Stop Chasing AI Hype and Start Delivering Real Outcomes

Rhys Fisher, Associate Editor at CX Today, sits down with Kevin McGachy, Head of Solutions at Sabio, to cut through the AI noise that's plaguing contact centers.

If you're tired of flashy AI demos that promise the world but deliver generic chatbots, this conversation is your reality check.

Kevin breaks down the costly mistakes organizations make when racing to deploy AI, why voice is still king despite the digital self-service push, and how outcome-based strategies can protect you from vendor lock-in and wasted investment.

This isn't another AI hype session; it's a practical roadmap for CX leaders who need results, not regrets.

In this frank discussion, McGachy exposes the pressure-driven mistakes that are costing organizations time, money, and competitive advantage:

• The ROI Trap: Why building business cases solely around cost reduction creates friction and misses the real value of AI – like addressing high attrition through “zero human impact” automation that doesn't eliminate jobs

• Technology-Agnostic Design: How to avoid vendor lock-in by designing systems that can swap between LLMs (Gemini, GPT, etc.) based on performance and cost, rather than betting everything on today's ‘winning’ model

• Voice's Staying Power: Why customers still prefer voice for urgent or emotional issues – and how modern conversational AI creates fluid, human-like experiences that adapt in real-time versus rigid IVR decision trees

• Outcome-Based Pricing: Sabio's approach to only charging clients when AI successfully resolves customer interactions, eliminating the risk of paying for underwhelming results

Transcript

SPEAKER_00 0:08

Hello and welcome to TX Today. I'm Reach Fisher, the Associate Editor, and today I'm going to be speaking with Kevin McGaki, the head of solutions at Startfield. Kevin, thanks for joining me. How are you doing today?

SPEAKER_01 0:20

I'm very good. Thanks for having me.

SPEAKER_00 0:22

Absolutely, absolutely. And yeah, I'm looking forward to the chat. You know, today we're going to be, you know, inevitably the competition will be rooted in AI, but you know, we're going to we're going to try and keep it a little bit fresh by really, I think, sort of digging down into the actual application of the technology for CX and customer service professionals who are kind of working with it at the at the coal fit, I guess. So on that point, we all know how quickly AI uh innovation is moving right now. From your perspective, what's the biggest mistake you see organizations making when they try to just kind of keep up with AI?

SPEAKER_01 0:58

Yeah, sure. I mean that there's quite a few different um challenges, I guess, we've seen, or kind of trends that we we identify. So some of them are around the the sake of just enabling AI for the sake of enabling it without really understanding what's the objective, what's the use case, what's the value that we want to try and drive from it. And we see that, I guess I understand it, but business cases are so often centred around um the ROI and about the cost reduction. Um and then you know that creates a bit of concern or friction inside businesses, uh, where actually that doesn't necessarily need to be the case, right? We have lots of high attrition in our industry, and there are ways in which we what we call the zero human impact factor, which allows you to look at actually what should attrition rate versus how you might be able to automate that demand. So I think really how people are approaching building out their ROI models to understand where they should do investments, um, along with looking at like what is the technology we're enabling and why are we enabling it. It's just not really approaching it from the the right perspective, I would have said.

SPEAKER_00 2:05

Yeah, yeah, that's really interesting. That's definitely ROI is kind of the big link we always see with AI, but like you said, there are there are definitely other ways to look at it. Um kind of leading on from what you said there, I guess when you talk with CX and IT leaders, what fears or perhaps pressures are driving that urge to kind of constantly switch platforms and chase the latest tool, like you mentioned?

SPEAKER_01 2:30

Yeah, sure. I mean I think everybody recognises that if you aren't acting now, then you're probably going to fall behind. And actually it's a bit of a race in that the you know the the chance of catching back up in the future just becomes ever more difficult. So I understand that like we we have lots of execs who are pushing and have a lot of pressure to get going. I think we we also see with the the lot of hype that exists around the market that is so much communication about what the models can now do, etc., that um whilst one proof of concept is nearing an end, someone's already decided that we have to change the model and jump to the next thing. And actually, like for our perspective, that's not really the best approach to take. You know, if you're able to drive an outcome with the technology that you've chosen to build your POC on, then perhaps that is still the right choice. Um, there's obviously ways in which to test the next one and see if you evolve. But um, I think people are just like, now this function's been released by the vendor, we should turn that function on and let's just see what happens. Rather than again, it goes back to the first point about building up the right ROI, the right strategy about how you go about um deploying the solution. But again, everybody's under pressure to find some cost reductions and to not fall too far behind, the catching up becomes impossible.

SPEAKER_00 3:46

Yeah, yeah, it's definitely it's it's a tricky situation for organizations right now. It's interesting you mentioned outcomes there, because I know that's something that Sabio often talks about. It's outcomes rather than technology. In practical terms, I guess, what does an outcome-led AI strategy actually look like inside a business on day-to-day?

