Hyper-Personalization Using AI-Powered Marketing

Jake Burns:
So maybe you could start off and just tell us a little bit about what is the technology stack that you're using? Because I'm talking to so many customers right now that want to get started with AI and generative AI of course, and a lot of them are just stuck where to begin. So what advice would you give for them?

Vijay Chittoor:
I think the first, specifically in our domain as we think about the nature of AI, it all starts with first having a large amount of data. So in our case, the data is all about first party data of consumers, which is organized at a per brand level. So essentially each of our customers has a large repository of data, which they may or may not have been tracking historically, but with Blueshift, we make it easy for them to get started on that data unification journey, which often I'm sure you're finding this in your experience is one of the key steps to graduating towards AI. So I think the step one is really about having that rich data well organized, being able to capture that in real time, being able to unify that data. But then second, I think when you think about the advice we give to everyone starting on that AI journey is really to think about the end customer first.

And in our case, when we think about the customer, we really think about how can you use AI to deliver personalized interactions to the end consumer? And for us, a lot of that means thinking about customer AI. And when you think about customer AI, it's really about taking that customer data, the first party data we talked about and translating that into the who, what, when and where, of how to engage with the customer. So when you think about traditional marketing, which is often very manual, not driven by AI, you start making maybe flat decisions about who to target for a certain campaign, what offers to show them, when to reach out to the customer and which channel or where should you engage with them. And if you think about it in the apps in a world which doesn't involve AI, when you're making these decisions manually, you are oversimplifying it quite a bit and lumping a bunch of customers together and you're trying to say, well, this whole segment, let's just target them with this one offer.

But the reality is that people, the end consumers are unique individuals and they need to be responding to it differently. And what AI does really well is even when the human market is sleeping in that moment, it's able to make that decision at an individual customer level and make millions of these decisions in aggregate. And I think that's the kind of decisioning engine, and that's the kind of decisioning power, the power personalization that AI gives you. So when we advise people on how to get to that AI journey, start by organizing that data, second thing, customer first, think about the use cases, but then be able to leverage the advantage of AI that it can make decisions at scale, it can personalize to an individual and transform your end customer experience with those elements in mind.

Jake Burns:
Absolutely. Yeah, that's a great point. It's really about personalizing the experience. As a manual process, it would be just too laborious to be able to do that for any human to do even if they were working 24 hours, right?

Vijay Chittoor:
That's exactly right. Yeah.

Jake Burns:
But with AI, presumably it gets it right more often as well because it's using more drawing for more different data points.

Vijay Chittoor:
That's exactly right. And I think you touched on something important. You're thinking about the end customer journey. And if you think about it, a lot of people have been talking about how customer journeys have become much more complex in today's digital world where so many different touch points have emerged. And in that complex, because of that complexity, there are millions of permutations of the customer journey. So in some ways I think the customer engagement problem for today is really about nurturing each customer's self-directed journey because every customer is automatically on a journey with the brand. So, how do you recognize the journey that each individual's on? How can you be helpful to them in that moment and how can you do that at scale? And really that's kind where AI comes in and helps everyone. So, when we work with marketers, I think marketers are very good at being storytellers. But today the challenge is how do you take the kernel of the story but individualize it across all these different self-projected journeys. And that's where I think marketers can partner really well with AI. And that's been a very powerful partnership.

Jake Burns:
That sounds amazing. So let me ask you this, what is the role of humans in all of this?

Vijay Chittoor:
So, I think the humans are the core creative element behind all of this. There are also the strategic drivers behind all of this. So in some ways, I think when I think about a lot of automation technology, the first wave of automation technology, essentially, made it harder for humans to be more strategic and be more creative because I think a lot of that automation was conditional. And if this then that kind of rule-based automation.

And a lot of times marketers and other departments across the enterprise ended up just pushing a lot of buttons and knobs and it took the creativity and strategic thinking out of their jobs. And I think with a new way of AI, which is truly driving real automation where you don't have to be sitting around pressing if this, then that, kind of buttons, you're actually being challenged more and empowered more to deliver the strategic value and the creativity. You can actually now really think about the stories you want to tell to your end customers and use technology as an assistant to deliver those at scale and not be bogged down by fighting against your technology in some ways. So in that sense, AI has unlocked the potential of a lot of humans and we are very excited about that.

Jake Burns:
So kind of more of a co-creator relationship than replacing the human altogether.

