Synopsis: In this AI Leadership Insights video interview, Mike Vizard speaks with Gregory Whiteside, CEO for, about how conversational data is applied to AI.

Mike Vizard: Hello and welcome to the latest edition of the video series. I’m your host, Mike Vizard. Today we’re with Gregory Whiteside, the CEO for And we’re going to be talking about how conversational data is applied to AI, and what the benefits of that will be. Gregory, welcome to the show.

Gregory Whiteside: Thanks for having me.

Mike Vizard: So what exactly do we mean here? Where does conversational data fit in, how will AI be applied? What’s going to be the ultimate business benefit?

Gregory Whiteside: Yeah, so we’re helping companies that have large amounts of unstructured data, turn it into custom AI. And also extract from that data very valuable business insights. So we started Human First because we recognized that an increasingly large amount of this type of data was being produced by Enterprise across a lot of channels. So of course contact centers, but also things like product reviews and NPS reviews, and we realized that this data was really the key for companies to build very custom and high-quality, AI as well as to improve their product, their operations, and their customer experience.

Mike Vizard: Is this the data that’s floating around in our emails, or are these phone calls? I mean, what is the source for all this data, and how much and what types of it are there?

Gregory Whiteside: Yeah, we talk generally about natural language data, which is any form of data that’s unstructured in text. So, produced by humans, if you will. That includes things like call center logs, live chat transcripts, emails, NPS survey results, verbatims, really anything that’s unstructured. We started and focused really on making sure that for multi-term dialogue, so conversational data, our tooling helped make sense of that type of data, but it also works across simple things like paragraphs of text or documents.

Mike Vizard: And ultimately, what are some of the examples of the use cases for applying AI to that data? What can I do today that I wasn’t able to do yesterday?

Gregory Whiteside: Yeah, I mean what’s really important for companies today is to automate and to find ways to improve the customer experience. And a large part of that goes through, obviously, AI. And in order to have AI that really understands your specific business and your specific customers, you need to train it on your own data sets.
So this type of data is really what companies need to use in order to get to AI that is not generic, is not black box, and kind of gives insights but that are hard to action. It’s by building AI that’s really tuned to what you’re trying to solve with it, and by starting from that data, that you can get to results faster that lead to higher value.

Mike Vizard: So I take that data and I give it to you, and it goes into what? Some sort of AI model is trained. Is it a large language model? Is it machine learning algorithms? What flavors of AI are we using?

Gregory Whiteside: So we’re a general data productivity suite for natural language. The first real application of this suite and of this tooling was to help companies build natural language understanding models, so NLU. So it’s a self-serve SaaS tool, so all our customers use it themselves, and it’s simple. You ingest all of your data from different channels, and then Human First provides you the ability to efficiently explore that data using machine learning and AI capabilities. And it does that so that anyone, regardless of technical capabilities, can explore and understand and transform that data into high quality AI training data sets.

Mike Vizard: And what do you think ultimately will be the impact of generative AI, since you can’t walk down the street these days without somebody leaping out to tell you about their great new thing? But it seems to me we’re going to be living in a multimodal AI world with different things used for different use cases. So how do you see this all evolving?

Gregory Whiteside: We’re very excited, obviously, by what’s happening with generative AI. Because previously, the main problem that companies had to solve when it came to training custom AI was understanding the user input. Because the user input dictates what the AI should understand, and that’s really what you are trying to optimize so that the AI can classify, for example, what the customer is saying into the right intent, and then do something with it.
With generative AI, you now have the other side of the model that you also need to understand and optimize, and that’s the data coming out of these models. Because of course, with generative AI, there’s the word generative. We don’t control what those LLMs are going to produce.
And so it becomes a increasingly important problem to solve, is not only looking at the, for example, customer utterances going into let’s say a chat bot, but also understanding how is the LLM responding? And improving that response. And so that requires a lot of sifting through larger and larger amounts of unstructured data.

Mike Vizard: So we don’t have a great track record in technology when it comes to improving customer service, right? We first started out with these voice response systems, press five and nobody answers. And then we got to digital transformation initiatives where, frankly, we’re just pushing data entry from my employees out to the end customer. And a lot of people are kind of skeptical.
Will AI be different, and how so? Because, can we make the actual customer experience better and something that they may enjoy or prefer? Versus, is this just another alternative or another way of shifting the work around?

Gregory Whiteside: I think AI is definitely going to help us improve customer service. The earlier generations of AI, and of language models like natural language understanding models, I think a lot of companies got to the point where they were able to apply this type of AI to start automating and improving their customer service. The problem is that the language is a very long-tailed problem. And so understanding what customers are asking in the first place is a hard problem to solve.
And a lot of companies kind of hit a bottleneck, because they went with a very top-down approach to the problem of thinking they understood what the types of things that they needed to automate were. When in fact, there’s a large number of utterances and things that they didn’t know customers were asking, and that needed to be automated.
I think with LLMs and generative AI, we’re obviously seeing much more powerful models, things that you can feed data and it’ll do a very good job of information retrieval and of automating the process of answering simple questions. I think the key question is, of course, how much are we going to trust these models to take on bigger responsibilities?
And when it comes to things like transactional exchanges, where we’re going to depend on a model to go and do API calls, or to do some kind of backend transaction, this is where the question of trust and the question of really making sure that we understand how the models are behaving is going to become critical.

