Building Your Army Of AI Agents: What Marketers Need To Know

[00:00:00] Tessa Burg: Hello and welcome to another episode of Leader Generation, brought to you by Mod Op. I’m your host, Tessa Burg, and today I am excited to invite some other Mod Oppers to a special panel that we’re doing. We’ve all been hearing a lot about what AI and marketing can do, and being given all these examples.

[00:00:19] Tessa Burg: It can automate workflows, it can help you create content. Things that we should be doing, like optimizing our websites for LLMs and doing GEO style tactics, or even going as far to say we should have our own agents and the agents will act on our behalf. But one question that we get quite often is, how is this all even possible? And where do I start?

[00:00:45] Tessa Burg: So in today’s episode. The guests on the panel know how to make this happen. So I’m very excited to welcome Fabio Fiss, Aaron Grando and Javier Lopez, and we’re gonna be telling you not only where to start, but the technology that enables what feels impossible and makes it possible, and give you some examples that you can take back and start to explore on your own.

[00:01:09] Tessa Burg: So thank you all for being here. I’m super excited to kick this off.

[00:01:17] Tessa Burg: All right. We’re gonna open up with first having each of you introduce yourself, and then we’ll get into the very first subject about what is MCP in A2A. But first, Fabio, tell us a little bit about yourself and your role here at Mod Op.

[00:01:31] Fabio Fiss: Hey everyone. My name is Fabio. I’m a VP of Technology here at Mod Op.

[00:01:36] Fabio Fiss: I’ve been with the company for 10 years. Uh, I’m based in Miami, Florida, and I lead our development team and our digital experience, uh, group.

[00:01:46] Tessa Burg: Awesome. Aaron.

[00:01:48] Aaron Grando: Hello everybody. I’m Aaron Grando. I am VP of Creative Innovation here at Mod Op, and I work on Tessa’s innovation team. Um, and we are responsible for, um, identifying and rolling out AI technologies to our internal teams and figuring out ways to deploy them for clients as well.

[00:02:06] Tessa Burg: Love it, Javier.

[00:02:09] Javier López: Hey, my name is Javier. I’m based in Panama. I’m a senior software engineer. I work in Tessa’s team also with Aaron, and I do everything about software development and plumbing AI systems.

[00:02:26] Tessa Burg: Thank you again for being here. So a lot of our clients and a lot of our team internally are really familiar with how to use ChatGPT, and even some of us have begun to build our own GPTs.

[00:02:39] Tessa Burg: So we have a good sense that we can automate some highly repeatable tasks. We can start to do things we’ve never been done before. I say this in so many episodes, we’ve tested or reviewed over 300 apps. But what comes next in the next horizon of AI is really all about connecting those things so that agents can operate independently so that we get more accurate measurement on these workflows.

[00:03:07] Tessa Burg: And there are two. Um, technologies, or at least two that I’m familiar with that help with this, that I wanted to get defined first. And I think some people have heard of these, but they might not have a good definition. So, Aaron, can you start us off on beginning to define what is MCP and what is, um, A2A?

[00:03:28] Aaron Grando: Yeah, so both MCP and A2A are, um standards that, um, respectively, Anthropic and Google. And I think, uh, an overall like technology group of different companies have kind of come together to set some standards for themselves about how these systems can, you know, actually exchange information and exchange instructions.

[00:03:50] Aaron Grando: Um, so I think the, um, the main thing to think about MCP is that it’s not necessarily an API, I know a lot of marketers are used to APIs, but it’s a, um, it’s a standard for how you set up an API so that other APIs can reach out to your API and know what to expect and know how to respond. Um, and what MMCPs enable is essentially, um, agents to pull in information from external systems and incorporate those, um, that information into their prompt, uh, or their, their generations.

[00:04:24] Aaron Grando: So, um, you can think of stuff like a product information database pulling in info from, uh, an MCP provided by that PIM uh, provider. Um, A2A. It stands for agent to agent, and that is a standard that Google is setting up to allow agent-based systems to talk to each other and have agents provide context to other agents as opposed to other human beings or like human users.

