Keep Your Contact Center AI Stack Flexible Without Vendor Lock In - Content Guru

Picking the wrong AI vendor today could set your contact center back for years.

That’s the core message from Rhys Harris, Product Director of AI at Content Guru, who joined CX Today to break down why flexibility needs to be baked into every AI decision — not bolted on later.

The conversation covers a lot of ground. Harris draws a direct parallel between the mistakes organizations made during the cloud migration era and the trap many are sleepwalking into now. Over-reliance on a single provider, opaque consumption pricing, and contracts that couldn’t accommodate change all created what he calls “cloud regret.” The same dynamics are already playing out in AI.

On the question of what actually separates a genuinely multi-vendor strategy from one that’s just multi-vendor on paper, Harris is blunt: compliance and governance assurance, swappable AI models that can adapt to different languages and use cases, and vendor-agnostic orchestration. Organizations that treat these as checkbox features rather than real requirements will feel it.

He also pushes back on the rush to deploy. Forbes data already shows 25% of tech leaders have invested in AI too quickly, and Harris argues the root cause is almost always the same.

“A lot of organizations have just tunneled into taking a purely AI strategy without thinking about the outcomes. And I think the decisions that people will make is not about the technology but about the outcomes that they were looking to achieve first.”

The bottom line: what works today won’t necessarily work tomorrow, and the vendors worth betting on are the ones who can move with you. 

Transcript

SPEAKER_00 0:07

Hello and welcome to CX Today. I'm Nicole Willing. If you're exploring AI to improve customer experience, but you're also conscious of how quickly the market is changing and the risks of locking into a single vendor, this conversation is for you. We're looking at how to keep your contact centre AI stack flexible. And to do that, I'm joined by Reese Harris, who is product director of AI at Content Guru, an expert in this space with a clear view of the decisions, risks, and opportunities that leaders are facing. Welcome, Reese. Thanks for joining me.

SPEAKER_01 0:35

Okay, well, thank you very much and good afternoon. It's a real pleasure to join you and super excited for our interview today.

SPEAKER_00 0:42

Yeah, I'm sure we're gonna have a really enlightening conversation for our viewers. So when when um CX leaders talk about this concept of AI lock-in, what are they usually worried about? Is it the model, the vendor, the contract, the architecture, all of the above?

SPEAKER_01 0:58

Yeah, a combination of all of the above. I think primarily it's it's about vendor lock-in, and that's sort of comprised by a combination of the other three. You know, contractual flexibility is massively important, the technical architecture of any uh integrations that have been developed to help power the AI solutions is very, very important as well. Um, and then also the models and being able to select the the correct one for your capability. Um, you know, an orchestra has many parts, and you know, CX as a whole is, you know, and your CX um sort of outlook as an organization as a whole is is like an orchestra. And an orchestra has um you know a pianist, it's it's got woodwind, it's got the brass section, it's got drummers, and those on their own would not be as exciting if um or as capable if if if you just left them to yourselves, it wouldn't achieve the same outcome. Um and that's you know akin when when delivering CX. Um picking a an organization that you can work with that will chop and change those necessary parts to correct uh create the whole uh correct ecosystem outcome for you as a as an organization and as a consumer is is massively important. Um and when delivering CX, um choosing an individual AI provider who might be the leader now is not possible and you know will not be the same way um forever as well. So um ultimately end customers don't typically end end up being AI experts or maybe particularly forward-facing in in terms of of the future, um, and really you need to be thinking about your consistent competitive advantage and and differentiation. So um for me, you know, primarily concerned organizations should be primarily concerned about being locked into a single vendor, um, but that's underpinned by the other three, and and you need to work with a vendor who can offer you the partnership capability to be flexible in those three key areas as well.

SPEAKER_00 2:53

Sure. And I guess there's a risk, you know, that um an organization could kind of drift into lock-in without realizing it. So what are the early signs that that might be happening?

