Speaker 2: You’re listening to a new episode of The Brave Technologist, and this one features Andy Maskin, who is a director of AI Creative Technology at Publicis Sapient. He’s a Generative AI SME in both theory and practice with a current focus on genic marketing applications across his career. He’s focused on the way technology is impacting customer trends, led and built innovation labs, and his thought leadership has been published in major trade publications.

In this episode, we discussed. Brand visibility in AI as the metric to watch this year. Why they’re no longer asking, ‘do we rank on Google?’, but have shifted to, ‘does our brand show up when we use Chat GPT?’. A breakdown of what the MCP model context protocol is in plain English and how it enables AI agents to talk to each other using natural language and why genic AI fails without clean data and why companies need to avoid plugging in AI into unmotorized systems.

And now for this week’s episode of the Brave Technologist.

[00:01:00]

Luke: Andy, welcome to The Brave Technologist. How you doing?

Andy: I am good. How are you?

Luke: Good, good. Thanks for making the time to come here. We’re here at the AI Summit When people hear you speak at your talk today.

What’s the big idea you wanna land? So, the panel I’m on is about AI and creativity. And I think one of the key things that I deeply believe in. And a lot of people at publicists in general believe in is that AI will never replace great work, the role of humans and great work. Yes.

Andy: There’s gonna be a lot of automation. Yes, there’s gonna be a lot of disruption. Yes. We have to disrupt ourselves and learn new skills and think about the way creativity is executed, but the human spark of imagination is gonna still gonna be critical.

when you have this sort of sea of sameness of a lot of ad and marketing communications being generated by ai, you need a human in the loop to really take things to that next level.

To have that flash of human insight to tap into [00:02:00] the human lived experience that elevates. Work from good enough to Great.

Luke: Yeah.

Andy: And I think that’s critical and will continue to be critical even if the path of how you get there is gonna change dramatically.

Luke: Yeah. And it seems like it definitely is changing.

It seems like advertising is kind of that area where it’s really becoming a test bed for a lot of the generative video content, creativity and all of that too. you know, you’ve been at this intersection of technology and creativity and consumer behavior for many years, what’s the biggest shift you’re. Seeing and how people expect brands to show up in an AI powered world.

Andy: it’s interesting. I don’t think that the general population, the sort of average consumer cares a ton about what’s going on with AI in terms of. Because the end assets that they see are still videos, right? And they’re still images and there’s, it’s the same assets that they’re used to. Seeing how those assets produced is a little bit different.

So in terms of brands, I don’t think it impacts capital B [00:03:00] branding. As much as functional experiences. And I can, I can dive into that.

Luke: Yeah, yeah, please.

Andy: So people are experiencing chat bots, Gemini, and threat and so forth. So I think. Brands that have a chat feature on their website or mobile app or some way to chat with the brand are basically gonna have to do one of three things.

And one of them is to really up their game with their chatbot, where the chat bot is agentic, it can actually understand and solve. A consumer’s problem or really answer their questions with depth and genetically act to resolve their issue. ‘Cause one of the worst things, and we’ve all experienced this, is you go through a chat bot and you explain your problem at the end.

It’s a great, here’s an 800 number to call, and now you’re gonna be talking to the voice robot now. Right? Right. So you really need to up your game in terms of what your chat bot can do. If you can’t do that, then my personal view is that you need to then just replace the back end with [00:04:00] live. Agents,

Luke: right?

Andy: Altogether.

Luke: Yeah. Yeah.

Andy: Right. Yeah, you can, you can actually resolve issues and answer questions with nuance. Or the third option is just pull the chat feature from your app and your webpage.

Luke: Yeah. Yeah.

Andy: Because what you don’t want to do is lead with an experience that feels so much less satisfying than what people are now accustomed to in terms of.

Specifically text chat.

Speaker 3: The Brave Technologist is brought to you by the Brave Search, API access billions of indexed web results from a simple API call with the Brave search API join the leading names in AI and tech using the Brave search, API to power genic search.