[00:04:51] Aaron Grando: So the, the use case there would be for things like an agent system that may want to have a user request, a food delivery. And then, um, you know, that request goes to, um, the agent that is running the conversation and then that agent goes out and makes a request to Pizza Hut’s agent. Um, and both of them are speaking in A2A standards.

[00:05:14] Aaron Grando: And the, the conversation happens between those agents passing back and forth payment information and receipt, and then that eventually bubbles its way back to the user. So. Again, these are like enabling technologies that are really intended to set a standard baseline for protocols. And we’re really used to working with protocols when we’re working with digital products.

[00:05:37] Aaron Grando: The whole internet is based on protocols. HTTP you were probably familiar with. Um, so it’s about setting up these standards for applications to kind of agree on and, um, have, as you know, a kind of a lingua franca between, um, their functionalities.

[00:05:56] Tessa Burg: That’s a fantastic answer and very thorough. Javier, have you had a chance to test any of these emerging protocols?

[00:06:07] Javier López: Oh yeah, I, I’ve tested some, like sometimes when I’m developing, uh, I can use MCP to have agents talk to my database and, and help me debug issues. I, I wanted to give a, like a quick example for people to understand better MCP and A2A. I like to think about it in two types of workplace behavior. For example, sometimes you get hands-on, you open Excel, you grab the sales numbers, make a chart that’s like an agent using MCP to directly work with a tool or system.

[00:06:37] Javier López: But other times you use delegate, just Slack, a teammate you tell it to, to run the sales report and drop it. In the deck. And then that’s A2A, it’s just an agent telling another one that has the access, the tools and knowledge to, to do some tasks. So, so that’s like the, the, the difference between the two standards, like sometimes can be confusing, but I mean, I, I have used them, um, multiple times currently or recently.

[00:07:00] Javier López: Um, Claude, Claude chat release, uh, Anthropic You’re able to, to use chat to create data issues or to search your documents at Google Drive. So, so those are the advancement we are starting to get in the, in the field. Yeah.

[00:07:15] Tessa Burg: I love that. So just using it in sort of your everyday to make what you are doing more efficient.

[00:07:20] Javier López: Yeah.

[00:07:21] Tessa Burg: When we, Fabio, I know you work and lead the teams at, on the client projects, when you think about, uh, the power that the standard protocols can bring to existing applications. What are you seeing as some opportunities, um, in our digital experience work where, where this could really play a significant role?

[00:07:42] Fabio Fiss: So I’ll give you three examples of, um, things that are kind of starting and enhancing, uh, on our pipelines. Um, so one example that I see, uh, we’re exploring, right exploring right now with a platform called Active Demand, uh, in the senior living space, actually in the industry there. So one of our clients, uh, that works in that industry, they are looking at active demand as a platform to replace their CRM.

[00:08:09] Fabio Fiss: Plus some integrations with the OpenAI API and going through, uh, an agent to agent type of integration for automation, right? So, uh, they’re replacing a very specific narrow CRM that existed in that industry. With this AI based platform. Um, and, you know, it deals a lot with the user journey and, uh, not just the CRM piece of it, but bringing all that data into the system and then using that data for their workflows.

[00:08:39] Fabio Fiss: So, um, seems really powerful. We did some discovery last year on it, and this year we’re actually gonna go into full implementation. So, uh, we’ve been very excited about that one. Um, a second example, which is. More of an evolution of some things that we’re seeing in terms of demand from clients is, um, chatbots plus some other marketing integrations, right?

[00:09:01] Fabio Fiss: So we have a lot of demand that existed before, uh, for chatbots with platforms that marketers know super well, like HubSpot or Zendesk, and now they’re looking to understand what else they can do with agents that then can take this data and. New automations in their systems, right? So we’ve been exploring, uh, some things with HubSpot already with one of our clients in the real estate, uh, world or industry.

[00:09:27] Fabio Fiss: And, um, you know, there’s only more potential as, as these platforms also incorporate some of the protocols and some of the APIs from the foundation models. So we’re excited about, um, what’s possible with, with those, uh, platforms. And then finally, uh, more of an internal. Type of use case. Um, and it’s actually been very useful for our projects, uh, in audits.