SPEAKER_01 3:03

Yeah, and and that typically ends up being the case when you've picked a single vendor uh and a single AI vendor for multiple capabilities. If if you look at your portfolio and you think I've taken this and actually, you know, Organisation X has provided me three of my four capabilities at this point in time, you start to really begin to wedge yourself to a single solution, which might not be the best outcome for your organization than the next three or five years. So um ultimately you need to be conscious of working with an organization that can provide different AI models or different um products with with AI solutions underneath that can operate across multiple regions for the sake of data governance, multiple languages for for the sake of organizational expansion, um, but also flexibility in in terms of price and and capability because of the rapid changes that that are being seen within the AI space.

SPEAKER_00 3:59

Yeah, exactly. So then when um they're choosing those capabilities for the contact centre, what are the most common mistakes that teams might be making?

SPEAKER_01 4:07

Yeah, I I for me it it it goes back to rushing headlong into or rushing head on sorry, into um projects and and big transformational programs without really having a clear definition of what you would like to achieve first. Um so for for me, making sure that organizations have the correct key performance indicators and the correct targets of of what they're looking to achieve and that they're balanced targets as well, because you know there will be cost drivers, people will want to you know potentially cut um cut heads out of of their contact center spend or their CX spend, um, but ultimately you need to be looking at the the sort of key metric, which is ultimately the the CX and the capability that you're providing to your end customers because that ties to your organizational revenue. Um and from that and from selecting your key performance indicators and what are the metrics for for these sort of programs and the definition of success for these capabilities, is then about assessing the correct products or providers on the market and partners that you could work with to achieve that. Um, you know, it it's similar in a sense to cloud. You know, early cloud adopters um faced evolving regulatory requirements related to data storage, to privacy, things like GDPR legislation changed and came in. In the US, things like the California Consumer Privacy Act came in, and that really sort of hampered organizations because they were tied into big contracts that they hadn't properly thought through. Um, so you need to ultimately go back to working with a vendor that's going to offer you that that flexibility and the forward-facing guarantees. Um, data sovereignty and and you know personal information has become potentially a really crippling area for organizations if they get it wrong. So they have to be very um close to towards that and forward thinking in in that space. Um and ultimately defining these, not not by running before you can walk. So actually looking at can I achieve some of the KPI gains that I've been looking at for this overarching program by through an incremental approach and picking something that might be more of an augmented solution or a human in the loop based solution before actually saying, you know, I want to fully automate this portion of the of the journey and dive into that first. So um new technologies deliver the greatest value when adopted strategically rather than rushing into large-scale transformation. And it's already seen as you know, Forbes have said, that 25% of tech leaders already have reported investing in AI too quickly, and in my mind it's almost certainly because they didn't go back to look at what they wanted to achieve from it first, they just wanted to adopt an AI strategy to to appease the powers that be.

SPEAKER_00 6:52

Yeah, exactly. It does feel like companies have kind of thrown AI at things just because they need to be seen to be doing something in AI. Um you mentioned as well um the early cloud era, um, because you know there's there's similar similarities there. We've kind of seen that before where people got locked in, you know, with vendors, with CCAS. So um what did organizations learn the hard way that they could then apply to their AI decision making?

SPEAKER_01 7:19

Yeah, I I think you know that organizational knowledge is is really, really important as a sort of first point on that. Um the leaders that are coming through today may may in some cases not even have spent a huge amount of time going through that big cloud transition and and and migration, so may not have of some of the battle scars there that um some of the more um experienced colleagues within the industry may have. So, you know, the mid-2000s and the introduction of of large-scale public cloud infrastructure really shifted away from traditional on-premise IT to that um cloud, and and it was really driven by the low upfront costs, the elasticity, the ability to react to demand. Um and you know, I think it it's similar types of concerns, as you said, that that you will see when adopting AI. Um and it goes back to heavily heavy reliance in the initial days of cloud on one provider, left organizations really exposed to to major outages, service disruptions, or uh potentially long-term contracts with a vendor that wasn't flexible to work with different organizations and partner with them to achieve the needs and their ever-evolving demands as well. Um, it's not correct to make the best point-in-time decision for now and expect that to be the same in five years' time. You need to make sure you're working with an organization who are going to help you. And you know, we saw firsthand that uh over time organizations experienced cloud regret in in certain cases because they did uh experience unexpected costs because they were locked into consumption models that they haven't hadn't properly forecasted for, or you know, maybe were a little bit opaque. Um so I think again it comes back to the same sort of key principles that number one, you need to work with a CX organization and vendor that um is able to look at all the AI capabilities that are out there and available to help you select the best that are ultimately achievable ultimately best for your CX needs as an organization. Um, but also that you're not tied into usage-based costs, which may you know may be very unpredictable for for yourself as an organization as well.