Keep LLMs current with real time data, train foundational models, and bring the best of the web directly to the leading edge. Get started today at brave.com/api.

Luke: I mean, that seems like a huge opportunity too, for brands, right? Yeah. To get that feedback to, to iterate, to kind of leverage AI on the backend for, you know, iteration improvements, tailoring things. I mean, and this is kind of interesting too, when you think about like [00:05:00] how advertising has really gotten better and better around targeting certain cohorts, right?

do you see like now that content generation and AI and the experiential parts of AI are becoming in a way more to. Do you see some kind of merging between the two where maybe content’s created based off of a targeting cohort or, or something like that? I’m just kind of curious for your take on that.

Andy: Well, yeah. I mean, you always want to personalize an ad for the person who will be receiving that ad. Yeah. As much as you can. That’s kind of the ideal. A lot of that though, predicated on, you know, underlying technologies that have nothing to do with generative ai. Right, right,

Luke: right.

Andy: Has to do with. What kind of data you have, what kind of data are you allowed to have?

Right? What’s the responsible way to be managing that data? Right? And none of that has to do with the language models.

Luke: Right?

Andy: Right, right, right. It’s the, the, it’s a whole different that comes before the language models even get involved. So I don’t know if, if consumers expect the marketing messages necessarily to be tailored to them, although it’s in the brand’s benefit that they are.

But I think [00:06:00] support and there’s a relationship there. So this is what I was getting at with the chatbots. Yeah. Yeah. There’s an. Existing relationship between a brand and a consumer. The interaction should lend itself to the fact that there is already a relationship there. You did call last week,

Luke: there’s a history.

Andy: There’s a history, and the bot or the interaction point knows that history. Yeah.

Luke: Yeah.

Andy: So you cut through having to repeat yourself five times. First to the robot, and then to a person, and then to another person and so forth.

Luke: Yeah.

Andy: There’s, to your point, there’s enormous opportunity for much, much better consumer experiences.

Of brands, but you need to actually deliver a really good experience.

Luke: Right?

Andy: Right. You need to deliver, there’s an opportunity for it. Yeah. And you can seize that opportunity. Yeah. Or not, but if what you can’t do is stand pattern, right. With your existing meh chat bot. Right? Yeah. You need to deliver a really good customer experience that’s optimized, that’s agent that can resolve their billing problem by going into the billing [00:07:00] system and resolving their billing problem right there before having to transfer you to the.

Or six different people. That’s a pleasant experience. Yeah. And that’s the experience of the future. And I think it’s up to different brands to decide the path of how they wanna achieve that and get there.

Luke: Yeah, yeah. No, that’s, that’s fantastic. when you kind of look back on early generative ai, like wow moments, you know, what’s changed now that teams are moving from clever prototypes to real production workflows, we’ve heard, and

Andy: it’s interesting ‘cause I love me.

Clever prototype. Mm-hmm. God knows I have in the last three years personally created any number of clever prototypes that I love and show a lot of promise and that a lot of people can get excited about. But at the same time, I’ll be the first to admit that clever prototype is not a transformation made.

Luke: Right. Right, right, right.

Andy: And I think the nuance is when you wanna operationalize that clever prototype. [00:08:00] Deliver from your sort of dummy data to real data.

Luke: Yeah.

Andy: You have to fall back on systems modernization and systems integration. Right. Which you can use AI tools to help achieve. But it’s not the shiny super, you know, solve all your problem solution that people want to rush to.

Luke: Right,

Andy: right,

Luke: right.

Andy: And so I think the, the nuance is people wanna rush into these big agentic deployments have so much promise before they metaphorically pay off their technical debt,

Luke: right?

Andy: So if you have technical debt, you data debt, you have just issues in your tech stack that you haven’t been addressing, if you wanna leapfrog that and brush it under the carpet and try and do agentic AI without solving those first, those are the projects that tend not to work.

Out. Yeah. As scalable, real, ROI drivers. Right. And so it’s, it is an interesting dynamic where people see the clever [00:09:00] prototype and they want to operationalize it, but first you have to get the basics right. Right,

Luke: right.