[00:09:52] Fabio Fiss: So we’ve been exploring both our dev team and our uh, SEO team. We’ve been exploring using agents that we build either through GPTs or we build a little application, uh, to do audits. That we would have to do it very manually. So if we’re looking at things, for example, if the site is responsive, does it have connections with social media?

[00:10:14] Fabio Fiss: Does it have integrations that we’re looking for specifically in the code? We can build an agent that then goes and does that audit for us. The SEO team is also checking the content on a page to make sure that it’s all optimized for SEO, right? So a lot of the things that we would do with different platforms and different tools across many systems, we can all automate through these agents.

[00:10:37] Tessa Burg: I love that. And Aaron, you come to us from the Creative SBU and have a lot of years of experience and more the content creation, social media, creative design. Where do you see these playing a role in helping to elevate that quality of work and, and kind of give our creative teams different superpowers?

[00:10:58] Aaron Grando: I see both of these as really helpful at the, um, the part of the process that both pulls information into, um, an AI workflow and also kind of puts it back into the human workflow.

[00:11:13] Aaron Grando: So, for example, um, you could use an MCP to access, um, a product database and have, um, uh. Look at all of the different, um, you know, specific, uh, products that are maybe aligned to a specific attribute. Like you could select all of the. Um, high protein, uh, meals available from a specific, uh, brand. Um, and then you could do something in the middle, you know, using an AI assistant through like a conversational interface to maybe create that content drafted up, and then in the end, use an MCP that’s hooked up to a content creation tool like Figma, or one that’s set up to interact with a file system like SharePoint.

[00:11:57] Aaron Grando: To actually have the AI go out, make request to a tool. This is essentially what MCP does to go out and affect something outside of your chat environment. So it’s a way of connecting info from outside of, um, that AI work environment in that chat box to, you know, actual outputs and inputs that, um, affect what we do outside of that conversation.

[00:12:22] Tessa Burg: I love that. And then. This has come up a few times talking about pulling in data from external outside. If somebody wanted to explore who right now is providing or meeting these standards, how could they find that? Like, ’cause I could think of lots of different places. I would like to get data, but how do I know that they’re using the same protocol as me?

[00:12:50] Aaron Grando: With MMCPs, it’s really kind of like, um, a home brew solution right now in a lot of ways. Um, there are MMCPs that are out there that are offered, you know, either for free or for subscription. Essentially they’re data providers, data brokers. Um, for us, the way that we’re thinking about it is, you know, we have different kinds of data that we want, might wanna make.

[00:13:12] Aaron Grando: Accessible at different levels of permission. So we’re thinking about using MMCPs to allow different agents to access different types of information based on how we want to have, um, that permission delegated. Um, so we think about it as a little bit of traditional, like data domestication, data cleansing, um, data organization, and then eventually data serving.

[00:13:38] Aaron Grando: Uh, up through essentially an API, like you might traditionally, but it does allow us to work with a little bit more unstructured data, a little bit more different types of data in a way, pulled in through the same means and then processed with a, uh, a AI excuse me. Um, I think that answered your question.

[00:14:00] Tessa Burg: It it does, and I think. What we would want clients to know is some of this could be available and platforms they already have today, like their Yeah. Um, LiveRamp. Nielsen. I think that sometimes when you hear we’re gonna go get external data, uh, for whatever reason, sometimes people jump to like, scrapers like this.

[00:14:21] Tessa Burg: Like, this might not be legal, but in fact it really is about. Um, a lot of us as marketers have these data sources in our stack. It’s simply how do we connect the dots so that we can power these different solutions? And again, what I think is important for marketers is when we start with the challenge or problem we’re trying to solve, always flowing out that process.

[00:14:48] Tessa Burg: And after you understand the process, look at what tech. Is already underneath there because this may just be a matter of continuing to get additional data and features out of your existing tech stack. And I think that makes this a little less intimidating because these are new terms and it’s like, well, I don’t know if, if this applies to me, I don’t know if the team I’m working with knows how to do this.

[00:15:11] Tessa Burg: The team you’re working with may not know how to do this. However, it definitely can apply to you. And as we’ve said in other episodes, you know, AI features, AI access points. Um, data that you can leverage to increase the quality of your AI is already inside your business.

