Fireside Chat: The AI Symphony: How Analytics, Generative, and Agentic AI Work Together Drive Growth

Welcome, Jay.

Pleasure to be here, Garad. Looking forward to it.

Where are you today?

I'm based out of Washington, D.C. area.

Well, we're very close, very close neighbors. I can almost see you.

It will be a pleasure to be your neighbor too. So, all good?

All good. What is your favorite restaurant in Washington?

Ah, there's so many. But I absolutely love the brunch at Founding Farmers. We have a few options to go to, but we usually visit the one by McLean, which is closer to closest to us. Definitely one of my best places to hang out.

It is one of the top 10 listed restaurants in the Washington D.C. area. In Founding Farmers has probably about what, 15 or 20 restaurants on the East Coast. What I love about it is it's all fresh.

Yeah. Farm to table. Yep, exactly. So.

So the last time we talked about AI, we talked about so many things and I what I found fascinating is your terminology of comparing this to a symphony. So what, what does an AI symphony mean to you and how does it apply to selling? How should we think about that? Or sell us?

Yeah, absolutely. So, Gerard, what I really mean by that is obviously AI has been around for a long, long time. I know we talk about it day in, day out today, but if you think about it, it has been around for a little over 15 years in context of commercial adoption of AI. And obviously AI has many flavors and very distinct applications across many use cases. Right. So this notion of symphony is really important, especially at this point in time when we are talking about AI so much. There's a lot of noise, there's a lot of great innovation coming at an unprecedented pace at us. Really, if you take a step back for a second and think about it, most organizations over many years have obviously built some AI and machine learning capabilities to drive those insights, especially in commercial context, specifically for sellers and, you know, things like what's my sales forecast going to be, what's my score, how do I prioritize my jobs, tasks in a day, things of that nature. Right. So those insights have been around for quite some time. I do see a lot of app organizations having actually mastered that to some extent over time just because of the machine learning capabilities. Right. And now if you kind of fast forward it a little bit over the last two to three years, more specifically, we're seeing a lot more of generative AI capabilities, which has a amazing power of consuming a lot of enterprise data, both structured, unstructured, and put that in a very precise summarized way that could be applied in many use cases. Right. So really being able to take the contextual and conversational nature of this data and then make it applicable in various conversations, various situations makes it really compelling. If you fast forward further to today, we are talking all about agent AI. We are shifting from tell me how to do it, how do I do this task to do it for me. Which basically means you are relying on AI to do a few tasks for you. The idea here, when I say isymphony is really about bringing all of these different forms of AI and putting it in context of customer journey and making it really work for a seller. You can pick one use case, be really passionate about Agent Ki, but if it is really adding a very small incremental value for a seller in a day, it is not going to really move the needle for you from top line. So I think this notion of AI symphony is really going to bring all these forms of AI and really amplify AI can do for sellers in commercial context and also for other departments more broadly.

Yeah, so we, we were moving sort of from ChatGPT to the age of the AI agent and then next year there's going to be something totally different. And of course orchestration is right now a critical function. I want to go back to real basics. What are right now that you see the key forces of change?

Yeah, great question, Gerard. Obviously there are many, right. We are seeing a lot of convergence of factors like business urgency, the pace at which we are seeing innovation with models really improving each day. Right. We talked about ChatGPT2 and now we're ChatGPT5 and every model really kind of really taking that a notch higher each day. And then the access to how much of enterprise data and also outside an enterprise, there's access to all this data in the form of social media for B2C scenarios, even for B2B in many contexts. How are you able to really take all of this data and process it and apply it to a use case has become a lot more accessible. And then lastly, this is obviously not news to any one of us. There's tremendous amount of excitement across the organization, starting all the way from the boards to the C suite execs and all the way down. Right. Everybody's intrigued. How do I make this part of my life? There are obviously other sides of it where people are thinking about is it going to take away my job? So clearly a lot of excitement, interest which, which, which brings it as a absolute need for like force that's changing how people are thinking and forcing them to move in a certain direction. Right. But if you were to Just like play it back. Right? The business urgency specifically I see a lot of leaders kind of really talking about like hey, we have to get on this board train of like adopting AI, otherwise we are missing out. So clearly there's a fear of missing out. And when you put that in perspective, some organizations have already started seeing some top line growth, be it the productivity numbers, but also in terms of the revenue impact it has. So it comes in the form of really sales teams kind of being productive by using some of this agent Ki capabilities or generative. Alex, I want to send out an email. I want to, I can draft a quick email. So it's a matter of urgency to obviously act now and not miss out from your competitors by not adopting and waiting for the right data to be set up and things of that nature. Right. And I kind of talked about like how this, if you put it in context of like also the models and the innovation, I think the moment is right now. Right. You have the right technical capability to, to really get on that journey and then a lot of times you have great ideas, great innovation, but you don't have the right excitement for decision makers to act on it. So there's clearly excitement there. So it puts a lot of momentum behind because of all these different forces kind of coming together, which I feel is making it a very interesting journey for all of us, especially in the enterprise space.

