Agentic AI is changing customer experience. Is your organisation ready? Skip to main content

Agentic AI is changing customer experience. Is your organisation ready?

Tom Denbigh

Reflections from our Opticon Unwrapped roundtable on what it really takes to make agentic AI useful, trusted and scalable. 

Every year, Optimizely's Opticon gives us a view of where digital experience is heading. This year, the direction felt unmistakable. 

AI is moving from something we prompt to something that increasingly works alongside us. Virtual teammates can join meetings, make recommendations and take action. Experiences can be generated around the individual rather than built page by page. And customers will increasingly use agents to research, compare and take action on their behalf. 

But the thing I keep coming back to is the growing gap between what is becoming possible and how most organisations actually operate today. 

A few weeks after Opticon, we brought together digital leaders from financial services, charity and enterprise technology for Opticon Unwrapped. The conversation went beyond the technology itself. We talked about readiness, data, governance, personalisation and what happens to the customer journey as agents become part of it. 

Here are the themes that stood out for me, and the questions I think digital leaders should now be asking. 

Capability is outpacing readiness 

The pace of change is extraordinary. New AI capabilities are emerging in months, sometimes weeks. Most organisations simply cannot change at the same speed, and nor should they try to adopt everything that becomes possible. 

The question is no longer just can we build this? It is whether you can make it useful, trusted and scalable. 

The agent itself is actually quite a small part of that. The bigger challenge is everything around it: governance, decision-making, knowledge, operating models and how quickly your organisation can move from an idea to something that creates value. That enterprise-readiness question was a recurring theme in the roundtable. 

The question to ask yourself 
If an AI capability could materially improve part of our customer experience tomorrow, are we actually set up to adopt it, govern it and learn from it quickly enough? 

That, to me, is a much more interesting measure of AI readiness than how many proofs of concept you have running. 

Readiness starts with data 

AI is only as good as the information it can draw on. 

If that information is outdated, incomplete or poorly governed, AI does not necessarily stop and tell you. It can give you the wrong answer with complete confidence. At Opticon, we heard this neatly described as 'garbage in, gospel out.' 

That becomes even more important as AI moves from answering questions to making recommendations and ultimately taking actions. 

Trusted sources, ownership, accuracy and governance therefore aren't things to solve once you have worked out your AI strategy. They are part of the strategy. 

The question to ask yourself
Do we know which content, data and knowledge we would trust AI to use on behalf of our brand and our customers today? 

If the answer is no, that is probably one of the most useful places to start. 

Manage agents like teammates 

One of the ideas I like most from the Opticon conversation is the concept of AI agents as virtual teammates. 

But if we take that idea seriously, we also need to manage them like teammates. 

You wouldn't employ someone and, on their first day, give them unsupervised access to every system, every customer record and permission to make any decision they liked. 

Agents shouldn't be any different. 

We need defined roles, appropriate permissions, clear guardrails, monitoring and explicit points where human judgement remains necessary. Data sovereignty and understanding where information is processed, stored and potentially used to train models also become fundamental enterprise questions. 

The question to ask yourself
If AI starts acting rather than simply advising, are we clear about what it should be allowed to do, what it shouldn't, and where a human needs to stay in control? 

Governance shouldn't be there to slow innovation down. Done properly, it gives organisations the confidence to go further. 

Personalisation could finally become genuinely personal 

We've talked about personalisation in digital experience for years. 

The problem has rarely been the ambition. It has been the operational reality of creating enough content, variants and rules to make genuinely individual experiences possible at scale. 

Generative experiences potentially change that equation. 

Instead of deciding which predefined segment somebody fits into and showing them a version of a page, experiences can increasingly respond to intent, context and behaviour as they happen. 

That starts to move us from personalising pages towards individualising experiences. 

For CMOs, that's a much bigger opportunity than simply getting AI to produce more content. 

The question to ask yourself 
If the operational limits on personalisation disappeared, what would we actually want the experience to be for each customer? 

And then the harder follow-up: Do our data, content, technology and governance give us the foundations to make that possible? 

Your next customer might be an agent 

The other significant change is happening on the customer side. 

Customers are beginning to hand research, comparison and even purchasing decisions to AI. An agent could compare options, understand loyalty benefits and complete most of a transaction before asking the customer to confirm. 

That changes some fairly fundamental assumptions about digital experience. 

We've spent years designing websites primarily for people. Increasingly, we need to think about experiences for people and agents. 

AEO and GEO are already part of that conversation. Emerging approaches such as WebMCP point towards the next step: AI being able to interact directly with digital services, completing forms, making bookings or carrying out tasks rather than simply finding information. WebMCP itself is still emerging rather than an established standard, but the direction of travel is important. 

The question to ask yourself 
If more of our customers' discovery, research and decision-making happens through AI, how does the role of our website and digital experience need to change? 

That isn't a question for the future. It should be shaping decisions about content, platforms, data and customer journeys now. 

Faster doesn't automatically mean better 

There is also a less exciting but equally important side to all of this. 

AI makes producing the first version of something incredibly easy. But a polished first draft isn't necessarily a good answer. 

We are already seeing pressure to deliver more quickly simply because AI makes greater speed possible. The danger is that we optimise for volume and speed rather than quality, differentiation and customer value. 

The opportunity should be the opposite. AI should create more space for human judgement, expertise and creativity, while removing lower-value effort. 

The question to ask yourself 
Where will AI genuinely improve an outcome for our customers or our teams, rather than simply help us do the same thing faster? 

That distinction matters. 

So, where do you start? 

There is a lot happening, and nobody has a complete map of where this ends. 

For me, that's exactly why the answer isn't to start with the technology. 

Start by understanding how your market and customer journey are changing. Look at where AI could genuinely create an advantage or where it could disrupt the relationship you have with your customers. Then be honest about whether your organisation has the foundations to respond. 

From there, prioritise a small number of opportunities that can create meaningful impact and use them to learn. 

That might mean creating a more relevant, individualised customer experience. It could be preparing your content and digital estate for AI discovery. It could be introducing agents into marketing operations, increasing the pace of experimentation or enabling customers' agents to transact directly with your services. 

But don't try to do all of it at once. 

How Netcel can help 

This is the thinking behind our Impact Framework and the way we guide organisations through their agentic journey. 

We don't believe the answer starts with an AI maturity score or a list of technologies you should adopt. It starts with understanding what is changing around you, where your organisation could be exposed, where the real opportunities are and which of those opportunities are worth pursuing first. 

From there, we help you connect the ambition to the practical foundations needed to deliver it: customer experience, content, data, technology, experimentation, governance and the way your teams work. 

It's a journey from understanding the opportunity, to defining where to focus, proving what creates value and then scaling what works. 

Because the goal isn't to adopt agentic AI faster than everyone else. 

It's to make better decisions about where it can create an impact, and build the capability to act when it matters. 

Explore Netcel's Impact Framework