Author transcript
Hello and welcome. I'm Rob Wilkinson at CX Today and today we're taking a closer look at the future of AI powered customer service and why automation alone is no longer the finish line. If you're challenged with improving service efficiency without weakening the customer experience or if you're exploring AI as an opportunity to deliver faster, more thoughtful resolution. And stay with us because today I'm joined by Karthik Suri, chief product and corporate strategy officer at Sprinklr, who leads the company's product vision and strategy at a time when contact centers are under real pressure to deliver a lot more with a lot less. Welcome Karthik. Thanks for joining me.
Thank you for having me over, Rob. I'm I'm thrilled to be here.
It's really great to see you. Um, just for our audience's benefit, let's give them a little bit of context as to this conversation. So, what is changing in customer service right now that makes what you're calling the age of and feel so relevant, which I love, by the way, I have to say.
Yeah, I' I've been a fan of Jim Collins's book and the power of and is fantastic. Um I I mean let me take a step back and why you and I like and admire as we were chatting before the the age of or the power of and is that it allows us to bring together seemingly disparate things and discover a truth that is lying at the intersection of these things and that power is actually extraordinary right so that's what that's what I'm delighted about as we enter this new era which is the age of and in the AI world look on the consumer side the experience that uh people are are expecting or the demand that people have as of now is whether it's physical or it's emotional is that we meet them where they are right we meet them where they are whether it is their channel of communication or whether it is modality of communication or whether it is the approach to that particular communication some people want to get in and get out uh in terms of what they want out of their interaction with the brand others want that deep emotional connection with the particular brand or uh or or to whom that they are whose product or service that they are availing themselves of and and because of this things like containment have now have a negative connotation. Containment is a very inside out term right now. So that is one. The second is on the provider side or on the side of the brands right they they should demand right they demand and they should demand if they already do not that we negate trade-offs trade-offs I call Rob as the era of or right like is is it assistive or autonomous is it cost or quality is is innovation or safety those are fundamentally false choices so people are tending not to just buy software but they are tending to buy outcomes. They're tending to want connection. And for these outcomes, they are not just unidimensional anymore for companies, right? It doesn't mean just cost containment. It means cost containment with an extraordinarily high satisfaction rate that fosters loyalty. And that loyalty in turn becomes customer lifetime value. So bringing all of this together is an extraordinary important point. And the power of AI is that it actually enables us to do this if it's thoughtfully crafted.
It's it's a real it's like you're asking for the world, but it's only what customers want, right? Just know who I am, know the context, you know, give me the solution that I'm asking for.
Yeah.
It's been difficult for so long to actually achieve that.
You're absolutely right with AI now. the art of the possible is is significantly amazing compared to what it was.
So I guess um your your Sprinklr's view is that service leaders should no longer have to choose then between priorities such as automation and personalization or efficiency and customer experience. So where are those trade-offs showing up most painfully for the kind of organizations you're speaking to every week?
Absolutely. Uh uh let's unpack uh unpack that question because I love it. Right? Uh my mom's uh favorite phrase is that the worst interaction someone has with you defines the trust more than the best one ever will have. Right? If you apply that that's the that trust or the speed of trust as the as the underlying criteria there. The question becomes not whether AI can do a particular task or an activity but what degree of autonomy do we want to give AI for it to complete the task finish the resolution and leave the customer satisfied in this particular world. So not every AI outcome carries equal risk, right? an AI resolving some kind of a low stake FAQ or an address change or a booking or anything of that category or that zip code uh h has a very different profile than one that is handling a compliance request in a regulatory uh industry or or it is about about uh essentially how someone is in an emotionally charged state at that point in time. It may be actually a low-risk task if you think from a company perspective that but the person needs that human touch. So we bring these two dimensions together and then we determine what is that governance frontier because of the fact that not AI carries equal risk. So if you think about a a a a graph and a line right in the middle and let's just call that the governance frontier right in that continuum where can AI resolve autonomously and where is it assistive and where is it fully human like handoff is going to be the name of the game in terms of this and we have enough data enough customer context and enough tool to do this extraordinarily effectively But the ethos of wanting to do that, the ethos of not like you know thinking about containment as something I've taken care of and let's move on to the next one. But let me deepen my relationship with the customer. Let me deepen their relationship with the brand becomes the fundamental of designing this process or designing this particular system. So that is where that's what is going to separate winners from losers in how they treat and meet customers where they are in their particular journey. Be it compliance, be it an emotional need or be it one that needs a human touch there. At the same time, cost is a factor and therefore the more we automate uh and take give an autonomous authority to AI in the right side of the governance frontier, the more we get to reinvest in deepening that emotional connection between the consumers and the brands.