Yeah, it's hard to put all those pieces of the puzzle together so it adds up to a picture that we can use as a playbook. So what do you see as the most profitable and promising usage scenarios of AI in sales? And I'm also reminded that you wrote an article in Harvard Business Review. Can you tell us about that?

Yeah, sure. Like I said, the adoption application is going to be across the board. That's why when organizations talk about, hey, how do I think about this? The recommendation is always to look at the whole domain as a business function of the organization and look for all opportunities across the board. To your point on the commercial and customer journey point of view, the way I would like to think about it is the front office, middle office and the back office. Right. Every seller kind of in some way or the other engages across their actual selling processes to engage with your customers. And then you want to go to your mid office or middle office to take and onboard your customers or send a code, think of sales ops and insights, things of that nature and then your back office to actually drive the consumption, retention, customer support, all of these different capabilities that come to life. So if you look at some of the most applicable usages that we are seeing as of today, especially at the top of the funnel, right. We look at a lot of AI SDR being one of the key use case that is being applied. It is interesting. A lot of customers have capacity constraints with their SDRs, right? So you tend to go after just your leads that would come in the top of your queue, go after them, and then there's a huge backlog of leads that go untouched. So if you think about it, there's a real opportunity that you can drive by unlocking that capacity in the form of agents, kind of doing that outreach for you and helping you convert more and move those leads across the funnel. This is where Gerard, I want to touch a little bit upon the angle of Symphony and how you bring all forms of AI back in this use case of sdr. More specifically, you cannot just go and do outreach on all leads. It's not the most effective strategy. Ideally, you want to score your leads using your predictive, good old predictive capabilities and maybe even leave your top high quality score leads for the human reps to go. As organizations are experimenting, learning more about deploying these agents, you want to start with your core leads, help nurture those leads, track your conversion, and then over time use more and more of the generative capabilities of drafting that outreach email and personalizing it, and then have agents kind of do that outreach for you. Right. So these are the kind of use cases where you have to not just think about agent AI. Though the article I wrote was primarily focused on Agent Guy. I wanted to think, just if you think about it in this context, it's really. That's where the magic is happening, where all these are coming together. Right. And then to your other point across the journey, the SDRS is one of the use cases we are seeing a lot of. There's obviously use cases around sales enablement coaching. AI has the capability to process a lot of information, understand the context of the seller in a particular deal, and propose recommendations on improving the pitch. Right. We're obviously also seeing a lot of productivity hacks for reps, which is in the form of, hey, I need to send out this email. Can you like draft an email? Right. There are SS and copilots of the world which are already set up for reps to get benefits from that. There are also a lot of applications. When you think of about all the proposals and quotes that you have to generate, there's a ton of content generation there. So there's that happening. And then obviously from Back office, there's a lot of potential from capacity, freeing up capacity and generating documents and offering pricing recommendations, making sure the margins. So all the insights and coupled with your agent capabilities just amplifies your ability to have an impact across the journey. Obviously there's ton of other applications Gerard, but I think that's how I would think about it. And most of the details we would offer in the article were also kind of centered around the front, middle and the back office.

Well, obviously there's a lot of promise, there's a lot of potential there a few sales organizations that score amazing victories with AI. However, when we zoom back, there's something that stands in the way from sales leaders fully embracing AI. And I want you to speak to that and explain what is the resistance and also explain how we can overcome that.