Vijay Chittoor:
That's exactly right. And the co-creator analogy, in some ways, sometimes we talk about this idea that everyone becomes an editor, and in a literal sense, people who are now writing are able to get first drafts quickly and are spending more time in editing. But at a more strategic level, you start thinking about the work of humans. Everyone I think in every role across every department, the enterprise is getting elevated to that level of an editor and they don't have to, they'll do less of the grunt work of maybe coming up with the initial drafts and the initial writing.

Jake Burns:
Less of that undifferentiated work and more personalization and kind of final touches.

Vijay Chittoor:
That's exactly right.

Jake Burns:
Yeah. So let's talk about the skillset needed to create a company like this because most of the enterprises I work with, they all want to work with AI, but it's a very hard skill to recruit for because data scientists and anyone in the field of AI is very hard to recruit nowadays. They're very valuable. So what has been your approach to recruiting and getting this talent within your organization?

Vijay Chittoor:
That's a great question. I think part of it is recruiting right, but part of it's also setting the right culture. So when we think of the recruiting, I think we've been fortunate from day one to have AI talent in the company. My co-founder Manyam serves as our Chief AI Officer and he has done some very impressive work dating back to days when the AI was not as much a buzzword. So I think it's great to start with someone like that and build sort of the foundation of the team in the right way. So there's definitely a lot about looking for the right skillset and the talent, but then equally I think the culture is important. So you have to kind of set the right framework for the whole company, not just the machine learning and the AI engineers, but the whole company to sort of be able to leverage these technologies and to be able to take them to customers and make the customers successful.

So when we think about culture, we talk about five core culture values at Blueshift. Those five values, when we take the first letter, they form the word “MORPH.” So the first one M is for make new mistakes. And that's kind of a surprising one a little bit because why would you ask someone to make mistakes? But the key part is sort of the making new mistakes, which is all about learning rapidly, being able to try things out, but also having that constant learning culture and element of curiosity and learning. So we start with that because I think that's very critical, especially with new technologies like AI. Second, we talk about obsessing over customer success. So that's the O in MORPH. And again, I think when you think about technology, for it to be really valuable, you really have to have the end customer in mind.

So again, obsessing over that, whether it's our technology teams or even our marketing and sales and customer success team, everyone's kind of obsessing with that customer success. R is for raise the bar. So we challenge ourselves to be the best version of ourselves and truly thinking about what's the best innovation that we can deliver to our customers. So that's about raising the bar. The fourth one P is for play as one team. So a lot of this innovation for us to make our customers successful, we have to play as one team across the entire company, starting with the folks who are developing this all the way to frontline facing customer teams and so on. Finally, the last one, H is for have fun, seriously. That's just acknowledging that all of this work is going to be hard, but we'll create a culture where we make it enjoyable for everyone to come to work and have as much fun building this and enjoying the journey as much as looking at the destination.

Jake Burns:
You mentioned before, I think it was the M right, make mistakes, that might sound scary to some people. How do you ensure that you reduce the cost of failure so those mistakes aren't catastrophic?

Vijay Chittoor:
Yeah, I think that's absolutely key. So I think there are, when we talk about make mistakes, we talk about make new mistakes and it's much more of the emphasis on the learning culture within the company. But equally, we are talking about obsessing our customer success. There are a lot of use cases that we serve for customers which are mission-critical. And again, as we are grounded in this idea of obsessing our customer success, you have to take anything that's mission-critical, very, very seriously. And that's not the area where you want to be making mistakes.

But balancing that innovation, which can happen behind the scenes as we develop, but then taking the final developed product in a form which truly meets the bar for obsessing on customer success. That's kind of the two things that we have to strive for. And on that note, I think it's great to partner with fantastic companies like HAQM because we rely on HAQM for a lot of our infrastructure. It needs to be reliable, it needs to be performing, needs to have low latency, all of that. And I think that's the mindset we take when we think about delivering products to our customers and obsessing over that customer success.

There are things where you want, when you're prototyping things, when you are building things internally, you do want to make mistakes quickly. You want to have a culture where everyone wants to experiment, but you also want to have a culture which is cognizant of the time where it's not okay to make mistakes and is very grounded in that notion of obsessing our customer success and making sure that we are taking our responsibility to the end customer very, very seriously.

Jake Burns:
Have there been any challenges along the way? And if so, how did you overcome them?

