Which AI in CX Use Case Should Go First? 

By Jeff Tropeano

There’s an old joke I really like about a man crawling around under a streetlight, patting the ground. A stranger stops and asks what he is doing.

“Looking for my watch,” he says. “I dropped it.”

The stranger gets down and helps him search. They cover the whole circle of light twice and find nothing. Finally, the stranger asks, “Are you sure you dropped it here?”

“Oh, no,” the man says. “I dropped it over there, in the alley.”

“Then why are we looking here?”

“Because this is where the light is.”

The technology industry does the same thing. Vendors often build where the engineering is easiest to show, not where the operational impact is greatest. Organizations buy in, helping them “search under the streetlight,” investing in AI in customer experience (CX) use cases that look impressive in a demo but do not meaningfully improve the work customers, agents, or leaders experience every day.

The strongest AI in CX roadmaps do the opposite. They sequence use cases around readiness, risk, and the operating conditions that make the technology work. The hard, high-impact moments, regulated guidance, emotional interactions, customer-facing automation, anything that calls for judgment, usually need cleaner data, stronger governance, clearer ownership, and tighter escalation paths before they are ready. That’s “where the watch really is,” but it might be harder to find with less light.

Start Where the Team Can Catch Mistakes Early.

We recommend starting with use cases that support employees and improve execution before moving into the most sensitive forms of customer-facing automation.

This is typically the work your CX operations team already knows how to review.

• Post-contact summaries give supervisors and agents something concrete to check, including the issue, outcome, next step, and required documentation.
• Knowledge support lets teams compare an AI-recommended answer against approved policies, knowledge articles, and the customer’s actual situation.
• Quality analytics gives quality assurance (QA) teams a way to see whether AI is flagging the right patterns across interactions.
• Internal search and insight extraction help leaders test whether AI is finding useful trends across known data sources.
• Simulation-based training lets trainers review the scenario, coach the response, and adjust the curriculum before customers are affected.

Keep in mind these are rarely the flashiest demos, but they’re easier to govern, easier to measure, and more likely to strengthen the operation before the organization takes on higher-risk automation.

Match AI CX Tools to the Function They Need to Improve.

Once the early use cases are clear, group them by the work they are supposed to improve.

A training problem does not need the same AI CX tools as a root-cause visibility problem. For example, long ramp times may point to simulation-based training, performance support, and targeted coaching. Poor visibility into why customers are calling back may point to automated quality, analytics, and pattern detection. Pressure to reduce customer effort may eventually point to self-service or customer empowerment, once the journey and escalation model are ready.

This keeps the roadmap from turning into a shopping list.

When we ask our clients what function they are trying to improve, and how much autonomy they are ready to support, it usually means looking at categories such as:

• Attracting and developing talent
• Assisting live interactions
• Empowering customers
• Digesting data and unlocking insights
• Securing and optimizing operations
• Embodied AI and robotics

The category that belongs first is the one that is most ready, most valuable, and most manageable now.

Sequence the Risk Before You Scale.

A strong AI in the CX roadmap usually moves in stages.

• Phase 1 starts with clearer value and lower operational risk. These use cases are easier to review, measure, and adjust, making them a stronger basis for building confidence. In many contact center environments, that includes knowledge support, post-contact summaries, simulation-based training, quality analytics, and internal search or insight extraction.
• Phase 2 takes on a deeper workflow change. This may include advanced coaching loops, forecasting support, workflow orchestration, mature real-time guidance, and bounded customer empowerment. These use cases need stronger alignment, better data, and a clearer operating model.
• Phase 3 is where leaders should be most selective. Broader customer-facing automation, emotionally sensitive service moments, regulated guidance, high-stakes decisions, and embodied AI usually require the strongest governance and escalation paths before they scale.

We saw this with one healthcare client. The technology could automate highly sensitive patient interactions. The better design used automation behind the scenes to support the scheduler, agent, or nurse while the patient still received a human, empathetic experience.

That is where objective expertise can help. COPC Inc. helps leaders decide which use cases are ready now, which need more preparation, and how to sequence people, processes, governance, and technology together.

The Demand Won’t Shrink. It Will Grow.

It’s important to mention that there is fear running underneath a lot of these decisions: that AI is here to cut headcount to the bone. History suggests otherwise.

In the 1860s, the economist William Stanley Jevons noticed that a more efficient steam engine did not reduce coal use. It made coal more useful and affordable, which helped drive demand higher. The lesson is often called the Jevons paradox: efficiency raises consumption when it lowers the cost of doing more.

The same thing is coming for customer service. As interactions become cheaper and easier to deliver, customers will expect more of it: faster answers, more proactive support, better digital options, and smoother escalation when they need a person.

The winning operations will not be the ones that automate the most work. They will be the ones who use the efficiency to deliver more valuable customer experiences. That is why sequencing matters. Leaders are not racing to the smallest possible operation. They are deciding what kind of service model they want to build as AI changes the economics of delivery.

Before choosing the next use case, download COPC’s full executive guide, AI in CX, 2026: How to Modernize Your Contact Center Tech Stack Without Losing Control.