Vendors are in the best position right now to articulate how conversational systems used in chatbots are also able to run knowledge management, process agent evaluations and reduce the amount of after-call work time.

The pandemic era changed service delivery from the point of view of both the customers (creating higher expectations) and agents (moving to remote or hybrid work). What did not change was the fundamental organizational structure of centers: agents in teams, assigned to channels or queues, supported by supervisors focused on activity-based KPIs like speed of answer and call duration. When attention turned to agent-engagement issues three years ago, organizations looked to technology to knit together dispersed agent teams and provide them with the tools to do their jobs under new circumstances.

AI can do much to alleviate (if not solve) longstanding problems related to attrition, quality and performance management. What makes it transformational is that it makes processes more efficient and accurate by supporting agents in the background, preserving many existing structures for how agents are managed and evaluated. What we are seeing from vendors are applications targeted at use cases that have very specific automatable processes that already exist and that can be preserved but optimized. Considering how conservative contact centers can be when it comes to adopting new technology, that kind of assurance may make transitioning to AI systems easier than people think.

Through 2026, technology for agent management will expand beyond tracking and performance measurement to include a wide array of AI-enhanced tools. These are examples of use cases in agent management that buyers should explore. Each is supported by multiple vendors in the marketplace, making it possible to compare approaches:

• AI that uses LLMs to generate conversations for self-service chatbots can use that same facility to create summaries, identify dispositions and create notes about future actions. The value here is the immediate time savings that can shave seconds or minutes off each interaction. Reducing after-call work time increases the amount of time agents spend with customers and allows managers more flexibility in scheduling employees. Automated summaries and notes also serve to formalize institutional knowledge.
• A related application is in knowledge management. When an AI model has access to a large corpus of company-specific data, including product information, customer comments, postings from communities and many other sources, it is easier for employees to find the information they need. These tools can also automatically tag, summarize and extract key information from documents, improving search and retrieval. By identifying and recommending relevant knowledge resources, documents and best practices to agents based on their specific needs, they can speed interactions to quicker resolution.
• Natural language AI is a boon to evaluating and scoring agents in the quality process. Instead of a supervisor randomly choosing which calls to review, and only being able to score a tiny portion of the total, an automated system can review all interactions. It can track sentiment on both sides of the conversation and evaluate agent performance in a way that appears fairer and less biased. Feedback to agents is faster. By reducing the time between an interaction and its quality evaluation, you increase the chance that an agent will be able to use the feedback to improve subsequent performance. If feedback is available in real time, agents can take immediate steps to recover from a poor interaction.
• Real-time insights are themselves a powerful use case. Although uptake has been slow, agent guidance and assistance systems promise to empower agents with suggestions related to three key areas: the customer's intent, knowledge resources relevant to an inquiry, and processes related to compliance that must be followed.

These are just some of the use cases related to agents that have emerged so far; others are no doubt on the horizon. For example, the same logic that applies to agents (i.e., that you would use AI to automate repetitive tasks related to documenting interactions and spotting opportunities) also applies to sales teams that use contact center tools for outbound prospecting and lead qualification. Multiple vendors have added sales-focused modules to their platforms that enforce consistency and automatically capture content. It is likely that other types of uses will focus on automating inter-departmental processes and on surfacing relevant information for knowledge workers at critical moments.

In general, buyers and vendors alike should look to AI to solve for specific business outcomes. In the contact center, that translates to time saved on interactions and more productivity out of agents. While it may be possible to reduce agent headcount with these productivity gains, practitioners should instead be thinking of these gains as allowing for longer, more complex interactions that can be handled by the same agent pool. It opens space for finding agents with cross-sell or upselling skills and turning them loose on a larger pool of opportunities.

The takeaway for buyers is that modern automation and AI technology adds to agents' abilities without necessarily disrupting the overall structure of a center. It can unearth savings in speed and productivity that enable flexibility and enhance agents' ability to solve more complex issues. And it makes an organization's knowledge resources more consistent and available, empowering agents and making them more engaged with their work.

For more insights into the changes being wrought by AI systems, see Dave Menninger’s Analyst Perspective on Generative AI and Rob Kugel’s on broad business use cases.

For more information on contact centers and customer experience, visit our Customer Experience expertise area.

Regards,

Keith Dawson