Q&A: How generative AI is shaping end user experiences

• Generative AI is reshaping user expectations by enabling faster, more efficient and increasingly autonomous service experiences.

• Successful AI integration requires careful preparation, standardisation and risk management to ensure reliability and trust.

• Datacom is leading by example, embedding agentic AI into its own operations to deliver better end user experiences and outcomes, and create seamless zero-touch support experiences.

William Clancy: We’re really leading by example in this space, so we’re doing it for ourselves in the first instance and transforming the way we work at Datacom. We’re a Managed Service Provider (MSP) and by embedding AI into the way we work, we can be the north star for customers, and help push the boundaries of what’s possible. We’re designing services and products with a focus on how we can leverage agentic AI to provide a Zero Touch service – whether its assistance provided to customers or internal support from IT.

Jai Moran: When we talk Zero Touch, we’re looking at how we can reduce reliance on a human to fulfil a request or to resolve an incident. By doing so, we optimise the interaction and deliver the ideal outcome, as soon as possible, as efficiently as possible. We see AI playing a major enablement role here. When people jump on ChatGPT, Anthropic or Perplexity, they think, ‘Isn’t it amazing?’ And it is, there’s no denying it. But when you scale out AI across the business landscape, with use cases that are strict and defined, there’s a lot of complexity and configuration, trial and error. Trying to apply an AI overlay to technical services for a specific industry becomes quite niche.

As with any emerging tech, there’s no standard playbook. We’re dragging that technology into the sandpit and tweaking operations to make it congruent with the technology. We apply that AI overlay to so many different industries that it’s rare we run up against something we haven’t seen before, especially from a local perspective. So when it comes to the clients, we’re helping them clean up their own backyard first. Don’t waste money yet – make preparations! You need to be ready and willing before you turn on the functionality. 

JM: Generative AI sets high expectations for people who are using it. They’re free to play around with it, see how it works and how to adapt for your specific use case. It’s easy and available. It seems simple when you’re willing to fail. If I ask ChatGPT to write something and I don’t get what I want, I can refine it and try again. The result is amazing.

But if I’m an airline, providing critical services all over the globe, the capacity to fail isn’t there. The capability and the potential of AI is outstanding, but there’s still an inherent level of risk. Organisations need to be appreciative of this risk.

With standardised conversational AI, there’s a fixed intent. A set process to deal with a refund, for instance, where we can tell the chatbot to respond in a specific way. When we add a generative AI overlay, we say to it: ‘You have the ability to process a refund, change flight or educate customers on the desk opening times.’ We rely on it to make the determination, and there’s a bit of ambiguity and risk in the metaphorical ‘black box’ the AI sits in. Generative AI needs to be tried and tested to lock it down for real-world scenarios. It’s about meeting customer expectations with an organisation’s risk position.

WC: Organisations are complex, bespoke environments. You need to have ring-fenced that risk through standardisation and controlling the boundaries of what it can make up. We’re doing that work for ourselves and our services; we’ve created boundaries.

Without this knowledge, organisations have to make bespoke solutions, and AI is adding extra ‘bespokeness’ to your bespoke environment, which can be a tangled web of information. We standardise the process in a way that controls the AI. 

WC: We see this happening in two phases. Initially, we’re looking at AI supporting the human, while in the longer term, AI replacing the need for a human to complete low complexity, boring and repetitive tasks. In the short term it’s about focusing on ways we can get things done quicker and more efficiently with agentic assistance. Taking human-assisted generative AI and making it faster and more accurate.

JM: Taking it back to the agentic AI as a black box, it’s only as accurate as the information it’s fed. We can use generative AI to standardise uniform datasets into that black box, so it’s streamlined into a more concise view of the world. It’s like an infant until you educate it. We’ll use generative AI in the first instance with a human-augmented approach, and get that information in so that in the future it can be independent.

WC: Longer-term, we’ll see AI Agents leading our front-end channels for support. Near risk-free workflows, fully automated services, and delivery of true zero-touch IT support. Humans will only get involved if it’s a very complex issue, or with white glove support for VIPs or those who need it. More accuracy, less cost to deliver, and the end user gets a faster, more efficient service so their loss of productivity is significantly decreased.

JM: This extends past AI. Our goal is to create exceptional experiences for our customers and their end users. It’s about keeping the human at the centre, and we do this by embedding an experience management role in our engagements. The purpose of this experience management role is to be user centric: I don’t care about the server my application is hosted on, if it’s running slow, I just want you to fix it for me. That’s the ask for the experience manager: ‘Where do you suffer? Where is your experience poor? How do I improve that?’

Someone might be having a poor experience with a piece of software, but it might not be a technical issue with the software itself. It might be about the approvals process that takes eight emails, for instance. The experience manager sits in there amongst that and optimises the process, so the user walks in happy and leaves even happier.

Can AI play a part? One hundred percent. But can we improve it just by asking: ‘Do we need this? Can we automate this?’ That makes the engagement easier in the first instance. Beyond contracts and SLAs, it’s about you as a human and your user experience.

WC: Focus less on technology and have people looking at the actual human experience. Constantly provide feedback to the technology teams on how it can be improved. Identify where there is an opportunity to use agentic AI. You need some people to be focused on the experience – not just the technology – so you’re designing for the experience.

JM: When I take my car to the mechanic, I don’t want to hear the specific details about the engine. I want the mechanic to listen to me and my outcome and what I need, and then get my car running smoothly. It’s always about creating a better experience for the end user.