That's really clear that I I I love the way you you've kind of you've pictured that and I love the um the uh the front line analogy there that the governance front line. That's really really good. It helps to really understand the kind of ways that's impacting customers.
And honestly, Rob, these trade-offs don't have to just stop between like for example, as I said, like the degree of autonomy, right? When you think about like innovation and safety, right? People think about innovation and safety as a trade-off and and this is where we are moving customers and brands consultatively by thinking about it as a journey from gates to guard rails right from an afterthought to a forethought. So CX leaders who think about innovation and safety as a trade-off, they're not wrong, but they're living in a paradigm uh where they've observed these systems to be correct and they they within that operating model, it is legitimate, right? Like that model dictates that every AI initiative passes through five sequential reviews, legal, compliance, risk, security, privacy, uh all of them sign off on that. Each one has a queue, each queue has a clock. Now we are talking about 6 to 18 months from concept to production and that is a lifetime in the world of AI and the trade-offs are real. On the other hand, if you actually flip the process around on its head, right? And you say that safety is not a gate but it's a guardrail. It's a fundamental property of the platform itself. The same five checks happens but they are automated and they are parallelized and they pass into the build itself. the guardrails actually contain what can or cannot happen uh happen in in the a larger framework of things that get executed. Now you're moving from concept to production in days and you're containing the outcomes within the guardrails that you're providing. The guardrails can be at multiple levels. It could be at the platform level, could be at the system level, it could be at the prompt level and it could be at the handoff level. So the same enterprise rigor, different architecture, a shift left mindset and safety being a core feature or an attribute of the platform rather than a a a task that sits outside of it uh with the check boxes uh that that typically people are used to in the current paradigm. So that is where we need to move forward in terms of uh in terms of making AI an adaptive principle that actually you know that does not trade off between innovation and safety. So two examples of the power of the the the power of and in the new world versus the tyranny of or in the older world.
Yeah, I love that. I I and you're so right. the 18 month thing they can you if you look back 18 months to where where we were with AI and you think where we are now then why why would you set yourself up to fail I mean when you don't have to that's it's and it's such a yeah such a refreshing thought to be able to flip it on its head and build it in in that way. Um the thing that stands out to me is is basically what we're saying is that the the finish line now is isn't automation. It's the actual resolution of these these requirements. So why is it then that automation on its own is going to be the wrong measure of success? I guess how do we measure things now?
Yeah, that's a that's a great point, right? If you if you think about automation, right, I I believe that it is unidimensional and it is very inside out. Which customer has, you know, woken up in the morning and picked their favorite brand and said, I wish they automated more. Right? like that that's a that's a if you're talking fundamentally about customer experience management um I would rather have something that actually speaks to the customer and and creates that bond between the brand and the customer as much as possible because we all fundamentally believe that behind every brand is a human. So if you take that into consideration now and if you're a brand and you think about your customer you quickly realize that they give two hoots about whether you're automated or not. They want their issues resolved and better yet they want issues proactively identified and disposed off before they even realize that at that point in time you know uh at that point in time now you're building that deepening strength within the within the brand right now there they hate repeating context right I mean how many of us have had to repeat ourselves on something that we have just told uh you know uh in an interaction model with them not only has the has the has you know have have uh you know the companies forgotten the history of everything that I've dealt with them in the past including every product or service and they have all of this data but they've also forgotten the context of my previous interaction and they are not even able to assess the context of my current interaction right so that at that point in time I'm sure at the every previous touch point someone has closed that as a resolved issue is it resolved no it's contained it's deflected but you cannot clink champagne glasses because you've deflected it and your customer is still unable to get the value that they you know they derive out of your product and your service there. So you know at that point in time you're now thinking about resolution in a different dimension than containment or automation there. At that point you are like am I providing the value if not more of what the customer paid for in terms of the product or service they did you know that that that they consume from us there at that point in time resolution becomes your your your northstar and proactively seeking that and proactively whether it's an outreach or whether it's an autoresolution etc now goes from satisfaction to delight and that delight translates itself into loyalty. So you know yes uh I see a lot of specialists today you know advertising you know call containment rates etc around that but fundamentally you want to make sure that that it's a multi-dimensional approach absolutely cost management is an extraordinarily important our our you know uh you know customers see anywhere between 60 70% if not higher in terms of auto resolution autonomous resolution but those are for you know for capabilities that the customers want to be resolved, right? Like you know a booking change or an upgrade or a you know or a or a you know address change or an appointment scheduling in case of outbound etc around that. But fundamentally when they receive value they need to feel good whether it's from a a a a workflow or from a human and that's where like you know that's where the examples that speak to me the most u actually bring that emotion together that they feel heard they feel validated they feel satisfied and they want to deeply engage with us for furthermore.