SPEAKER_01 4:07

Yeah, it's a good point. So, like we um like we've kind of changed our mindset that uh like as a business we've always been the solution centered, so not technology. It's more about actually what's the problem that you have, what would the best solution look like, and then let's go and find the technology that would help you to meet that desired solution. Um, we've kind of gone a step further beyond that now, which looks at outcome-based commercial models as well. So um, with a lot of the hype that exists in the AI space and people have a lot of concerns about that ROI, like I talked about, um, our approach has been well, let's benchmark your performance of your organization today. Let's look at what an agentic AI offering could be in that space. So, actually, how could we help to facilitate and automate and allow customers to self-serve through the voice channel? Um, and if we do that, we look to be paid on an outcome. So only if we actually manage to solve that customer's interaction does a customer pay for it or the client pay for it. And that allows them to have um much more guarantees and assurances about what has been deployed. Like I say, it's not AI for AI's sake, it's very much what's the value to your business, and then if we only if we can demonstrate that value back to you does a client then pay. So it means all the massive marketing percentages of cost savings you can get, you know, those massive headline figures, that you have something here which allows you to um be guaranteed ultimately that that's going to end up happening, right? Otherwise, you're or you're you're not like you'll need to pay for it.

SPEAKER_00 5:36

So yeah, you mentioned voice there, and I think this is quite an interesting trend almost that we're seeing in the space right now. That you know, despite the the strong push towards digital self-service, you know, voice is still a really preferred channel by customers in a lot of cases. Why do you think that it's why is voice still so powerful even in this kind of AI-driven world?

SPEAKER_01 6:00

Yeah, I mean, and I think voice from an AI perspective has always been the hardest channel to automate, etc. So like it's kind of typically it's kind of come a bit last. We've all gone through kind of digital transformation projects to try and manage like the how do we get customers online and things. Um, but from the research that we've looked at, like you know, the market will tell you that voice remains the dominant contact channel. There's a lot of different reasons for that. Like we can speak faster than we can type, so therefore the pace at which we communicate allows us to to get across faster what it is we want to try and convey, particularly if if you have a sense of urgency or um particular emotional connection to the things that you want to talk about. Being able to convey that in words is much easier than trying to write it out in a really long message, especially most of our stuff is done on mobile phones today, so actually the pace at which you can type on a phone is even slower than it is on a keyboard. So there's a lot about the ability to get a connection when you speak to someone, to feel like I can just rabble on all my points, like get across to you what my messages are and allow you to digest that and then and work it out. But I think like we we're the same, like I'm the same, right? Despite the fact that everything I do is all about AI, if I really need to get my point across, I'm still picking up the phone. And we also have seen that there's a um perception that the kind of latest generations um don't want to use voice and they don't want to make phone calls, it's all about text messenger. Actually, we see that that's not quite the case. Um, and it may be when we talk about voice that um they become voice notes, not necessarily an actual live phone call. So like there's lots of different ways you interpret it, but um and actually voice notes is a great example about the communication style, right? If you think how easy it is to give a voice note and how annoying it is to listen to one, right? Like it just takes so long. So um, but it just shows you that it's for all of us, we see it as the easiest way to get across all of our points, like especially when it's uh immediate action that's required or uh something that you've got emotional connection to. Like we all find that is the the best communication channel.

SPEAKER_00 8:06

Yeah, yeah, it's interesting. As you were speaking, exactly what his picture was voice notes and kind of how popular that is, especially with younger people now. Exactly, it's exactly like you said, and kind of rather than sending 25 paragraphs to your friend on text, you can send the voice note now. Yeah, it makes perfect sense. Um but it's interesting, I guess, how voice has still evolved with AI, although they often kind of pitted against each other almost using AI, like you said, more in the digital self-serve realm. I guess how do you think modern AI has changed what's possible with voice automation compared to you know the old IVR experiences that uh people used to dread, I guess.

SPEAKER_01 8:45

Yeah, and I mean like I I've been responsible for building some of those experiences and you know, like the deflection to the websites, like I've done all of those kind of journeys. Um and it's where we we genuinely believe that the technology has evolved enough that you that's just no longer the way we should treat customers, it's just not how you should deliver a service. Um and if we think about the the differences in the traditional versus what's now possible, right? So that even using conversational AI, so the ability to say why you're phoning and go through a bit of um a flow, you are ultimately in a decision tree. You know, like what you say and how it responds is all prescribed, it's it's all written out. Whereas in the new world, um like the we were doing this with a client earlier, we're going through like an ID and V journey, for example, and we ask someone for their account number, nobody has their account number to hand, like that's just traditional. And so in an old world, you would then capture perhaps their postcode and some other details about them. But if the person then has the letter and finds that account number, they have no way of going back in the process to then give you that bit of data. Whereas the new world is very fluid, like we we have an outcome to achieve, so we still have to ID and V you. But if all of a sudden, you know, you say they don't have it, and then we say, Well, what's your postcode? And you then you say, Oh, wait a minute, actually, I've just found a letter. You can jump around that flow and still meet all the policy requirements and authentication requirements, but the customer's experience is like much more human-like. You know, it's like if you and I were like if I was the agent and you with customers, I would just naturally manage my way around that. So I think you you see a lot of that. The other um positive that's coming out of the use of AI and voice is around translation as well. So the multilingual element. It's definitely a you know, we service um customers well, so for Sabia would do it across the UK and Europe predominantly. Um, we have lots of different languages exist in that space. We as businesses have different customer bases, so using real-time translation allows us to manage those conversations in the customer's native language and the agents' native, and we manage to communicate and help everybody to stay together. And I think the the last part is that um the technology has changed so much in that the way in which you deploy it, the pace at which you can change. So, you know, the old world would be raise an IT request, it may go out to a supplier, perhaps. Maybe it's got a three or four-week turnaround even to have it actioned. Just now you're talking about making changes that are near real time, adapting to something the weather's just turned, and all of a sudden there's an influx of demand. Like you can manage that situation in real time rather than like having to either predict for it or um, like I say, waiting weeks for it to actually get actioned out.