Andy: And that may involve addressing technical debt that you haven’t been looking at for years and years.

Luke: Yeah.

Andy: But now the bills come due and you’ve got to deal with it if you want to move forward in an AI world.

Luke: Do you see AI is helping to kind of resolve some of that technical debt? Like be a, a, a tool in that process? Or, or

Andy: AI can be enormously helpful in addressing that issue. So at equivalent of Sapian, we have a tool called Slingshot. Which is an enterprise grade product to help address this exact problem.

Luke: Awesome.

Andy: Of going through your legacy stuff and wrap. Modernizing getting you ready for systems integration, getting you up to that sort of basic level where you can then launch into ambitious agentic ai undertakers. Okay. Right.

Luke: Yeah. Yeah.

Andy: So because our business has been transformation for decades.

Luke: Yeah.

Andy: And so [00:10:00] now that we have these AI power tools, we’re able to turbocharge that in a way that can get enterprises ready for the future much faster and with much higher quality code than you could do in a comparable amount of time or budget in the past.

Luke: That’s awesome. Yeah. Yeah. That’s cool. I, I love it that you guys are working within, working with the tools within your system already as part of this too.

the other side of that too, I mean. Around metrics, right? and, you know, how should creative teams start to rethink metrics around ai, and what’s one metric do you think will matter in the coming year that doesn’t necessarily matter that much today?

Andy: I kinda think for creative teams,

Luke: yeah.

Andy: I tend to zoom out a bit in thinking about this. At the end of the day, the metric is business results.

Luke: Uhhuh,

Andy: it’s moving product off the shelf,

Luke: right?

Andy: It’s selling your services. It’s ROI.

Luke: Yeah.

Andy: Like build a company, bring in more money than it’s spending.

And how does it return value to its [00:11:00] shareholders? That’s the ultimate metric that everything is judged on. Mm-hmm. I think, you know, in terms of creative output, again, the assets that get produced are the same. So it’s the extent that there’s better creative that’s driving better business results.

That is the metric. But I don’t think that’s a different metric than we’ve had time in Memorial.

Luke: Yeah.

Andy: I will say one metro that is going to evolve and become increasingly interesting over the coming year is visibility of brands to chat bots. Yeah. Right? Yeah. So some people call this a EO, some people call this GEO.

Yeah, yeah. Whatever it is. The point is we’re SEO is about showing up in search. The question is how does show up in chat?

Luke: Yeah.

Andy: And we’ve been spending, like, my team’s been spending a lot of time thinking about this. Many, many, many other marketing organizations are spending a lot of time thinking about this.

‘cause people are not clicking on those links in the chat answers.

Luke: I mean, even conservatively, you know, web traffic’s down 20% because of these AI summarizers and things like [00:12:00] that too. I mean, what’s the current school of thought thought process on that? Like where’s your head at, you know, currently with how brands will play into.

These new interfaces and in this new way of you know, querying and, and doing all this stuff.

Andy: Well, it, it’s interesting because I passionately believe that this is not a solved problem.

Luke: Yeah.

Andy: That we don’t know yet.

Luke: Yeah.

Andy: Right. Any, and, and in my personal opinion, anyone who says we’ve totally nailed how to take over an answer for a given question is not necessarily looking at the full picture.

Right. It’s a very nuanced situation. It depends on the brand, it depends on your industry. Depends on the kind of taxonomy you’re, you’re working with. And you know, there’s a lot of, it’s not even just retooling yourself and disrupting yourself to appear in the answers and to appear in chat bot answers more often.

Luke: Right.

Andy: But it’s also measurement. How do you measure this stuff? ‘cause it’s not the traditional way you’re buying keywords.

Luke: Yeah, yeah.

Andy: You’re buying questions, but everybody phrases a. [00:13:00] Question in a different way.

Luke: Yeah.

Andy: And because these are probabilistic models, if you change the question a little bit, you’re gonna get a totally different answer that your brand may not be in.