SPEAKER_00 9:29

Yeah, yeah, definitely. So then kind of you know taking that by a lens of how they evaluate vendors, um are there like three non-negotiables that would tell them they're buying flexibility rather than getting locked into a closed ecosystem?

SPEAKER_01 9:43

Yeah, I I think the the first one um would be compliance and governance assurance, just simply due to the risks that sit around those and making sure that um the partner that you're working with gives you the the flexibility and and the transparency about how that data is going to be controlled, processed, subcontracted in certain cases, and and audited across the regions that you work in. Um ultimately then swappable AI AI models and swappable models for the use cases or or for your industry is important as well. Um being able to adapt those models to different languages if you're a multinational organization and people that you know purport that as an important part of their value add is really really key to picking somebody that's going to be with you across the length and breadth of your AI journey. Um and then ultimately vendor vendor agnostic orchestration. So working with an organization that that partners with best in class providers to assist your use case, that offers um you know flexibility and and transparency in the selections that they've made and why they've made those selections be that through uh the use of of benchmarking reports or or other metrics as well.

SPEAKER_00 10:59

Yeah, that leads into uh what I was gonna ask next, which is like what's the difference between you know an AI strategy that's multi-vendor on paper, but one as you use the word swappable, what makes it genuinely swappable in practice?

SPEAKER_01 11:11

Yeah, so it it it's a combination of architecture, contractual flexibility, um, and you know, I I think again it a lot of the themes for me go back to to key performance indicators and metrics, defining what's critical what's the critical outcome for your business first, and then making sure that you're working with a provider who will help you know work towards those needs as well and offer you the correct flexibility if if those needs aren't aren't being met. Um, you know, in in traditional CX that's typically focused very much around outages and about stability of services. Um but for AI, you know, there's an opportunity to take those KPIs a bit further and look at things like quality and latency and things like that for different use cases as well. So you know the the vast majority of uh CKS CX software options are rigid. You'll have to use their own AI tool or uh you know maybe something from a very closed ecosystem of organizations that they work with. Um and it may look like um they have on paper multiple different levels of of API, AI API integrations, but really you know the organizations that are truly multi-vendor are looking out for your use cases, the the performance indicators that are important to you, but then also the the hosting, um the data sovereignty and the governance options. For us, it's it's really key for uh to be able to offer that flexibility and and give customers the opportunity to plug their own large language models in, to maybe you look at using the latest and greatest that's only available through a hyperscaler and to discuss transparently what those options are. But then also, you know, if you've you know have a use case where you really need that close attention to, I need to know physically where my data is, being able to provide that option as well. And and for us and and and for the strategy that I look after within our brain orchestration layer, um, being able to offer those three points of of flexibility is is is key for us as well.

SPEAKER_00 13:12

Yeah, makes a lot of sense. So I always like to end with the foil looking question. Um looking ahead um you know for the next year or so, what do you think good will look like for organizations? You know, what decisions might leaders be glad they made early in the in the process?