Yeah, very, very relevant question. I think literally everybody, it's top of mind for most leaders to see. I think there's three really big things to talk about Gerard, in this context. The first one is, I mean we talked about the momentum and excitement on this topic obviously which what it resulted in is kind of fail fast approach, which is always a recommended one to like let's go just deploy some AI capabilities and see where it goes. Let's learn from it, which is great. Right? That's always a great idea to go for it. But what happened over the last year and a half is a lot of organizations embarked, emerged on that journey and now are kind of struggling in the pilot purrogatory phase where those AI investments are not really scaling which basically is resulting in them not really unlocking any value that they originally anticipated. Right. So that's one big bucket of challenge that we are seeing each day with most organizations. And then there's a real challenge around this tremendous amount of noise. Right. If you think about every day, if I were to just go on the Internet and try to learn about AI and what it can do for my business, if I were to be a CEO or whoever, there is just so much of information which makes it very difficult for leaders today to understand what is the real potential I can capture today. Right. There is obviously a lot of forward looking vision of what AI could do and how much it can automate and change our lives, disrupt our lives, which is broadly true. But at the same time, how much of it can technology do and achieve today is still a bit of question that has to be fully understood by each use case. Right. Then the third one more from the tech enablement, we're seeing a lot of leaders really stuck in this platform choice, there's ton of vendors out there, obviously a great time to be offering AI solutions. At the same time this also relates back to the pilot. If I am experimenting quick I would ideally want to start small, pick something that's not big to build. But at the same time all of these different point solutions are not going to scale. A lot of CIOs and tech leaders are currently grappling with this question and idea around how do I make a scalable platform choice that will really help me make and move the needle across the enterprise and not just by business domain, but give the business leaders enough acceleration by adopting AI, Right. So I think that those three are the big buckets and just to double click on some of those. So especially on the pilots. We saw a article that came out a few weeks ago, I think from MIT, roughly 4%, 4 to 8% are the ones which are actually capturing value. So if you think about is a very low percentage. I mean tech investments traditionally had this challenge of not fully unlocking value be it the CRMs of the world and even the ERPs and even more specific domain specific applications had a traditional challenge. But especially what makes it unique with AI especially is how do I kind of I know this use case worked, the experiment didn't fail, but I'm still not going to get value out of it tomorrow or day after. So how do I kind of change the dynamic or sequencing here to make real impact to my business? And if you think about what is really causing this, the underlying reason is a use case is never going to move the needle. It is really about how is my business going to change my business process or workflows and tasks can be automated through agent. AI needs to be looked at holistic. You've got to really assess the impact each one of these cases can make and have a clear roadmap of what am I going to work towards. Right. So really thinking about scaling starts when you actually pick a use case, when you actually evaluate what's the impact it has and not after you've already done your pilot. Right. What you can learn from pilot is the technical feasibility of it, how are the customers and how do you measure it right. Is there real conversion or was it more of a outreach but not really moving the needle. So I think really making that, tracking that, identifying the right use cases, putting it all together and having a plan before experimenting is a big unlock there. Also I think the agent AIs technology Gerard, I'm sure you had a lot of great conversations over the last Couple of days, set out all the different solutions that are out there which work well, really well. Right. But if you think about it, it is not an answer for every single problem to tackle today. Right. And the reason I say that is if you. I mean sales is a very complex psychology. It's a complex art to have. Not everything is science about sales. We all know that and that makes it very unique. Right. Human conversations are extremely complex and and nudging a customer towards a sales motion and towards closing a deal is not as easy as giving all the information and. Right. Metrics in front of them. Right. So there is a little bit of style, personality and the ability to handle more complex conversations. Still very much. AI agents are not there. Right. Especially when there's a lot of context and multi turn conversations that are happening. We are seeing a lot of the AI agents kind of underperforming when you measure them against human baselines. Right. So I think the idea here is to absolutely pick the use cases which are right and ripe for agent AI and really action on them versus going after all the complex use cases first and then thinking that oh, agentiki is not ready and kind of thinking it's a failed experiment. So picking the right use case is extremely important. And just lastly on that, the organizations and the platform choice I was talking about also goes to the buy versus build framework that usually IT organizations had. And typically the idea would be to go with a buy or some kind of a platform which is low code and it really gives you the time to value and speed to market with delivering your solutions for your business stakeholders. However, Agentica is not very unique. But the challenge is a lot of these platforms are also evolving and developing. They have a great promise to offer in terms of what can be possible and how the roadmap they have planned for each of these big platforms and big vendors that we often hear about. But what we are seeing is there's still kind of, it's a bit of work in progress and we talked about how fast AI is evolving and innovating which makes it very difficult choice for leaders in it, especially to think about the scale and picking a platform when it is still growing and evolving. So I do see that as a big challenge a lot of leaders are facing. There's plenty more. But obviously absolutely every conversation I have end up going in one way or the other. I've done a lot of pilots not getting value. Hey, how do I pick my platform choice? That way I can actually scale all my use cases and obviously making sure you have a Right. Structure in place from the systems and tools point of view too.

Just got an audience question. Gene wants to know which tools does McKinsey use to multiply their consultants time?

