How to upskill human agents in an AI-powered contact center

From chatbots to intelligent IVRs to real-time agent guidance, AI is fully embedded in the contact center. Where does that leave human customer service agents?

Half of all contact center leaders believe AI will reduce workloads in the next two years, but 87% are still planning to create new positions or backfill roles. The need for human agents isn’t going away, but the nature of the role is changing.

There’s no doubt that AI is already resolving customer queries that would have previously gone to a live agent. And as generative and agentic AI technology continues to advance, it will increasingly deflect calls to human agents. Gartner has gone so far as to predict that agentic AI will resolve 80% of customer service queries without human intervention by 2029.

However, there will always be some cases when it’s best for customers to speak to a human agent, especially in situations involving:

• High complexity (e.g., the customer needs to complete a process that requires multiple inputs and systems, such as filing an insurance claim or matching with a health coach)

• Emotional stakes (e.g., a customer recently lost their job and is calling their utility provider to ask about payment assistant plans)

• High upsell or cross-sell potential (e.g., a satisfied customer shows interest in a premium product or service but needs to talk through the decision with an agent)

So while humans will field fewer straightforward queries, they’ll likely spend more time on the most complicated types of interactions.  

“Agents used to have a breather between complex calls,” says Creovai COO Simon Black. “Now, self-service deflects all the simple questions. Agents are left with only the hardest, most urgent, most emotional ones.” This puts agents at risk of burning out and making errors, which negatively impacts both contact center operations and the customer experience.

Contact center leaders must work with their agents to build the skills necessary to handle these complex interactions—and to collaborate with AI. Below, we’ll look at five skills agents need to succeed in an AI-augmented call center (and how managers can help them build these skills).  

Emotional intelligence and empathy

AI may be able to mimic empathetic communication, but it can’t truly empathize—and customers can tell. Almost 60% of consumers say companies have lost touch with the human element of the customer experience.

Empathy is especially important when customers call about an emotional issue, such as an inability to pay their electric bill, a medical procedure that wasn’t covered by their insurance company, or money lost due to fraud. Agents must be able to recognize the customer’s feelings, mirror their energy appropriately, and de-escalate sensitively.  

Agents with high emotional intelligence make customers feel heard and understood, strengthening trust and leading to better call outcomes. Emotional intelligence also improves the agent experience, with empathetic agents feeling less burnt out and better equipped to handle tough calls.

What managers can do

• Train agents through role-playing and simulations. Organize role-playing sessions with team members or use role-playing AI to simulate emotional customer scenarios. This lets agents practice empathy in a safe environment and get real-time feedback.

• Coach agents with relevant call examples. You can use conversation intelligence software to pinpoint emotional calls based on the phrases customers use. Bring examples of emotional calls into coaching sessions and highlight where empathy helped (or could have helped).

• Prompt agents with contextual real-time alerts. You can configure AI-powered agent assistance software to trigger coaching pop-ups to agents when it detects certain customer emotions during interactions. These pop-ups can remind agents about empathetic language they can use or actions they can take on behalf of vulnerable customers.

Critical thinking

While many customers are happy to turn to a self-service channel for simple questions, 68% say they still prefer to speak to a human agent when trying to solve a more challenging issue. Agents must use critical thinking skills to solve these challenges, which may involve piecing together information from multiple systems, troubleshooting an edge case, or finding a solution to a new problem.

AI can help in these situations—for example, agent assistance software can surface relevant customer data and make context-based suggestions. But agents still need to critically assess what the AI suggests and determine if it truly solves the customer’s issues. And in uncommon scenarios, where there may not be existing guidance or knowledge base content for agents to draw on, they need to investigate potential solutions and determine what they can do for the customer. Without strong critical thinking skills, agents could get bogged down in process confusion or make errors after blindly trusting AI.

What managers can do

• Simulate tricky scenarios. Create multi-step simulations that require agents to draw on information from multiple sources to solve a problem. You can also have your agents review examples of AI-generated suggestions and make judgment calls on whether they are appropriate to use.

• Foster a culture of knowledge sharing. Create peer learning groups where agents can discuss how they have solved challenging issues. You can also use conversation intelligence software to identify calls related to particularly tricky topics and look at how the top-performing agents resolved those issues. You can coach your agents on these examples or even use them to inform your real-time coaching prompts.

• Use real-time workflow software. This software lets you integrate multiple data sources (e.g., your CRM, your knowledge base) into a single interface so agents don’t have to scramble across multiple systems to find information. It also delivers step-by-step guidance for structured processes, allowing agents to spend less time trying to remember processes and more time thinking about how to solve challenging customer issues.