Luke: Right. Or, or, or may not approve. Right. Like

Andy: how do you measure it? And so this is a very elaborate way of saying, don’t, I have my, you know, the experiments that we’ve done Yeah. In terms of trying to make pages vary. LLM friendly

Luke: Uhhuh,

Andy: right? So that is kind of table stakes. Just do it early. Do it now.

Yeah. Because the next version of Gemini, the next version of GPT are training on the internet right now. Yeah. So in order to get your brand into that metaphorical conversation, you wanna be thinking about these things right now, and then see how you show up, measure how you can try and measure. And then try and see where are you showing up, where are you not showing up, what topics are you the answer for?

And what topics do you wish you were the answer for? And then [00:14:00] try and develop a content strategy around that.

Luke: Like do you think people will be measuring recall from these chat experiences where brand names are mentioned? Like, like, like, oh, we, we saw this in an output and you know, a day later they bought the product or some type of connection like that.

Like are you guys thinking about it from that angle at all? I’m just kind of curious. Well.

Andy: the team that I’m literally on. Yeah. We are a B2B brand uhhuh. So we’re not, it’s not transaction consumer like did they buy the product?

Luke: Sure.

Andy: So that’s, that question is not necessarily part of my day to day.

Got it. But it is something I think about. Yeah. Yeah. ‘cause I have been in a number of different marketing technology roles throughout my

Luke: career,

Andy: and I think the extent which you show up is one metric.

Tying that back to purchase. That’s, I mean, it’s hard as it is

Luke: totally

Andy: before chatbots enter the equation.

Just trying to, you trace a SERP result through to a purchase of, you know, maybe a car or something. High value, where there’s just so many different inputs is tricky.

Luke: Yeah.

Andy: Versus banner, [00:15:00] click, click, click, clickstream, and now I bought the product. Right? That’s, that’s, you can do attribution that way.

Luke: Right.

Andy: But depending on what it is your brand is selling.

Luke: Yeah.

Andy: Attribution can get tougher. Have the variability of chat answers into the mix. Oh

Luke: yeah.

Andy: Then it becomes really difficult.

Luke: Yeah,

Andy: and I don’t necessarily yet have the gold standard answer to that yet. I think it’s something that a lot of people are figuring out.

We’re all trying to take our best educated guess as to how to. Of that.

Luke: Yeah.

Andy: But that, I think to your question about next year, I think there’s gonna be a lot of attention paid to, are we showing up in people’s chats?

Yeah.

Andy: And if not, how do we fix that?

Luke: I think it’s refreshing too, to hear if somebody say we don’t know the answer yet to something.

Yeah. Especially in a conference where people seem to have every answer to everything. I, I think it’s, it’s fantastic. Let’s switch gears a little bit. Let’s talk about MCP, you know. For people hearing the acronym MCP for the first time, what is MCP and why should they become familiar with it?

Andy: So thank you. [00:16:00] I’m, I’m a little bit of an MCP fanboy, having played with it for the last six, seven months or so. so MCP stands for a model context protocol. It was a standard released by Anthropic a little more than a year ago, and it’s very elegant in its simplicity and what it does.

Is it provides a standard way for AI agents to communicate with each other.

Using natural language. So for the benefit of, you know, non-technical people, before I get technical, it’s somewhat analogous to a standard that exists in, in the physical world with books in a library.

So let’s say you go to a library and you go to the nonfiction section.

There is a standard way that an index is presented at the back of every book.

Luke: Yep.

Andy: So if I need to know about like aqueducts or something, right? And I’m looking at a big shelf of all these books. Even if I go through a number of different books that look promising, what I’ll do is I’ll flip to the index and I’ll kind of have this nonverbal communication with the book.

Of do you have information on aqueduct [00:17:00] and if so, how much? And you see how many pages of the book are written about that topic that you’re interested in. And if not, you put it back on the shelf and you pick up the next book. That’s sort of an unspoken standard that we have an expectation of.