Yes, it's a it's such a strong uh call to action to say take a step back, put yourself in the customer's shoes, have a look at the journey that they've been on and and would you accept that if it was you? That's what you know companies need to do more of. And I guess looking at it through from the customer's perspective, um how does that look on the flip side of this? So when we get good AI powered service, what does that feel like to the customer rather than what we're experiencing now? and how or what should change in terms of the experience when AI is doing its job properly.
Yeah. You know, you go to a neighborhood store and uh you know, I I done this all my life grow, you know, uh growing up through my childhood, right? You you have that deep intimate connection with your neighborhood shop. They know you. They know your parents. They know what you've bought before. They know what your favorite things are. You know, they call you by your first name. They they introduce you. they start speaking in in like you know dialects and slangs that you actually can relate to even more. That connection at scale is what we want deeply personalized contextualized so they know the context of my engagement with them from time equals zero right from the first point of engagement or better yet even from my discovery stage. Uh they they they have memory of of uh essentially who I am and what I do. they know preferences autonomously like you know just as second nature uh that that that has happened um the favorite candy that I like you know the the you know the the the sports good or the sports magazine that I always go after as soon as I enter the neighborhood store, my favorite fruit right you think about this together doing that at scale is what AI fundamentally enables us to do unrivaled data right so you know companies need to capture not just what people say to them but what people say about them right across the board. So they they they are in a position to understand deeply uh any form of signal that that they have deep deep customer context. My interaction can be physical, it can be digital, it can be voice. It has to understand where I'm coming from in terms of that industry. Uh they need to understand the language that I'm speaking. If I say this is mid, it doesn't mean it's average. It actually means it's poor. and if I'm a if I'm a you know Gen Z who's using that language etc. So it understands deeply what I'm not just explicitly communicating but implicitly communicating and at that point in time it also then helps me you know with the proactively uh you know as I'm starting to approach personalize that experience for me or after the fact if I have an issue with that it should be as seamless as possible in terms of like you know I put my old magazine take a new one because I had picked something up wrongly or whatever it is in a neighborhood store going back to that allegory needs to be able to do all of this. So that deeply hyperpersonalized experience with deeply contextualized information about me, not only what I've said to them, what all the subtle signals that I've sent implicitly explicitly about them elsewhere, all of that coming together for me in a proactive way is what makes me feel that emotional connection. That is what we want to do at trillions of data points of scale across you know thousands of companies with you know billions of uh uh consumers involved.
I I love that analogy the local store analogy. It's great. I won't ask you what your favorite sweets are today but maybe another time. [laughter] I think what I would like to ask you though is is we need to look at the technology side of all of this. So we know what the art of the possible is, but just kind of for our audience's benefit, talk about what a modern CCaaS platform uh needs to have in order to support them as CX leaders if they want that speed and quality without um you know the expense of one or the other. So it's not it doesn't have to be a trade-off. But how does that happen?