[00:15:29] Aaron Grando: I think to your point, the important thing there for me is that so many of these challenges that we’re trying to solve and innovation, you know, throughout DX, are the same problems that we’ve always been trying to solve.

[00:15:40] Aaron Grando: Like it’s, it’s the same things, uh, just a new package for them. It’s the same data. In a lot of ways MCPs don’t give us access to new data. It just gives us a new way to access it and a new set of instructions for our AI coworkers essentially to, uh, understand how to use that data and what they’re doing in our, in our chat interface or our API workflow.

[00:16:05] Tessa Burg: Aaron, you just said something that can really open up a whole other Pandora’s box, and so we’re gonna go there. You said AI coworkers. And I think that that is a future some people get very intimidated by. But Javier, I’m gonna throw it over to you because you’ve already started using AI as sort of an assistance.

[00:16:28] Tessa Burg: So how far off do you see having an agent who’s more like your coworker actually being.

[00:16:39] Javier López: Well, I think it depends. It depends of of the expectations you have. ’cause yeah, they can start being maybe a coworker right now, but for what tasks? Because there is a lot of responsibility in many tasks. And when an AI is your coworker, you are the one responsible for what he does.

[00:16:57] Javier López: Right. So, so that’s, that’s like, that’s a challenge when you start giving ai, for example, tools to create a to-do list and organize a calendar. If the AI does something wrong. I mean, the AI is not responsible at the end of the day is, us that pick the solution, right? So I think you can start having AI help you in many tasks, but you have to, to think very.

[00:17:21] Javier López: Totally about, about what, what is it gonna do, what happens if it goes wrong? And you have to start evaluating those things. So right now I use AI for, for many things in my, in my day-to-day things. But I, I have to review what it does. ’cause at the end of the day, I’m responsible for, for that work. So I think they are starting to be there now, and I think this year we are gonna have, uh, be able to, to use, Hey, can you fix this Word document for me?

[00:17:46] Javier López: And it’s gonna do it very well. But I think for, for use cases, like generate a report I’m gonna present to my boss, then that’s gonna take a little more time. Or, but you’re gonna have to, to intervene to, to see the output of, of the AI coworker. Yeah.

[00:18:02] Tessa Burg: Yeah, I agree. And you gave some examples that entry level employees used to do, or even junior developers do this research focus on bug fixing.

[00:18:15] Tessa Burg: Focus on testing and QA is another area where AI is really advanced. And I know there’s a concern in the market, like what will entry level folks do. And in fact, just last week they reported that the number of jobs went down and there’s no specific evidence that says it’s ai, but there’s these quiet reading between the lines that boards and.

[00:18:43] Tessa Burg: CEOs have said, well, we’re just not gonna hire because we are seeing AI be able to do a lot of these entry level worker tasks, and we wanna see how this plays out. Um, I, I certainly have an opinion, but, uh, Fabio, I wanted to get your thoughts. How do you think this is going to impact. Teams, especially if you think about web development, ux, our entry level folks coming into marketing, advertising and technology as we go forward.

[00:19:15] Fabio Fiss: I have three thoughts on that level. I think, um, first I feel that automation and looking for efficiency, it’s a natural path all the time, right? So if you look from the CEO point of view or any like director, marketing director, they’re gonna look for that automation process and that efficiency process, right?

[00:19:35] Fabio Fiss: Uh, so that’s kind of something that anybody just starting in the, the industry should have in mind already that. Um, you have to start with that concept in mind that these tools are there to make you more efficient and to elevate your skills too. Right. Um, the second thought I have is that, um, there’s always this.

[00:19:57] Fabio Fiss: Fear of a new technology replacing, um, jobs, right? And replacing people in general. But if you look at from more of an optimistic point of view, um, it’s, it’s usually not the case in the long term. What happens, and I guess the best example is social media, right? If you look at just 20 years ago, there was no such thing as a social media manager or a social media.

[00:20:23] Fabio Fiss: Content creator. If you look at video production, there was no such, there was no such concept of a, uh, content creator, uh, in the way that YouTubers are seeing today, right? So you may get a, an impact in the short term of how people operate on their day-to-day. But in the long term, you’re gonna see brand new.