Digital dexterity

As contact centers adopt more AI tools, it’s increasingly important for agents to be digitally dexterous: able to adapt to new technology quickly and confidently. This might mean getting comfortable with an AI-powered agent desktop, effectively using a real-time assistant, or seamlessly switching between providing voice and chat support. Agents who resist new tools or rely on outdated workflows will fall behind; those with high dexterity will use tech to improve speed and accuracy.

An agent doesn’t need to enter their role knowing about every contact center technology under the sun, but they should demonstrate a willingness to learn and an understanding of how technology can help them work more efficiently.

What managers can do

• Offer ongoing tech training. Schedule training sessions when your contact center rolls out new technology. Create screen-sharing tutorial videos or “tech tip” newsletters that agents can review whenever they need to. You could even schedule regular office hours or Lunch and Learn sessions and encourage agents to come with questions or feedback on the technology they’re using.

• Build tech skills development into the agent role. Make tech skill development part of the agent job description so agents see it as a core responsibility, not an extra chore. Carve out professional development time so agents can get familiar with new technology and feel confident using it to solve customer issues.  

• Communicate how technology benefits agents. Some agents may be resistant to change, especially if they have been with the contact center for several years and are used to working a certain way. When introducing a new technology, make sure you’re clearly communicating what the agent will get out of it. (For example, you could talk about how a new QA automation platform will enable agents to get better coaching.)  

Consultative sales skills

In industries like financial services and telecom, agents often have opportunities to cross-sell or upsell. Whether the customer decides to buy can come down to the agent's communication skills. Agents need to be good at listening to customers, evaluating their needs, framing offers in a way that will appeal to them, anticipating objections, and delivering effective rebuttals.

AI can help with cross-selling and upselling. For example, you can use conversation intelligence software to identify the most effective rebuttals to common objections, then trigger real-time pop-ups with these rebuttals during relevant sales conversations. However, a successful sale still requires uniquely human skills. Agents must use their best judgement, pay attention to verbal and tonal cues from the customer, convey confidence, and establish themselves as a trusted consultant.

What managers can do

• Provide data-backed sales coaching. In your coaching sessions, review transcripts to identify missed opportunities or communication slips. You can use conversation intelligence to measure talk-listen ratios, sentiment, and agent behaviors; use this data to coach agents on being more engaging or persuasive.  

• Build sales prompts into real-time agent assistance. Equip agents with AI-powered agent assistance software that suggests sales offers at the right time. For example, if the software detects a customer talking about how they had a good experience with a certain product, it could prompt the agent to offer a related product as an add-on. Train agents to integrate these offers naturally.

• Take a customer-centric approach: Remind agents that upselling isn’t pushy if done right. Encourage agents to frame offers around solving the customer’s problems. Training sessions should reinforce this mindset.

The ability to collaborate with AI

Working collaboratively with AI means knowing when to rely on automation and when to take control. Agents must be comfortable handing off parts of the interaction to AI but also know how (and when) to jump in.

AI can be great for certain tasks, such as summarizing calls, suggesting disposition codes, or answering questions about an interaction. But it can still make mistakes or misunderstand a query, and agents must be able to review AI outputs critically and make adjustments as needed.

What managers can do

• Set clear guidelines around AI usage. Document and communicate what AI should be used for and when agents should take over an automated interaction. If AI outputs (e.g., automatically generated email responses) require human review, make sure you’re assigning people to this task (and notifying them when they have content to review) so nothing falls through the cracks.

• Train agents on chatbot handoffs. Run drills where agents practice escalation scenarios like a chatbot failing to answer a question or a frustrated customer demanding to speak to a human. Teach agents to verbally acknowledge AI handoffs and review information already shared with the chatbot to keep the customer experience seamless. Debrief after these drills to discuss the agent’s communication skills and problem-solving approaches.

• Reinforce that AI is an assistant, not a replacement. Communicate that rather than replacing agents, AI is intended to help them operate more efficiently so that they can focus on more high-value work. You should also coach agents on how to review and fact-check AI outputs; emphasize that they still need to think critically and use their judgement rather than assuming these outputs are always right.  

The new agent experience

The role of the contact center agent already looks a lot different than it did a decade or two ago—and the pace of change is only accelerating. But that’s a good thing for contact centers and the agent experience. AI enables agents to spend less time addressing frequently asked questions, handling routine issues, and doing basic administrative work. Instead, agents get to focus on higher-value work: solving tricky problems, helping customers get more out of their company’s products and services, and strengthening customer relationships.

Of course, the increasing complexity of the agent role can also lead to burnout if agents don’t have the training and tools they need. It’s up to contact center leaders to help agents develop new skills and harness AI to be successful in their roles. By investing in their agents and embedding AI in the contact center, these leaders can boost efficiency while still keeping the human touch in customer service.

‍