And it’s good because some people put the index in the middle. Maybe an index could be non alphabetized. Right? Right. The fact that it’s a standard helps us with our research in the analog world enormously. Now let’s talk about agents who live in this digital real,

Luke: right? Right.

Andy: MCP is a way for AI agents, if they just have the URL of the other agent in authentication where it’s applicable, can ask another AI agent if. Effectively, Hey, what can you do? What information do you have?

Right?

And so if an AI agent has kind of like a work for a human would be like a little book of phone numbers we used to have, right?

Luke: Oh, yeah. Yeah.

Andy: It’s a, it’s able to kind of say, okay, these are the MCP servers that I have access to.

Here’s a task that I have been given for by my user, by my company, whatever it may be, which of these services [00:18:00] might be useful? And we’ll call those services and say, Hey, I’m trying to do this task. What can you give me? Like, is there information you have? Is there a task you can complete, like breaking a task at the subtask and then calling different AI agents using MCP to accomplish that task?

And I had an incredible aha moment around this earlier this year where I was doing one of my prototypes and I set up an MCP server and then I. I went over to Chachi PT and I had Chachi PT call my MCP server with a question, right? And I’m looking at the logs of what, what CHATT is sending and what my server’s answering back.

And there’s in the logs an English conversation going on where Chachi PT asked my server an English question, and then my server responded with an answer in English. Yeah. And then Chachi had a follow up question based on my answer, and then my server provided more detail based on the follow up question.

And I’m reading this log. This is like, this could be like a teams chat.

Luke: Yeah, yeah, yeah.

Andy: This is like two people chatting with each other, but it’s actually two AI agents [00:19:00] chatting with each other. Probing. They say, oh, I need this piece of information. Do you have it? And all of that is enabled by MCP.

Luke: That’s awesome.

Andy: And that is why I love MCP, and it opens up so many possibilities for different AI agents either communicating externally or MCP or AI agents using MCP to communicate with each other within the walls of an organization.

Luke: Right.

Andy: Can you resolve this? What do you have? What information do you have here? And using semantic natural language and situational awareness of whatever the task is to accomplish things.

Luke: And it’s auditable, right? Like you’re saying,

Andy: it’s completely auditable. Yeah, yeah,

Luke: yeah.

Andy: You get the whole logs. I mean, I’m still in that phase of wonder when I see robots talking to each other, other in English it. Right. Right. Of like, why did they phrase a question a certain way? Right. What was their reasoning?

Yeah. And then what was the reasoning on my side? Yeah. With my LLM?

Luke: Yeah.

Andy: That, that I’m using the API for of answering the question in a particular way.

Luke: Yeah.

Andy: It’s, I’m still in that phase where it feels [00:20:00] like magic. So I’m, I’m really bullish on MCP as a technology that will help accelerate the adoption of AI agents, but critically.

Those MCP servers need to be sitting on top of clean data and actionable tools.

Luke: Okay.

Andy: Right?

Luke: Yeah.

Andy: Because if you’re an AI agent, you’re trying to resolve a question a situation for a customer. And you ping an MCP agent who’s supposed to be able to help. An MCP agent is like, I can’t because I’m not connected to anything.

Oh. I’m not, I’m not integrated with any of our systems.

Luke: Yeah.

Andy: I, I just have a stack of information to give you that’s been, you know, in the Vector database, that’s not nearly as useful as if there’s tools underneath those MCP servers.

Luke: Right.

Andy: And that goes back to resolving technical debt that goes back to systems modernization.

And that is what such a huge focus for Sapien is around enabling agentic AI through. The use of AI tools to get you to that place where all of this magic stuff [00:21:00] can happen.

Luke: That’s such a great breakdown too. Just, I mean, because you hear people talk about, okay, you’ll have agents, managing agents and you have these kinds of like workforces or whatever, but it’s a great way that you kind of broke down MTPs so that, you know, these are the ways that people, this will kind of happen, right?