[00:20:44] Fabio Fiss: Uh, jobs and positions to surfacing. So I feel like that’s a na, another natural path of any new technology revolution. And then the third, uh, part of my thinking is always that look at the ways that the. Technology companies are positioning these technologies as well. They’re not showing these technologies.

[00:21:06] Fabio Fiss: There was a way to replace it. Uh, we were talking about that before. For example, with Microsoft, they call it a Copilot. They’re not saying that you replace the pilot, it’s just a copilot. So you’re gonna have an assistant to keep like working with you. So the way that I feel about this is that, um, you are always gonna be as a developer or some other type of leader in the team.

[00:21:28] Fabio Fiss: Uh, or even as an entry point person, uh, you’re gonna beat the conductor and you’re gonna have instruments on your orchestra, and you’re gonna be able to orchestrate, uh, the, the, the music and the songs. Um, so this is how I envision, um, these AI tools and, and, uh, platforms for the future.

[00:21:48] Tessa Burg: Yeah, I agree. And Aaron.

[00:21:51] Aaron Grando: I mean, we have talked a little bit about it. We recorded an episode yesterday and we thought about this problem a lot and I’ve been thinking about it a lot too. And where my head goes is that especially for people that already work in technology like us, um, we have a lot of really transferable skills that where we are taking an expertise that we’ve been able to.

[00:22:13] Aaron Grando: To build like a pretty good moat around technical expertise is hard for people to break into. It’s hard for people to get up to speed, especially with how fast technology, internet technology has always moved. Um, but I. I could see that moving out to kind of diffuse across the entire organization where there could be opportunities for an engineer on every single type of team now, as opposed to being siloed in like an engineering team or a technology team, where having somebody who.

[00:22:44] Aaron Grando: That has technical background, maybe using an AI coding assistant, like we talked about yesterday. Um, be able to take concepts from engiineering and apply them to like, you know, domain specific business challenges. I think that only kind of, it speaks to the, the YouTube creator versus the videographer, um, mentioned that Fabio just had, where there’s going to be more people doing technology engineering work, there may be less engineers, if that makes sense.

[00:23:14] Fabio Fiss: And I think to add to that, the role of the senior engineer may evolve to be elevated as well, because you may have to be more of a mentor rather than being the person that’s just doing the code day to day. Right. So I think one thing that we’re thinking right now internally is actually how do we change our mentorship programs?

[00:23:33] Fabio Fiss: In-house for developers that are growing in expertise, but need to assist people that are just starting their journeys as as a developer as well. So I think that mentorship capability will increase. It’s the same way that happened in the past with a developer that was just coding to become an architect.

[00:23:51] Fabio Fiss: Right? So I think those skills will evolve even if you’re not doing code in the same way, uh, that we, we were doing before.

[00:24:01] Tessa Burg: Yeah. I agree. For some reason, like I have very little concern for young people and that maybe that’s because the people I, young people I’m surrounded by are like so innovative.

[00:24:14] Tessa Burg: But just for example, you know, my daughter who is 11 is already producing her own movies and I feel like AI for all of us, whether you’re 11 or you’re in your forties. It’s giving you the ability to create, develop, elevate in a way you’ve never done before, and that will lead to different positions. So while there probably is a silent pause on hiring for the roles that exist today, those who are awesome problem solvers and always are curious about how they can use the tools in front of them to create, solve problems and elevate.

[00:24:56] Tessa Burg: Will be positioned really well for the, for the jobs of tomorrow. Um, Javier, I saw you raise your hand unless you were just Yeah.

[00:25:05] Javier López: I wanna, I wanted to add to what Fabio said. Like recently I saw a, a very prolific maintainer in the JavaScript ecosystem, commenting, he’s a senior engineer and he said that now with AI he can mentor and now he also has time to code.

[00:25:19] Javier López: So before he only had time to mentor Right. And now he has the time to go. So, so we, we don’t know how these new tools are gonna affect. But they are, they are changing the way, the way people work. So people are able to do many more things. Uh, like for example, I also read an article today from Martin Fowler, which is a big software, uh, famous person let’s say like that.

