AI can improve customer experience by helping staff find reliable answers, identifying when customers may need support and making guidance more relevant.
The test is whether customers can achieve what they came to do—with less effort and appropriate help when something goes wrong.
Through the AI Awards, I examine examples of organisations applying AI. On the AI Ireland Podcast, I speak with practitioners about how technology changes their work. In my keynotes and leadership workshops, I connect those perspectives with the decisions businesses face.
Three AI Awards 2026 finalists illustrate different approaches to customer experience. Taken together with Irish consumer research, they offer a useful way to decide where to start.
What do Irish customers want from AI?
A 2026 study commissioned by CCMA and conducted by W5 surveyed a nationally representative sample of 1,012 respondents between 21 April and 11 May.
Its published findings describe greater acceptance of AI for straightforward, transactional tasks, alongside a preference for human support when issues become complex or sensitive. 77% wanted to be told when AI was being used. Read CCMA’s published research findings.
My interpretation is that convenience and access to people should be designed together.
A quick answer is useful. A quick answer that leaves someone unable to resolve their problem is not a successful customer experience.
Three Irish AI customer experience examples
The following examples come from the published AI Awards 2026 finalist summaries. They describe shortlisted projects, rather than independently verified performance results.
| Organisation | Published approach | Leadership question it raises |
|---|---|---|
| Laya Healthcare | Supports advisers with policy-specific information, member context and references | Can staff find and verify the right answer? |
| Headon Boxing Academy | Identifies behavioural signals that may indicate cancellation risk, supporting staff outreach | Can earlier attention lead to a helpful conversation? |
| SSE Airtricity | Uses generative AI and smart-meter information to provide personalised energy guidance | Can customers understand and act on the guidance? |
Explore the AI Awards 2026 finalists.
Laya Healthcare: improve the information behind the conversation
Laya’s Knowledge Hub shows why an AI customer experience project need not begin with a customer-facing chatbot.
Helping an adviser locate relevant information can be a valuable place to investigate.
For your business, examine a question that staff struggle to answer consistently. Identify where the answer lives, which version is authoritative and who maintains it.
Then test whether an assistant helps the adviser reach an accurate answer with less searching and correction.
Measure: answer quality, time to resolution and repeat contact—not simply how quickly the tool responds.
Headon Boxing Academy: connect a signal with a useful response
Headon’s Pulse illustrates a different application: helping staff decide where to focus attention.
A prediction does not explain a person’s circumstances. A change in behaviour could have several causes.
The important design question is what happens after an alert. Who reviews it? How do they approach the customer? What would helpful outreach look like?
Measure: the usefulness of the intervention, false alarms and customer response—not just the number of people flagged.
SSE Airtricity: make personalisation useful and understandable
SSE Airtricity’s finalist summary describes an evaluation approach that checks accuracy, tone and relevance.
That is a useful combination to examine in any personalised service.
An answer can sound tailored while relying on an incorrect assumption. Test whether the information supports the recommendation and whether customers understand what to do next.
Measure: accuracy, comprehension and successful action—not simply whether the message sounds personal.
These proposed measures are my recommendations for evaluating similar projects. They are not claims about results achieved by the finalists.
The podcast insight: technology still depends on human connection
In my conversation with Kevin Magee, Chief Technology Officer at All human, we explored the relationship between people, processes and technology.
One point captures the discussion:
“I don’t believe you can take the human element out.”
Listen to Kevin Magee on the AI Ireland Podcast.
For leaders, my takeaway is to examine the whole journey.
A customer might receive an automated answer, speak to an adviser and then need a specialist. If information disappears between those steps, the customer still experiences a broken service.
Ask whether the next person can see what has already happened. Can the customer correct a misunderstanding? Is there a clear route to someone with authority to help?
The quality of those transitions matters as much as the quality of an individual AI response.
How to choose your first AI customer experience pilot
Start with an observed problem rather than a preferred technology.
| Problem you observe | Approach worth investigating | What must be in place |
|---|---|---|
| Advisers spend too long searching for answers | An internal knowledge assistant | Current source material, access controls and review |
| Staff notice disengagement too late | Predictive signals supporting outreach | Appropriate data, a defined response and outcome tracking |
| Customers struggle to understand their information | Personalised explanations | Reliable inputs, tested guidance and access to help |
This is a decision aid, not a recommendation to automate every situation.
Choose one task with a clear owner and enough regular activity to evaluate. Some problems may be better addressed by correcting confusing information or simplifying the process first.
Measure the customer’s outcome
Before a pilot begins, agree what success means.
A useful scorecard includes:
- Resolution: Did the customer complete the task or resolve the issue?
- Accuracy: Was the information correct and appropriate?
- Effort: Did the customer have to repeat information or contact you again?
- Human support: Could they reach suitable help when needed?
- Staff workload: How much checking, correction and follow-up remained?
- Cost: What did setup, operation and maintenance require?
Shorter interactions can look efficient while transferring work to customers or colleagues.
Record a baseline and compare similar cases. My guide to measuring AI value explains how to structure that assessment.
Use the 5Ps before expanding
I use the 5Ps to connect an opportunity with the organisation’s readiness.
People: Who is affected, and what support will customers and staff need?
Process: Where does the journey become difficult? How are exceptions handled?
Platforms: Can the systems share appropriate information and support a handover?
Proprietary Data: Is the information accurate, current and suitable for this use?
Products & Services: What should improve for the customer, and how will you know?
If the team cannot answer one of these questions, make it a preparation task before expanding the pilot.
Questions business leaders often ask
Does AI customer experience always mean a chatbot?
No. It can include adviser support, predictive signals, information retrieval and personalised guidance. Choose the approach around the customer problem.
Where should a business start?
Select a recurring problem with reliable information, a responsible owner and a measurable outcome. An adviser-facing pilot may be worth considering when staff can check outputs before customers rely on them.
How do we know whether AI is improving the experience?
Measure resolution, accuracy, customer effort and repeat contact alongside cost and staff workload. Faster responses alone do not establish better service.
Bring these examples to your next leadership event
Customer experience is a useful subject for an AI keynote because it connects technology with decisions people recognise: how to respond, when to intervene and what a good service should feel like.
I explored AI and human experience in my keynote at Three Ireland’s Forward: Fast conference.
For your event, I can bring together Irish project examples, practitioner insights and the 5Ps to help your audience examine its own opportunities.
Discuss an AI keynote for your event.
About the author: Mark Kelly is an AI keynote speaker, Founder of AI Ireland and host of the AI Ireland Podcast. His work connects practical AI examples with business leadership decisions.
Share This Story, Choose Your Platform!
Planning an event or leadership offsite?
Mark Kelly delivers evidence-based AI keynotes and hands-on AI leadership workshops for boards, executive teams and conferences across Ireland and Europe.