Yeah. Like within these, these standards. So, so there is a spectrum of views kind of around a genetic ai that from everything from, it’ll change everything to, it’s a hand. Wavy hype cycle kind of thing. You know, based on what you’re seeing how disruptive do you think it’ll actually be? and how long will it take?

if you had to predict, like, what timeline would that take for it to become like effective?

Andy: Well, I think that it’s gonna be an incredibly big deal. Yeah. It’s gonna be profoundly transformative. But to our point earlier, I think it feels like a hype cycle.

Luke: Yeah.

Andy: Because people jump at trunk.

Implement it without getting the basics right first. Yeah. Without getting their tech debts sorted out and they hit a wall where the pilot program fails or whatever it might be. ‘cause they didn’t get, they jumped too far too fast. And then they hit the [00:22:00] Trav dissolution. And they’re like, well this is all garbage.

Right. Didn’t work.

Luke: Right, right.

Andy: Without going about it in. Methodical way of really assessing your landscape, coming to terms with what you need to get sorted out.

And then implementing that. So I think in terms of adoption, it’s gonna vary on how much technical debt you pay. Yeah. You have large, vast organizations, it may still have COBOL lying around.

Right,

Luke: right, right, right.

Andy: I mean, my team is very fortunate. We don’t have a lot of legacy code within the marketing group that’s holding us back from doing lots of. Sort of act, sort of activating AI capabilities within our team globally. So that’s good. But I know we’re fortunate in that regard and a lot of organizations are not, and Publicis Sapien more generally exists in, in Parts and help them get to that place where they can.

Achieve that. So, I don’t think there’s gonna be, wow. Now everybody’s ready. Yeah. Kind of moment,

Luke: right?

Andy: Next year.

Luke: Yeah.

Andy: I think it’ll be more similar to like, past [00:23:00] cycles. I could say back,

Luke: yeah.

Andy: Right? Yeah. Some companies were ready to do website tomorrow. Some companies really had to get their act together before they put up and like an e-commerce site,

Luke: right?

Andy: Like, if you were gonna move back. In the late nineties, if you were gonna move into e-commerce, you had to get your factories straightened out, your supply chain. Oh yeah. All of this non-digital stuff.

Luke: Yeah.

Andy: You had to do before you could sell one widget online. Right,

Luke: right,

Andy: right. And if you try to open a website and say, Hey, these are our products, they’re great.

You can’t buy them. You have to walk to one of our stores. Not nearly as effective as e-commerce right then and there.

Luke: Yeah.

Andy: And so it’s a difference of where are you in your maturation? Yeah. As a company. And there’s a spectrum of that. And so I think it’ll be a slower evolution. Where Agentic will, you’ll wake up one morning and suddenly lots of people are using agentic very effectively in their in their industries and their, in their organizations, rather than a big bang, like, oh, now it all works.

Luke: Right.

Andy: For every company.

Luke: Right, right. Yeah. No, and I think was kind of curious, did you’re, you’re in the marketing world, right? Yeah. Like, do you think this is a especially tough [00:24:00] thing in marketing to grapple with? Because marketing tends to kind of, it’s a showy, industry, right?

Like where, this is where you are gonna kind of preview the future, right? do you think it’s particularly difficult in the marketing space with handling adoption or, not moving too fast or, you know, not painting a picture of the future that isn’t real or that, that type of thing.

Andy: Well, I mean, for us, yeah. And, and you know, I think a lot of what Sabian communicates here to the world is about how we can use our, the technologies that we’ve de developed, the AI powered products that we’ve developed

To help people and organizations get to where they need to be to see the future.

Yeah. We try to be very clear-eyed in communicating what an organization needs to do to get to that place. And also we have the tools to get them there.

Luke: Awesome.

Andy: Right. Yeah. And, and be very clear-eyed about it.

Luke: Yeah.

Andy: Because, you know, if we know as well as anybody

Luke: Yeah.

Andy: Having done systems integration, system modernization for decades, you know, as well as anybody.

But if you try [00:25:00] to like slide the shiny thing in on top of technical debt

Luke: Yeah.

Andy: It’s like building a mansion on top of sand.