[00:25:42] Javier López: And he said that now the developers are able to do things that before there was too many busy work, like reading the documentation, how we gonna implement this? Even if it was simple, you spend hours, you’re reading how, how this thing work. Now with ai, you can. You can just ask a few questions I and can help you, like, get the, the base parts together.

[00:26:00] Javier López: And now the developer with his knowledge takes it from there and, and finish working on, on something that before he would’ve taken hours, now he is taken maybe 30 minutes. Yeah.

[00:26:10] Fabio Fiss: Yeah. And I wanted to add to that too, ’cause uh, recently I’ve been watching a lot of, uh, content about how to use, um, these tools that generate code.

[00:26:18] Fabio Fiss: We were talking about vibe coding before right? In another, uh, episode. So, uh, one of the. Initial, uh, applications that a, a regular developer can start is just a notepad application, right? Uh, so you. And, and these days with these tools that are available, you can code that very quickly. Now, in just a matter of hours, you can ask through prompting to create a Notepad application.

[00:26:40] Fabio Fiss: So what I I think is gonna happen is that people that are just starting out, that barrier is gonna be gone of just being able to start a hello world type of application or very simple type of application. And now they’re gonna be able to. To be even more creative because the entry point for that is already simple to, to, right?

[00:27:00] Fabio Fiss: So you don’t have to spend hours and hours sometimes weeks, right? When we, we had to in the past to build a simple application. So now you have more space and time to be more creative. So I think that’s what’s gonna happen in a, in a sense for people that are just starting out, which is, if you think about is what happened again with social media, when you remove that barrier of publishing, like publishing what used to be very hard.

[00:27:23] Fabio Fiss: Right. And then you made publishing very easy. Now you create a whole nother industry of creativity just by making that barrier of entry lower, right? So I think that’s what’s gonna happen in a sense.

[00:27:36] Tessa Burg: Yeah, I love that. So I’m gonna recap a couple of things before we move into the next section. So, at the beginning of this, we talked about some of the enabling technologies.

[00:27:48] Tessa Burg: Make building an agent army or simply automating workflows possible. And we brought it down to the marketer level. So I, I just googled to the most common marketing platforms because I’m very positive that any marketer listening to this already has some of these platforms inside where they can start accessing that data using MCP.

[00:28:10] Tessa Burg: So we already mentioned Nielsen Trade Desk. DV 360, if you’re doing any digital media and platforms, um, most of the ad servers, CTV campaigns, any of your web analytics. So again, when you’re thinking about, Hey, I wanna optimize or automate this workflow, what? Lay out that problem, lay out the process.

[00:28:32] Tessa Burg: Tech, you have some tech below that can help be the external data to help make sure that how you’re optimizing can start to be self-sufficient using existing agent technology. And then the other thing we talked about is opportunities in the market, uh, as it relates to jobs. You know what, that this will impact the types of roles available.

[00:28:54] Tessa Burg: But the next section I want to get in is. If we are directing and orchestrating as Fabio put it, if we are conducting this orchestra of agents and we do have assistance working with us to, uh, operationalize independently against workflows, what does that mean for what we’re building for clients, for example, apps and websites.

[00:29:18] Tessa Burg: Someone sitting at a buyer business, we’ll use B2B first. As an example, I’m an engineer. I need to source different components and parts to complete a project. I wanna evaluate what’s available. I might have an agent go do that for me. Um, same with any market research firms or professional services organizations.

[00:29:38] Tessa Burg: I wanna track trends in the market. Um, I want to better understand where. The financial impact of mergers and acquisitions is going. So when we think about, there’s so much work that is put into research and starting to design, uh, solutions and for buying, purchasing, even consideration set. I. What are we doing?

[00:29:59] Tessa Burg: What are we thinking about today, Fabio, as we look at the role of the websites, the role of apps and other traditional tools that have delivered this information to clients, and how do we think that’s gonna evolve as agents emerge?

[00:30:14] Fabio Fiss: I think, um, if I had to think about just, um, the way that we talk about websites to our clients or applications, there’s still your.

[00:30:23] Fabio Fiss: Uh, hub of information, right? Uh, so the website is your place of ownership, of your content, your data, and I think that’s also a way to think about how these tools, the applications, and the websites can become sources of truth for data as well. So I’ll give you two examples of things that happen with clients.