Vijay Chittoor:
That's a great question. I think as we are taking some of this innovative technology to market, there've been a couple of interesting challenges over time that we have now tackled around specifically around AI. So the first one I would say is really about making AI trustworthy and explainable because it's being deployed in enterprise situations. And our customers want to make sure that the experience they're delivering to the end consumer is consistent with their brand and is a smooth personalized value added interaction. So, when you're going and telling the enterprise that the AI is making all these decisions, how do you convince the customer experience team, the marketing team, that the AI is making the right decisions because they would not be able to audit each and every one by looking at it manually because that would almost defeat the purpose. So really I think a lot of the ways we try to solve this challenge, and we've been very successful at that now, is about making the AI explainable at multiple levels.

So, how do you ensure that a non-technical marketer, for example, could come into the Blueshift platform and understand the effects of the AI before it is deployed? And to understand it, you'll create a UI where somebody who just doesn't fully follow all the different parameters of the AI can still understand enough about it by understanding whether the model itself is one of high confidence, by understanding the nature of data that went into maybe the modeling itself, the features that got extracted and were used. By maybe looking at user interfaces that explain how this AI might've made a decision for a hypothetical customer in a certain segment, for example. And I think building all of that into our applications user interface has been a key thing in being able to deploy that AI with confidence. And that's something I would urge everyone who is taking AI technologies to market to think about, where you want the humans to be able to collaborate with that AI. And in order for them to collaborate successfully, they need that AI to be explainable, intuitive, interpretable.

Jake Burns:
So in other words, the AI will give an answer, but it needs to explain how it got to that answer.

Vijay Chittoor:
To a large extent yes, or that answer should feel so intuitive that it feels right, and there should be enough evidence that without looking at millions of decisions, you should be able to be convinced that it's still doing the right things for the end customer.

Jake Burns:
So one of the things everybody wants to know is what does the future hold, right? I mean, nobody has a crystal ball, but if you were to guess two, three years into the future, how do you see AI and maybe more specifically generative AI shaping this in terms of customer experience?

Vijay Chittoor:
That's a great question. So, when you look at the first wave of generative AI over the last few months, we've been hearing so much about generative AI and everyone's talking about it's taking the whole world by storm. I think a lot of the initial use cases for Gen AI have been about creating content and more and more variations and being able to reduce the complexity, the time to create new content. And if you think about the world of customer experience, which you touched on, historically, for teams that are trying to deliver customer experience, there's been a big content bottleneck in order to produce the right content to personalize each and every interaction. If you think about millions of personalized interactions, how do you create millions of pieces of content? So in some ways, the first thing that Gen AI has done is to eliminate or at least reduce the bottleneck on content creation in many variations of the same content.

But where it's going to go next is really about combining that Gen AI with what we call a customer AI to deliver true personalization. So what customer AI would do is really predict what each individual wants or what variation of content might appeal to each and every individual. And what Gen AI could do is to actually create all of that content in real time or near real time and be able to make that available. So simple examples of that could be to take a promotion that a brand sends out through an SMS and the copy be different for each and every customer. Can that be informed not just by sentiment and things like that, which are easy for generative AI to manipulate, but also by customer AI, which really understands which offer the customer is interested in. And when you bring those two elements together, we feel that'll unlock the next level of the holy grail of personalization in some ways. So we are very excited about the future where Gen AI and customer AI come together to deliver great customer experiences.

Jake Burns:
So is there any advice that you would give out to folks who are out there, they're maybe in the beginning of their AI journey from someone who's been at it for quite some time. We have a lot of people who are starting right now. What is some of the top things that you would advise that they consider?

Vijay Chittoor:
If you think about it, there's a lot of latent potential in companies. There's a lot of latent knowledge, there's a lot of latent data which could be used to deliver value to end consumers. So how do you unlock the value of all of that? I think traditionally brands have started with human-driven efforts. They have then done some bit of digital transformation to use some technology to start unlocking the value.

But really, with AI, everyone in the enterprise is now enabled to actually deliver that transformational value to end customers. And if you start thinking from that lens, you start thinking about how you would redo your entire strategy, your processes, if you had to build everything AI first, some of this would need processes to be reinvented.

Jake Burns:
Yeah. So you mentioned something, I think you kind of alluded to the democratization of this technology, right? Giving it to the hands of all the people in your organization or more people in your organization. And also what I'm hearing is take the data you already have, which is probably going largely unused, and use this technology to unlock the insights in that data.

Vijay Chittoor:
That's exactly right. Yeah, that's exactly right. The democratization. Because this technology is becoming mature to a point where non-technical users will be able to use it. And as that happens at scale across many enterprises, it just unlocks tremendous enterprise value. So thinking about how that can be the biggest driver for your growth by getting this technology in the hands of customer facing teams into the hands of a lot of other non-technical functions in your company is going to be the key to scaling.