Mike Vizard: Do you think there’s also generational issues here? I mean, I’ll talk to some younger folks who prefer to engage with a bot. They don’t really want to talk to a human. And of course, there’s other folks who are like, all they want to do is talk to a human because they can’t quite wrap their heads around how the machine’s supposed to work.
Do you think this will change? It feels like the way we interact with computers is changing, and they’re going to be more natural language driven, and the machines will understand us more than we understand them.

Gregory Whiteside: I think AI has the potential to improve, really, the CX. And in the case of customer service, certainly provide increasingly better experience. I think you’re right. In the last few years, chat bots have gotten a bad rep, because they ultimately had very little AI in them. And even though we talk about the field of conversational AI, it’s really just with these large language models, and ChatGPT opening the floodgates, that we’re at a point where the personalization, and the fact that every response can really be adapted to the context and what the user asked, that it will feel much more like talking with a human.
And I think ultimately, bad customer experience comes from long wait times, humans not having the right answer, and there’s such a high turnaround in these industries that it’s very hard to keep people that have great knowledge of a business.
So I think AI obviously can help improve the customer experience, and I think it also has the opportunity to help the humans that right now are doing this job maybe get to better, higher quality type work where they’re working alongside the AI.
I don’t see, necessarily, the case where the humans will be replaced. Rather, I think they’re going to be really helping and managing the AI.

Mike Vizard: Do you think also, we’re going to get to a point where you can’t be competitive if you don’t have some sort of AI capability? Because the customer is going to expect it, and if you don’t have it, you’re going to be viewed as somewhat antiquated.

Gregory Whiteside: I think that’s, yeah. I definitely agree with that statement. There’s really, of course, infinite use cases now for applying the types of models that we’re seeing coming out. I think most enterprise are still at a very early stage of figuring out exactly where they will apply this in a safe and trustworthy way. And so I think we’re at the very early stages. But from our experience, we’re seeing all of our customers and all of the companies in this space kind of very, very heavily looking into what they should be doing with these large language models. Because it’s undeniable that they will transform businesses.

Mike Vizard: Where should folks get started? I think a lot of folks are looking at this stuff, and there’s so many possibilities now that they’re a little overwhelmed. So is there someplace that they should be focused on. And say, start here with something small or interesting, or maybe big enough that it matters to the business? But what’s the path forward look like?

Gregory Whiteside: Yeah, I mean, that’s a great question, and it’s really ultimately what we set out to help solve with our data-centric tooling. Because we believe the ground truth in your business is really in this data. It’s telling you what customers are feeling, it’s telling you where there’s operational problems, it’s telling you where there’s product improvements that can be done.
But the problem is, until now, that insight has been locked away. And very, very hard to access. And most of the companies that we interviewed before starting Human First were using Excel as a tool to do that exploration, and to understand their data.
So I think the starting point really starts from the bottom up. Going in the data and, as you said, identifying those low hanging fruits that can be turned into automation, or that can simply be turned into better ways of doing that within the business.
I don’t think everything necessarily needs to go through AI. Not everything needs to be automated, and a lot of the value you’ll get from this type of data, again, can inform simply how to better change your products so that it serves its purpose. But what AI will do is help accelerate that process of exploring the data, and making sense of it much, much faster.

Mike Vizard: Ultimately, what will be the impact? I mean, I don’t think customer service reps are going away, per se. But certainly, roles and functions are going to change. So as you look over the horizon, what does the future of customer service look like from your perspective?

Gregory Whiteside: I think it’ll be better. Because where things can be automated, they will be. And in a better way than what we’ve seen in the past, with kind of very hard-coded flows. I think that humans will be used more efficiently. So the triage layer of customer service is going to be highly improved with ai, allowing humans to take on kind of the edge cases, but the high-value ones where their ability to do that for other humans will make the difference in how the customers perceive the brand.
I think ultimately, AI is going to really help businesses that harness it stand out and provide a better user experience. Which is today, one of the biggest moats that a brand can build.

Mike Vizard: Folks, you heard it here. AI is here, it’s not going away. So now it’s all about how well you’re going to use it. So the time to start figuring that out is right now. Gregory, thanks for being on the show.

Gregory Whiteside: It was my pleasure. Thanks for having me.

Mike Vizard: All right. Thank you all for watching the latest episode of We invite you to check out this one and others on our website. Until we see you again next time, stay safe.