[00:30:44] Fabio Fiss: We talk about a lot, uh, uh, we talk a lot with clients about data integrity. Uh, if you have, for example, an investor relations, uh. Piece on your website, right? And how important that data is to feed other platforms. Right? So, um, same way with industries, for example, in the real estate world where a lot of their data is in their content management system, I.

[00:31:08] Fabio Fiss: For, uh, properties or offices or things that then need to be pulled to other platforms. Sometimes we’re pulling from external sources into the website, but now I see a world where the website can also be that source of truth for data gathAarong, and then ultimately for AI agents. To consume that data and move it into automation.

[00:31:31] Fabio Fiss: Right? So that’s one area that I see a lot of our applications and websites, um, becoming more and more the source of truth for data, data gathering. You know, that you can control your privacy policies there, you know, you can control your data integrity. You, you know that there will be a human in the loop in that process so you can maintain the quality of the data.

[00:31:53] Fabio Fiss: Um, so that’s how I see this evolving.

[00:31:57] Tessa Burg: I agree, and it really puts a strong emphasis on making sure you understand what your buyers and target audience really need to make a decision. Right there we talk about, you know, AI can help us or does help and is already accelerating content creation. Just creating content will not be enough in an agent world.

[00:32:18] Tessa Burg: These are gonna be optimized by people to only do and to gather exactly what they need and want. And so really understanding those needs wants across the journey will be very important as we continue to optimize websites and build websites that matter and drive performance.

[00:32:34] Fabio Fiss: Yeah, I can give you one more example of, um, another type of client or industry, uh, in the healthcare space where they have a lot of information on their websites, right?

[00:32:44] Fabio Fiss: So, uh, one of the things that we’ve had requests from different clients, uh, more recently, is about being able to gather all that information. Make it consistent and, uh, accurate. So you can use it for chatbots and other types of interfaces that interact with the user, right? But their website is the source of truth for that data that’s been regulated, has been reviewed in that specific industry, which is so important with compliance and, and things of that nature. So,

[00:33:13] Tessa Burg: Yeah, that’s a great example. Aaron, what about on the B2C consumer side?

[00:33:19] Aaron Grando: Going off of what Fabio just mentioned about keeping a human in the loop, I think these technologies are one of the enabling pieces that we have there. Um, a really common MCP pattern that I think we’re gonna see emerge is using an MCP integration into, uh, an AI chat experience that does allow for that human in the loop, um, approval or rejection of a concept or an idea before it takes like a further action.

[00:33:46] Aaron Grando: So I think that’s one major place, but um, in a lot of ways I think it kind of speaks to that we’re kind of shifting from what I’ll, you know, just generally call classical computing, um, paradigms where, you know, one plus one equals two always, um, a hundred percent of the time. Um, we’re, we’re moving into a little bit of a more probabilistically like driven set of paradigms when it comes to interactions and, um, data retrieval.

[00:34:16] Aaron Grando: Um, so where those two things overlap, um, will be where we wind up. Um, there are gonna be applications where it’s much more important to have fully accurate recall of a very specific set of data. And then there are also gonna be applications where, um, we want to have the more like. Probabilistic AI driven, um, predictions coming out of the system, and these technologies are kind of the, the glue that kind of lets us put those two pieces together and say, okay, right now I want to treat you like a database, but when I’m brainstorming ideas or asking you to generate content, let’s set aside the database and start to think more out of the box, a little bit more probabilistically, like, let’s kind of mix it up a little bit.

[00:35:02] Fabio Fiss: At the same time, um, I wanted just to add that, uh, data structure is really important. Huge. Now more, more than ever. I think this is one part that we’ve been bringing up to our clients in the sense that SEO is evolving right as well to meet the demand of the models that are using search to find the right references.

[00:35:25] Fabio Fiss: If you don’t. Treat your data on your website or your application in a way that’s well structured and, uh, with schema markup and in a way that the model can actually understand it, you won’t surface in, in their search right as Google is also evolving their search with ai. Uh, open AI is taking a huge space of, of that, that, you know, market share now over time.