So we use productivity tools. We have a homegrown tool called Lilly that we use. There's plenty of press and articles out there, so would strongly encourage you to read again. I think there's ton of options to kind of take your own capability and then build agents off of it, making sure productivity can come in different forms depending on your role and depending on the things you do. But we do a lot of that agent AI and generative AI agents that we've already built homegrown on our side.

But so I have a confession to make. Before we did this event, I went online and I said, who are the, the top consultants in gentic AI? And, and I got, of course, you know the list, you know your competition. And then I ask what are their philosophies in terms of implementation? So we can do this right? How do you hire the right consultant? But what I found is that the advice was all over the place. There was no consistent vision that the big consulting companies share on how to do this right. So I want to know what is Jay's golden standard for doing this right?

Yeah, I mean, again, I would give it more from my personal experience here, more so than speaking for other organizations or ours for that matter. And Gerard, to your point, there's no one size that fits. All right. And I would also caveat saying that we are all learning how to do this right. I think that's important to recognize that we are in an ever changing world of AI where it's very rapid, very much unlike what we've seen before in terms of the pace of innovation. But if you take a step back and kind of think about it. So to your point, should we go big, small? It really goes back to what do we envision in AI? What role do we envision AI to play in an organization? And it is typically a horizontal capability that applies to all the different business functions, specifically to one domain, different capabilities within the domain and so on, so forth. So it is really very much a horizontal. It's not of some people kind of think about it as like, hey, I got to just put a dot across my end to end journeys. There is like moments where I would fit in. But I think there's a real need to take a step back and think how is AI going to be a muscle for an organization and a horizontal capability? Right. And having a clear vision about what that means and how do I. And it goes back to like a business function, a business capability, and you have tons of workflows and tasks and processes. And there's a intentional need here that organization should take to kind of take rethink and reimagine. And we keep talking about sales teams and commercials, commercial functions specifically. So I think there's a need for us to think about is this how I would reach out to the customer or should we really deploy a human capacity to do this task versus an agent? Right. So that rethinking is going to force us to get creative about deploying agent, AI and other AI capabilities. So I think that's basically the big thinking angle there. And every organization is going to be forced to go down this path of like, hey, what is this going to do to my business? What is this going to do to my business function or department? And when it comes to execution, we talked about the experimentation and learning being extremely important as you scale this, both from the risk point of view, also from the angle of learning. To actually deploy this at scale when you are in the process of scaling, you should absolutely have a strong measurement framework to assess the impact. You fine tune the agents. The first time you actually deploy something, you're probably not going to hit it right with the tonality of the agent. A lot of this is conversational. That's something for us to be very mindful of and it's very important. Just automating workflows is not the end goal. But how do we actually have a conversation which is going to make someone feel like clicking on it, which drives the conversion rate, which provokes your customer to ask follow up questions. So to be able to do that, I think there's a real need to fine tune that over time and that makes it really interesting. The last thing I would also say is on one side we are saying to your point, go big, do all of these things, rethink and then start small experiment. And I think that's where, Gerard, I feel like this has to be a bit more top down and bottom up. Right. The top down angle is you should have a clear vision of where you're going. What do you think you can do with your technology today and kind of build off it that way? And bottom up is learn as a organization, how do we deploy this, how do we bring this into our workflows lives and democratize the ideas that each of your sales teams, be it SDRs, reps, account managers, come up with ideas on how things can be actually improved. In their life by deploying AI. Right. So really bringing them together is where I see it's not a framework. I think fundamentally that is the only way organizations can actually move fast, learn from the bottom and then have a clear vision as to how to sequence and scale them across the organization.

That brings us to the beginning, which is orchestration between the top and the bottom.

That's absolutely right. Yep.

Well, thank you so much, Jay. Where can people learn more about your work, your papers.

Happy to share my LinkedIn and feel free to engage or LinkedIn. Always a pleasure to speak to people and share more ideas and also learn from everyone.

Yeah, I, I, you know, what I'm taking away from my conversations with you is that sales is an art and it is a science. And as science like AI increases the potential for faster, better, cheaper results, we need to improve our artistic talent, our capacity to shape the future in so it corresponds to our hopes and dreams. So as we become more technologized, we need to become more human and, and become more creative so we can imagine a much better future.

Really well said, Gerard. And that's why I would love to be your neighbor. That way I could learn from you every day. But you said it really well and I think the art is going to come with time. We don't know but to your point of science is where I think we've been focusing to kind of automate use our AI capability. But really well said. Fully agree with that.

Thank you, Jay. We have a nine minute break and we are going to come back with Abe Avashti who is a partner at Deloitte. It's going to be fascinating. Jay, I think you, you want to watch your competition.

Pleasure, pleasure having me here. Thanks Sara.