Yeah, that's a fantastic question, uh, Rob, and and one that that, uh, that I'm I'm deeply passionate about. U, first and foremost, if if I were to design this from scratch, I wouldn't even just limit to what what what is the attribute of a CCaaS system. It's an important attribute. What is the attribute of a customer experience system, right? All the way from discovery to purchase to engagement to care, right? So CX, good, bad or indifferent is associated with care and resolution etc. when there is a problem. I would love to have a system that is a end to end customer experience management system that can deeply engage uh with the customers through their life cycle and through any channel of choice. So let me you know that's the first part of the uh the prelude to the answer that I want to provide. The next big portion towards that is any uh any system that that that is used in the context of the question right there needs to have a few things that are that are extraordinarily important. It needs to have context first and foremost. Customers should feel like when they are when they are interacting with uh with a brand that they are actually deeply known and frankly we have all of this data about that particular customer. But we pretend we don't because it's a specialist solution that is doing one sliver of thing or it is not a holistic capability there. But it it the holistic nature with deep customer context is I would say the number one thing that the system should have number two is the ability to balance what is the right approach as I had mentioned for the degree of autonomy. The degree of autonomy should be a thoughtful planned uh uh approach uh both methodical and mathematical in terms of you know what gets done autonomously versus what gets handed off to a human. When it is handed off to a human or when it is being autonomously is resolved it should still feel human because we have human at least at one side of the brand. Right? which means that the design ethos, the the the conversational uh the conversational patterns, uh the brand value, the brand DNA, all of them should be beautifully integrated into that particular experience. My one of my favorite examples is one of our uh customers from the Middle East had an extraordinarily specific way that they wanted to communicate like with emojis with colloquialism with a slight sense of you know sense of humor etc. So making sure that in your autonomous like you know uh capabilities that brand value and that brand ethos and that communication style does not get lost and that persists when a human agent speaks to you. So you do not want this is robot this is agent uh kind of an approach to life in general right uh so that has to be a big part of it four is uh obviously cost right you need multifold ROI that should be crystal clear which means that you need to have the right autonomy but the right handoff and some and things that are in between so which means that the journey goes from autonomous to assistive to entirely human meeting the customers where they are in the journey on the infrastructural or the platform level. This then should have enough intelligence to take your conversational analytics provide quality management help you plan and engage your your human resources fairly effectively. So because again those people have to be deployed at extremely high value situations in the most effective way uh that one that actually embraces a community for collaboration etc. So if you think about connection, you think about community and you think about customer, you can bring all of this together beautifully in a contextual way that persists across to drive delightful experiences. And when your resolution is autonomous, it should also feel like magic. Like you know Arthur C. Clarke says that uh technology advanced enough is indistinguishable from magic. That is what I want uh the the people at the other side of the brand to feel and have access to human connection whenever whenever it is necessary.
Yeah, I I love the Arthur C. Clarke uh comment. I also love the blank piece of paper going back to like a if you were and this is actually something we could challenge our audience to do. how just take a step back, start from scratch, pretend you don't have anything, get that blank piece of paper and design how it should look for your customers and then compare that to actually what you've currently got going on. And I think there'll be quite a difference uh if if people are being honest. Um, and if they do go down that route and they look at um, like moving towards a more suitable uh, kind of a more comprehensive option, um, there's some things that they're going to need to consider because um, there are there's then the chance that when you've got AI and human agents working together so closely that actually a customer journey can become quite complex or emotional or it's like a high value customer, people are going to be worried about what happens then. So can you give them any advice around how you how you kind of protect that that those scenarios?
Yeah, absolutely. Um I also wanted to add as I do that to the fact that I also want to make sure that people do not think a human interaction is a defect in a process of autonomy. Right.
Yeah.
I myself am proud in reporting the the percentage of autonomous resolution as I just mentioned 76% to you for one of my customers there 90% in my best-in-class etc. But I want to be crystal clear that a human interaction in that particular even in those scenarios are actually not defects. They're actually we're giving more time and care and and and handling time to the humans to do a better job in connecting with other humans through the process. Right? So in all of this the underlying ethos is that AI helps humans be more humans with respect to that. I will give you one example right in a world of autonomous connection and robot-based resolution. one of our co-pilots, right? Our co-pilots took the customer uh co-pilot is an assistant AI technology that helps a human agent do their job better. In this particular case, an A/B test showed that beyond an autonomous resolution, a co-pilot when handed off to a human um had a 30 percentage point difference in the human uh advising the customer on the next best product to buy. So, not only are you delivering auton autonomous resolution, you're delivering the right human handoff at the right time, but through that right human handoff, you're converting what was a contact into a, you know, a into a loyalty moment, right? A 30 percentage point increase in conver in in conversion after a satisfaction satisfied resolution, you know, deepens the relationship with the customer with their next best product to buy. That is how growth flywheels are spun right. So that becomes the that becomes the mindset tool set and skill set that we have to collectively adopt towards. So coming back to your question first uh unshackle yourself from this tyranny of or uh the cost versus quality the innovation versus uh versus uh safety uh you know the speed uh you know versus you know uh uh speed versus you know deliberately uh handling uh at the right time. Autonomy versus uh assistive nature all of these are false choices. you can actually find sweet spot leveraging the power of and that that actually brings you to that magic the magic is in the middle in finding that sweet spot right there. So uh you know first of all from a uh from a design philosophy basis unshackle yourselves from the tyranny of or there number one. Number two, get understand that context is extremely important and persisting memory and knowing that customer deeply whether it is systemically or at a human level is going to be extraordinarily critical for them to feel valued, validated and heard because as I've mentioned multiple times behind every brand is that human that wants to be validated, valued and heard is number two. Number three is set your governance frontier. You can have extraordinarily good results with the right governance frontier approach. Things that can that you know what degree of autonomy that you want to give to to AI and where is the right moment either either from a compliance or regulatory perspective or from a brand impact perspective or a basic human connection perspective that needs that human touch there. If you designed all of these together with a holistic system that also is able to analyze and provide you continuous improvement recursive learning capabilities, have a learning loop that actually makes the next interaction better than the previous one with the right decision traces that brings the power of human and the power of AI together, then you have a winner.