SPEAKER_00 11:28

Yeah, yeah, that's really interesting. I like what you said about it being fluid, I think that's a really good way of looking at it. And like you said, the old the vision a lot of people have of what it's in their head is still that kind of that regimented IVR system. But yeah, being able to have that flexibility now, I think, is a massive, massive change and definitely an improvement. Um so just I guess the final question here if you were advising a TF leader that's planning the 2026 roadmap today, what would you prioritize to future-proof the AI and voice strategy without kind of locking themselves into the wrong systems that, like we touched on at the start, that they're gonna be stuck with and maybe regret as the technology, as everything moves forward?

SPEAKER_01 12:12

Yeah, I think like so how we've been approaching this and therefore as how we would recommend to others is um to design in a technology agnostic way. So, you know, today's latest model, Gemini 3, is the leader in this space. Like, I mean, no doubt in a couple of weeks' time, like OpenAI will trump up with the next model, you know, like it just happens all the time. And so if you lock yourself into one of those vendors, then actually you create a challenge for yourself that if someone else can do something better, then you miss out on the opportunity to use that model or to use it. Um, and also we see that like the older generation of models, like some of which um, if you use minis and things, can be a little bit cheaper, are actually great at doing the job, right? If if you look at the task that you're trying to solve, if you look at the value and what the outcome is you're trying to drive towards, you may have multiple models in play at any one time, taking advantage of what each one has to offer. But that's a like that's a very big architectural design decision that you have to take at the start of that journey. Um, and I understand that others maybe um have uh an alignment with certain hyperscalers or certain vendors, and therefore like they don't have as much flexibility in that space. But we find that to be massively important that the way in which you design it and the way in which you go um about building your journeys is agnostic as much as possible of set technologies so that you because for me that's the only way future-proof, right? Who knows what the next provider is going to be that we haven't heard of yet? Like, you know, you just don't know, and so you have to keep yourself in that kind of open space. The the second part is like everyone knows that um building these journeys and building these models only works as well as the data that you can provide it, and it's not that you have to do a massive data like transformation project before you can do AI, there's definitely tactical ways or being very regimented and controlled and the use cases that you start with to make sure you you don't cause yourself challenges, but um knowing that you are going to have to make systems accessible to these technologies and you are going to have to fix some of the data elements at some point behind the scenes. So kind of having that as part of your overall strategy is important and not having it as a standalone or two separate kind of business areas, which is what we we sometimes see as the case. Um, and then lastly, is like as part of your overall strategy, you have to consider um what's your business operating model longer term, you know. Like we see it as a world of um human and agentic agents, managing customer demand, working side by side, you have like a quality team already internally. They'll just be assessing the quality of both types of agents, and maybe maybe they don't even know one's agentic and one's human, they just look at the scorecards that come out and feedback and things. Um so I think really having a good I think having an open platform, understanding that data needs to form part of your AI strategy, and lastly considering what is your overall operating model, are probably the three top areas that um we see people need to focus on that we're trying to help them do today.

SPEAKER_00 15:20

Yeah, thanks for that, Kevin. I think that's uh yeah, a really good detailed kind of overview of the of the key areas that people need to be considering. I think just more generally the chat, like I said at the start, we all know how ubiquitous AI is right now, so it can sometimes be a little bit tricky to have sort of a new conversation around the subject, but I think in really digging into the details about specifically, like you said, the outcomes. What can this AI actually deliver when it's implemented correctly? I think yeah, it's gonna be really helpful for our audience. I'm sure they're gonna really get a lot from the chat. So yeah, thank you again for your time.

SPEAKER_01 15:52

No, no worries. And like I said, that for me, the clients and an organization should look for vendors and partners that are going to help them to deliver their outcomes. And something like an outcome-based pricing model, you know, where where you only pay on success, allows you to have a bit more um security in your own decision making that you know that you're going to have a partner there to help you. Um, but absolutely look for the value of what it is you're trying to achieve.

SPEAKER_00 16:16

Yeah, I think that's great advice and a great place to end things. I did just want to quickly thank our viewers as well for tuning in. If you enjoyed this chat, remember to like and subscribe to the channel and head on over to CX Today for more articles like this, or for more stories like this, rather. Uh, until next time. Thank you for watching.