What if your AI assistant could do more than just generate content? Imagine an AI agent that works across your entire marketing workflow, enhancing both creativity and productivity—are you ready to transform your marketing team?
I am here at Opticon 2024 in San Antonio Texas and getting the opportunity to see a lot of inspiring ideas from some of the world’s leading brands and hearing all about Optimizely’s platform and how it enables 1:1 personalization, streamlined content operations, and incorporates the latest generative AI features.
Today we’re discussing how AI agents are revolutionizing the marketing workflow with Kevin Li, SVP of Product Strategy & Operations at Optimizely. We’ll explore the concept of AI agents, their role in enhancing creativity and productivity, and why marketers should be excited about them.
Resources
Optimizely website: https://www.optimizely.com
Wix Studio is the ultimate web platform for creative, fast-paced teams at agencies and enterprises—with smart design tools, flexible dev capabilities, full-stack business solutions, multi-site management, advanced AI and fully managed infrastructure. https://www.wix.com/studio
Don’t miss Medallia Experience 2025, March 24-26 in Las Vegas: Registration is now available: https://cvent.me/AmO1k0
Use code MEDEXP25 for $200 off registration
Register now for HumanX 2025. This AI-focused event which brings some of the most forward-thinking minds in technology together. Register now with the code “HX25p_tab” for $250 off the regular price.
Connect with Greg on LinkedIn: https://www.linkedin.com/in/gregkihlstrom
Don’t miss a thing: get the latest episodes, sign up for our newsletter and more: https://www.theagilebrand.show
Check out The Agile Brand Guide website with articles, insights, and Martechipedia, the wiki for marketing technology: https://www.agilebrandguide.com
The Agile Brand podcast is brought to you by TEKsystems. Learn more here: https://www.teksystems.com/versionnextnow
The Agile Brand is produced by Missing Link—a Latina-owned strategy-driven, creatively fueled production co-op. From ideation to creation, they craft human connections through intelligent, engaging and informative content. https://www.missinglink.company
Transcript
Greg Kihlstrom:
Welcome to Season 6 of The Agile Brand, where we discuss marketing technology and customer experience trends, insights, and ideas with enterprise and technology platform leaders. We focus on the people, processes, data, and platforms that make brands successful, scalable, customer-focused, and sustainable. This is what makes an Agile brand. I’m your host, Greg Kihlstrom, advising Fortune 1000 brands on Martech, Marketing Operations, and CX, bestselling author and speaker. The Agile Brand Podcast is brought to you by TEKsystems, an industry leader in full stack technology services, talent services, and real world application. For more information, go to teksystems.com. Now let’s get on to the show. What if your AI assistant could do more than just generate content? Imagine an AI agent that works across your entire marketing workflow, enhancing both creativity and productivity. Are you ready to transform your marketing team? I’m here at Opticon24 in San Antonio, Texas, and getting the opportunity to see a lot of inspiring ideas from some of the world’s leading brands, and hearing all about Optimizely’s platform and how it enables one-to-one personalization, streamlined content operations, and incorporates the latest generative AI features. Today we’re discussing how AI agents are revolutionizing the marketing workflow with Kevin Li, SVP of Product Strategy and Operations at Optimizely. We’re going to talk about the concept of AI agents, their role in enhancing creativity and productivity, and why marketers should be excited about them. Kevin, welcome to the show.
Kevin Li: Thanks, Greg, for having me. Good to see you here again at Opticon.
Greg Kihlstrom: I know, returning champion here. Absolutely, we’ll get you a plaque or something if you come back next year. Nice, nice. So let’s, before we dive in here, and you know, certainly lots to talk about, why don’t you tell us a little bit about your role, which I think is a little bit of a new role for you, and a little bit about your background.
Kevin Li: Yeah, so I lead up product strategy and operations here at Optimizely. A little bit of background, I joined a company five years ago now, I can’t believe it’s been that fast with all that, through an acquisition. And I took over product strategy, which covers M&A, corporate development, acquisitions, but also our portfolio strategy. So we obviously grew as a company from 100 million to 400 million, that was a public press release. And so looking after the portfolio strategy. And on the operations side, I also have product analytics, documentation, competitive intelligence as well. And yes, I appreciate that I recently got promoted to roughly the same role, just a slightly bigger title, I suppose. Congrats. Thank you. Thank you.
Greg Kihlstrom: Well, yeah, let’s dive in here. Again, lots to see and hear in San Antonio here. I wanted to focus this conversation on AI agents, and I’ve talked about them a little bit on the show. I had somebody from Microsoft to talk about some of it, talked with a few others, but I wanted to focus on really specifically the role of AI agents in marketing. Optimizely recently announced the launch of AI agents embedded within the Opal platform. First of all, maybe for those a little less familiar, what exactly is Opal? And then can you talk about what’s an AI agent in this context of marketing?