Luke: Yeah. Yeah.

Andy: Right. It looks beautiful, but it’s gonna start sinking.

Luke: Totally.

Andy: so one of the things we’re very clear-eyed about when you working with organizations is being very clear about where they are.

Right? And how we can use our products to get them where they need to be, such that agentic will work for them.

And you’re not gonna have pilot programs that stall out because the underlying technology is not. There yet. And so we have modernization tools that will work right now to modernize your old code. Awesome. And we have you know, product Bode, which is an agentic platform to then implement a agentic AI solutions on top of your modernized tech stack.

Luke: Cool.

Andy: So we have that suite to get you there and then to sustain that momentum going forward.

Luke: Yeah, so part of the step I think, you know, like you’re, you’re illustrating here is just really kind of having some, some fundamental conversations with, you know, the brand you’re talking to about like, okay, here’s where we’ve looked at your stuff and here’s where we need to clean [00:26:00] things up and get things in the right spot, and then how we can get you there.

I think it’s really interesting. I mean, ‘cause there’s so much, I mean, just to show what Agen can do involves, you know, okay, here’s kind of painting the future. And even when you’re looking at stuff, and this is something we hear ‘cause we have browsers are doing agen. Right. So you’re seeing, okay, well the thing can make a tweet for me, it’s 10 steps.

And you’re like, I could just go make the tweet, right? Like, or maybe I just wanna make the tweet myself. You know, that, that type of thing where like, people are, people don’t really know exactly what it’s gonna look like. Right. But like, I, I feel like in that process you outlined, you’re probably gonna help with finding market fit for a lot of these things, or finding out how is it, how is gen really gonna be helpful?

Right? Like, are, are you guys finding that at all when you’re having these conversations?

Andy: Well, yeah, I mean, a lot of it comes from understanding what the pain points are,

Luke: right?

Andy: Where the business has opportunities for growth.

And, you know, particularly for agent tech, the where the customer experience gets slowed down.

Where, and then you see drop off in, in retention and sales and so forth. Or repeat customers [00:27:00] because there’s friction, right. Right. How do you zap that friction? Agents can be great at zapping the friction

For for customers, you know, when before public api I was in different other, different parts of Publicis and working with more consumer oriented brands.

And part of the vision that I frequently talk about and hear about and daydream about is a world without waiting on hold.

Luke: Right. Right.

Andy: Is is where the AI agent you’re talking to. Both, there’s two parts that are critical, understands what you’re trying to achieve. And maybe

Luke: that’s an order

Andy: rather than a problem. Right?

So eight understands you completely, and B can actually act.

Luke: Right.

Andy: Right. And can actually then solve your problem. And I think it actually is not hard, you know, when you’re talking to clients about transformation, to say the reason the agent can’t resolve the customer’s billing question is because it’s not connected to the billing system.

Luke: Right. Right, right.

Andy: So we need to do a little bit of work. To figure out what that systems [00:28:00] integration is going to be. Because when you make that connection, you’re gonna have a lot of happy customers.

Luke: Right, right,

Andy: right, right. You’re can have a really great experience where they can just resolve their problem and, or, or then, you know, upsell.

They may want to upsell or, or self upsell. I,

Luke: yeah, yeah, yeah. Upgrade. Yeah, yeah, yeah, yeah.

Andy: Right. Like they, they have such a surprise and delight with that moment of reduced friction.

Luke: Yeah.

Andy: That it’s just only good news for your bottom

Luke: right.

Andy: And so I think there’s enormous potential there. It’s not that hard to explain to people because everyone, everyone calls customer service at some point.

Everyone deals with other brands. We’re all human beings in the world.

Luke: Right? Right.

Andy: And so we all felt the friction ourselves at various points.

Luke: Yeah.

Andy: And agents can go zap those points of friction very effectively once you get to the point where they’re deployable and they’re implemented.

Luke: It makes you think about marketing from a bit of a different angle too, like around this experience where like maybe it’s not necessarily about advertising as much as it is about just improving the customer experience.