[00:35:49] Fabio Fiss: So, um, structuring your data from an SEO perspective, technically speaking is super important now, so that that will continue to evolve. Yeah.

[00:36:00] Tessa Burg: Well, this has been really exciting, Javier, before we leave. Is there anything, or, I’m sure there is, but what is something that you are looking forward to working on or sort of testing and experimenting with this year?

[00:36:16] Javier López: Well, I think, um, a to a agent, to agent communication ’cause it was announced maybe a couple months ago, but it hasn’t taken the shape really. Uh, the agents still are very controlled by. By humans, like we chat and, and we have to approve their actions. So I think they are going to start popping up agents that are completely let loose and seeing what they are capable of doing.

[00:36:44] Javier López: Uh, for me it’s very interesting. Uh, so, so that’s something I’m very, very excited, excited about to see. Yeah.

[00:36:51] Tessa Burg: What I just you saying that I think might be a surprise to some people listening ’cause there are so much hype around agents. I, that’s why I like keep calling it an agent army is ’cause I feel like some people think there’s this massive army coming for their jobs and coming for all this work.

[00:37:06] Tessa Burg: Yeah. Reality is they’re not quite there and it is important to keep going back, revisiting, testing, and make the most of what’s in front of you. Like so many of the opportunities we see with clients is simply connecting the dots, connecting the data sources within your existing tech stack to help you get more value and elevate your work.

[00:37:26] Tessa Burg: So. I wanted, oh, Javier, do you have something else to add? Yeah, go ahead.

[00:37:30] Javier López: There was something else I’m looking forward to, like they’re starting to build these, these integrations like AI that work with you in the browser. Because there’s so many times that are, you’re using a website and doing busy work.

[00:37:43] Javier López: You have to copy from here, check this out, move this here, fill this in here. And now they’re starting to make agents that are able to, to control your browser. So you can use, Hey, can you help me create a summary from this and put it in my Google document? So you’re gonna start seeing that the, those kind of helpers are very useful and it’s a nice interaction between human.

[00:38:03] Javier López: And and assistant. Yeah. That’s something I’m very excited about. ’cause ’cause I get tired of going to the web and clicking here and therefore doing some things that, that should be automated. Yeah,

[00:38:14] Fabio Fiss: Definitely gonna change e-commerce. Um, um, I’m seeing that very clearly, how it’s gonna change, you know, specific industries in, um, like airlines and, you know, e-commerce, uh, driven experience. So.

[00:38:28] Tessa Burg: I agree. Aaron, anything, any last words, anything you’re excited about?

[00:38:32] Aaron Grando: I’m excited to see, um, if we can get A2A set up on any of the agents that we’ve built. Um, you know, I’m, I’m really, uh, expecting to see that we may start to. Lean into like very specialized agents that are good at a very particular task, um, and use a to a to make those task requests get routed to the right assistant and you know, have them done in a way that’s better than we can get out of our, like, super generalized agents right now, which are awesome, but.

[00:39:09] Aaron Grando: You know, I think there’s, um, there are some things that, um, you just kind of need to put your blinders on as people and get done. And I think the same is the case with agents. Um, agents that are super focused on a particular task are gonna outperform generalized agents, I think, for the next couple of years at least.

[00:39:29] Tessa Burg: Well, with that, we are going to end this episode. Thank you all for participating in the panel. For listeners who want to hear more episodes of Leader Generation, you can find them at Mod Op. That’s modop.com. And if you are inspired by this episode, but still feeling a little overwhelmed, have no fear, reach out to us.

[00:39:51] Tessa Burg: We can help you put together a roadmap, help you maximize the data and tech you already have, and start thinking about how you can upscale, retrain, and evolve your team’s roles and become that company of the future. Um, all right, well, I hope you all have an, a great week and it’s like, I’m, I’ll see you again soon, but I’ll literally just see you in the other meeting today.

[00:40:12] Tessa Burg: So thanks again for making the time to be on the podcast.

[00:40:17] Aaron Grando: Yeah. Thanks Tessa.

[00:40:19] Javier López: Yeah, thank you very much.

[00:40:22] Fabio Fiss: Thank you.