True continuous improvement. That sounds like uh holy grail kind of stuff. Um, [laughter]
we could talk about the new term is recursive self-improvement with the right dog.
Yes, I think um I think we could talk all day about this. I there's some of the numbers that you've shared there around performance impact are are awesome. I'd love to be able to dig down on those a little more, but I think we are struggling for time, so maybe people are just going to have to read a little bit about those. Um, but let's before we go just look ahead a little bit. Um so for anyone who's watching this who's considering making some sort of moves in this direction what what's one kind of simple practical step that a a service leader can take uh you know in the next quarter if they want to move beyond this thinking automation only and start thinking about the resolution side of things.
No absolutely first and foremost I would love uh for for service leaders who are thinking about that more than happy to have a conversation with respect to that. Um I would as I'd mentioned I would I would say that like you know you know pick a system that actually busts the tyranny of or ask for outcomes right think about how you what experience like you know from a blank sheet of paper what experience would delight your customers drive that loyalty drive that lifetime value for them and make them both satisfied and happy at any given point in time but you're you know managed for also cost containment. Uh so from those two perspectives you know do not accept one versus the other which means specify your outcomes. Start with the outcomes that you're solving for and work backwards into what works into that. And once you get there then you know deep contextual understanding and an ability to take action from the insights that you're getting from those from those plethora of signals is going to be the third big portion that you have to think through on how you design that particular system. And fourth and I would always encourage this is to ensure that you have a governance model and a governance frontier that actually helps you assess what is the degree of autonomy and it it's not a static measure right as I said every decision makes the next one better so from that particular perspective what is the dialup dial down that you want to have in such a way that your governance is built into the system and not extraneous to it and pick again pick a vendor that is in a in a that is partner to you in this particular journey, but who is also ready to, you know, share uh share the benefits of your outcomes in in a way that they are, you know, they are incentivized to deliver outcomes for you, not just a a fake measure or a vanity metric uh that they are committing to.
Yeah, that that's so important. I think um yeah, there too many of those
are the top five that come to my mind. Rob, I'm sure you have a lot of thoughts and you're an expert in the space. You know, I love uh listening to you. Uh therefore, I'm I'm pretty certain that that uh that you you know, your audience uh uh you know, I would love to listen from them on what are the other things that we should be doing.
Well, I think I think we'll be able to get to that. I think um I was going to go into a big long summary as to kind of this conversation, but I actually I think it it's very straightforward to sum up for our audience and and it's very simply put. It's it's the age of and and watch out for the tyranny of or. I don't think we need to say anymore. That that that covers it and I and I love it. Um Karthik for anyone watching who who wants to dig deeper on this topic uh or maybe to speak to the the Sprinklr team. What what's the best next step for them?
Oh uh please uh in invite invite you to sprinklr.com u check out what a holistic system looks like what the power of AI looks like what uh you know unrivaled data uh in deep customer context and a system that takes you from insights to action looks like um and I'm more than happy personally to have a conversation with you as you're looking through all of this you're welcome also to connect with me on LinkedIn or or with Sprinklr on LinkedIn uh and we are more than happy uh to continue those conversations in in any of those modalities
That's fantastic. Thanks. And of course, you can discover more articles, resources, and other videos just like this one at cxtoday.com. Uh, but for now, that is all we've got time for. Uh, thanks again to Karthik at Sprinklr for sharing all these thoughts. Um, I'm Rob Wilkinson, and thanks very much for watching.
Fantastic talking to you, Rob. Take care. Good to see you.