Right? Like I. I’m gonna steal the on hold example because I think it’s [00:29:00] fantastic.

Andy: From your vantage point of publicist what signals or patterns feel most indicative of major shifts have. Happening across creative tech and marketing right now.

So it’s interesting ‘cause Publicis s Sapient itself does not do nearly as much creative technology per se, as the other divisions within Publicis, but.

I do have vantage point of those other divisions because we have this power of one AI driven philosophy within globally, within Publicis Group.

And so there is an enormous amount of communication between subject matter experts and all the different branches of Publicis across the world.

Around best practices, around different tools. Like, Hey, has anyone done this before? Hey, we’re thinking about improving our visibility on chat. You know, what are some techniques people are talking about? We use this vector database or that vector database, what performs better? All sorts of technical stuff.

And even more business logic stuff, you know, common problems that in different ways, different clients face. And there are [00:30:00] clients that we. Serve along with other branches of Publicis. As sort of a power of one proposition.

Luke: Yeah.

Andy: So there’s enormous collaborative thinking, an enormous yes. And

That happens,

Luke: right

Andy: when you have so many experts. Who can communicate with each other so fluidly. That it gives me a view as to how much creative problem solving is happening globally throughout the network. And has been enormous value to me personally in solving some of the technical questions I’ve had.

And I’ve been able to, you know, some. Supply perspectives to other people in the network of subject matter experts. And so there’s a certain amount of, through the types and variety of questions that are coming up as clients transform. That is the way that I experience the transformation happening in marketing.

Luke: Awesome. '

Andy: cause just by virtue of the type of questions people ask. You get a sense of like, oh, they’re really [00:31:00] future forward. Like they’re really going out into the, the bleeding edge of what’s possible here.

Luke: Right.

Andy: And that inspires me. And then likewise, things that I experiment with can inspire other people who may be working with clients on a totally different thing.

Luke: Yeah.

Andy: So it’s that, that shared expertise and the type of questions, really, that’s what gives me the sense of how fast this is moving and how much change is happening.

Luke: Yeah. And a lot of these problems, I mean, with the dimension that AI adds to everything. Whether it’s a consumer problem or a B2B problem, a lot of the, a lot of the issues are similar, right?

Like where okay, like, you don’t want to, you wanna tap into this data set, but you wanna make sure that this piece, this, this set of data doesn’t kind of get into the pool. Right? Like that types of thing. So I, I think that’s a really, really great perspective. You know, for, I appreciate you kind of making the time for all this.

I, I’m sure people want to kind of follow along with what you all are doing. Where would you recommend that they take a look and where can they find you online?

Andy: Well, I’m on

Luke: LinkedIn. Okay,

Andy: cool. Andy. A-N-D-Y-M-A-S-K-I-N. There’s only one of me as far as I know on LinkedIn. And then Publicis Sapien is [00:32:00] publicis sapien.com.

Luke: Cool.

Andy: I, it’s funny, I was talking to somebody and they didn’t know I, I, I’ve come to pronounce. Publicis is Publicis. Yeah. Because it’s a French company and we all just refer to it by the French name. But I was talking to somebody who said, oh, Publicis. Yeah. I was like, sure. You know, I, yeah, that’s fine.

That’s awesome. Somebody publicis sapien.com if you, if you need help with the spelling and that’s, that’s where we have all our, our, our, you know, our products and, and our ai. Thought leadership and all that great stuff.

Luke: Awesome. Well, Andy, thank you so much for, for taking the time today. A really great conversation folks check out out on LinkedIn and yeah really appreciate it. Love to have you back too, to talk about some of this in the future.

Of course. Alright, thanks man. Alright, thanks a lot.

Thanks for listening to the Brave Technologist Podcast. To never miss an episode, make sure you hit follow in your podcast app.

If you haven’t already made the switch to the Brave Browser, you can download it for free today@brave.com. And start using Brave Search, which enables you to search the web privately. Brave also shields you from the ads trackers and other creepy stuff [00:33:00] following you across the web.