SPEAKER_01 13:28

Yeah, I I I think um you know it it goes back to, as I said, measurements and and ultimately the outcomes and the value that you're looking to achieve first. Um, a lot of organizations have adopted um and cheesily call it the the old McDonald's strategy of of AI AIO, but just have tunneled into taking a purely AI strategy without thinking about the outcomes. And I think the decisions that people will make is not about the technology, but about the outcomes that they were looking to achieve first, and then taking the right tools, be that AI or otherwise, to achieve those. So um, and to do that, you need to not only think about the here and now, but what the future might look like and and being open to change in that as well. So, you know, ultimately what the best tool is today may not be the best tomorrow. Um there's such a rapid advancement in in um AI innovation that you need the ability to swap every couple of months to to the latest version of a model to be able to achieve the best outcomes. Or um, you know, if as I said previously, if if your organization's expanding maybe into different areas that you're working with, a vendor that can offer you that flexibility for the those target markets and the different languages. So you know, the ability to sort of extend that in the same way as as as it was previously with user interfaces and making sure that if you were opening a uh uh uh uh you know an operating entity in a different area that the technology that you had and were contracted in to implement allowed your your end users to be able to utilize it because it had the correct language support. Um technology doesn't just change, but regulations and unfortunately the geopolitical landscape can can also change very quickly. Um and I think you know I've said this said this once and I say it again, you know, the European AI Act for me is is quite forward-facing and I feel presents a very nice blend between not stifling innovation but also ensuring that the use of AI isn't negative or harmful to ultimately the consumer populace, but also to organizations. So bearing that in mind, looking at organizations that are that are uh EU AI focused, I think is is is important, particularly in my opinion, um, because geopolitical situations can change and the ownership of your data, the location of your data um you know can can be impacted alongside that. Um there's a lot of um you know large language models is is a very broad um umbrella term. Most people tend to think about foundation models like Chat GPT or or um you know Claude, etc., when when they're thinking of of those, but there's a lot of specialist large language models that are uh smaller than those but much better for individual types of tasks that can be used. And I think uh if you're wedded into somebody who's very locked to a single large language model provider or in partnership with them, you miss the um capability that other smaller vendors have have to offer. Um and you know, we've we've seen people who are you know ultimately mandating, you know, I I want generative AI, I I want agentic AI, which are massive growth areas, but actually, you know, agentic AI might not be the best solution for the tasks that they have at hand because you know, like humans, uh in a way that these are thinking systems that aren't always going to take defined pathways. And ultimately, if you want to take a payment for your customer or um you know do something like cancel a bank card or something similar, organizations need those journeys to be correct 100% of the time, not 95% of the time, because it's it's gone off gone off the rails at that certain point. Um I think the the key thing you know for us it's it's similar to to a mortgage broker, right? You need to, if you went to the market to buy a mortgage, you'd want to go to a broker who's who's got multiple options, who can give you the best choice and flexibility to fit your individual needs. And consumers and customers need to be working with a a vendor who's going to be clear and transparent with them, who's gonna you know listen to their needs and and the outcomes that they're looking to achieve for their business, take a consultative approach to help them achieve those, and by doing that, select the best tools, the best AI solutions, the best large language models, the best product wrappers to help an organization in that way.

SPEAKER_00 17:50

Yeah, absolutely. So it's clear that leaders have a lot of different things to think about and consider. Um so thank you, Reese, for for bringing your perspective and you know some practical guidance there for our audience. Um, where can people go to learn more about Content Guru?

SPEAKER_01 18:06

Um so I I think you know ultimately the website's the one-stop shop, and there's plenty of access for um people through the website to to contact our team. Um our customer success function does a fantastic job of working in partnership with organizations because you know I I've talked a lot about metrics and KPIs, and that can be very, very scary for organizations as well. So there's a sort of plethora of different services ranging from if you're taking your first steps, human in the loop, augmented based solutions, all the way through to you know pure automation that that we can offer and ultimately guide our customers on and help them select the best underlying technology for those products as as well, um, to tackle all their language, data sovereignty or or or whatever needs as well.

SPEAKER_00 18:53

Sure. Well, thank you, Reese, for taking the time today. It's a really important conversation for our audience. Thank you.

SPEAKER_01 18:59

Thank you very much. It's a real pleasure to join you. Uh I've had a fantastic time today, and um, you know, if if if anybody that's that's watched this would like to speak more, uh feel free to reach out either via our website or to find me on LinkedIn. I'm I'm very happy to chat and offer my guidance.

SPEAKER_00 19:14

Great. Thank you. And to our viewers uh watching, um, to continue exploring these topics, you can visit our website cxoday.com. Make sure that you are subscribed to our newsletter for the latest updates, and continue the conversation on our LinkedIn community. Thanks for watching.