Kevin Li: Yeah, so Opal was essentially our AI co-pilot that we launched last year at Opticon. I think you were here covering that piece of it. So Opal is meant to be the interface that marketers and users essentially access a lot of the generative AI features across the board. So last year we had half a dozen, 10 plus features or so. This year we’ve doubled, tripled that, etc. AI agents, importantly, will be accessed through Opal. So think about Opal as the interface, if you will. That’s why we sort of gave it a little bit of a personified identity. And then behind that is where the capabilities such as agents, retrieval augmented generation, some of the more sophisticated capabilities will actually sort of come in through Opal itself for users. And then on the agents piece of it, I think this is a super exciting moment for sort of technology. It’s changing super quickly. Last year we were talking about like, oh, let’s embed generative AI into the features itself. And this year we’re now talking about agents. So the biggest difference is really the level of autonomous execution that can actually happen. So before, if you think about, you know, without AI agents, what happens is software vendors like us, sort of purpose-built use cases. So if a user is doing something like, I’m sure you’ve tried to hold like, oh, generate a content, write a headline, generate an image, etc. It’s sort of the task is very specific. And there’s not much sort of autonomous execution because it’s sort of input, output, right? You tell it to do something, you get something in return. Agents is a little bit different in that you typically are describing things in terms of the outcomes that you actually want to accomplish. And the agents are executing something sort of on your behalf. So there’s a workflow automation piece of it. There’s a little bit of more intelligence. And borrowing sort of a metaphor from autonomous driving, there’s sort of the six levels of autonomous driving. So what we’re also seeing in sort of marketing and MarTech in general is the degree of autonomous execution kind of going up as well. So agents is basically a big step forward to say, Can I have one agent do something, or maybe a cluster or group of agents do something, or an army of agents go do something? So that’s the path that we’re basically on.
Greg Kihlstrom: Nice, nice. Yeah, because that’s kind of the limitation when you’re just like prompt to prompt is things get very, you know, you can move forward, but that automation is key. Since its introduction, the Opal platform has seen a 500% increase in adoption. We’ve all seen rapid adoption numbers out there or whatever, but even with the rapid adoption of AI in general, that’s pretty impressive. What do you think has fueled that growth, and are there areas within that growth that are particularly notable?
Kevin Li: Yeah, I think for us, number one, it’s great to see. As part of my job, I run product analytics, so the adoption numbers we see on a day-to-day basis, it’s interesting. A year or two ago, when we first launched some of this generative AI stuff, it’s very much a I don’t know if you remember that time, it was a wild west, so people were just sort of like throwing things out. Does anyone want to use this? Is this even a thing they want to do? Now when we see this adoption, I think an important point is also that we’re seeing sort of sustained adoption. So people are trying it, but they’re also sort of sticking with it. So we’re sort of seeing behavior change that actually sort of tends to stick rather than like, you know, you try something and like, I’m not really going to use it the next time. That’s an important sort of part about the growth that’s pretty notable. I think the other part of it is for us as Optimizely, we sort of have three different parts of the pillars sort of powering this stuff. So the first part is we actually have our customers’ data. Well, they put their data in our system. So that’s their, we obviously it’s, we have to get their permission to use it. So they have to explicitly say, hey, this is okay. The second part is we have, the workflow itself. They already do work in our platform, right? So there’s no swivel chair of like copy and paste something or going somewhere else, et cetera. It’s already sort of part and parcel for their day-to-day tasks. And then the third thing is, as a sort of digital optimization, digital experience provider, we also have the experience. We know what the sort of eventual outcome sort of looks like. So those three things, the data, the workflow, and then the end resulting experience piece of it, all in one place is kind of how we think about why marketers are like, well, honestly, it’s just easy. We’ve lowered the barrier to adoption. You don’t have to sign up for another account and then get legal approvals on whether or not you can use it or not and move your data to a bunch of different places, et cetera, right? So if you think about you’re starting your AI stack from scratch, that’s a lot of pain and effort. Whereas if you’re already on Optimizely, you go, oh, actually, I can get all of it already where I work today.
Greg Kihlstrom: Yeah, because I mean, otherwise you’re like copying and pasting out of one thing into another. And as you would imagine, it’s very disconnected, right?
Kevin Li: Yeah, I think as someone and you know, you write a lot, right? And so if you were to sort of take a very simple metaphor as well, spell checking is a form of AI. But how painful would it be for you if you don’t get the red squiggly lines to tell you that you’ve misspelled something and you literally have to copy and paste a paragraph to some other browser to spell check for you, correct it, and then you have to paste it back in. You just wouldn’t use it. So it’s the embedded in the workflow. It’s right there as you type, you’re like, oh, I typoed, let me fix something. That’s AI sort of right there and there and has access to your data because it’s looking at what you’re typing. good metaphor to kind of really think about adoption in general. So it has to be where the marketer is already working today.
Greg Kihlstrom: Yeah, and rightfully so. There’s been a lot of focus on content creation and generating text images, all of those kinds of things. I think the workflow stuff, some of that comes from I do a lot of work in process and marketing ops myself, but I think the workflow and productivity angle is really powerful when combined with those as well. So, you know, why is introducing automation into the workflow and delivery components just as important as some of those creative aspects?
Kevin Li: Yeah, because I think at the end of the day, if you think about a marketer’s workflow, they don’t just want to create. So when we kind of think about the outcomes that they want to do, creation is sort of one aspect of it. It’s sort of like one pillar, right? But they also want the AI to not just say, hey, help me write something or create an image. They also say, Help me do something. So help me, remind me if a task is due. Create for me an entire campaign out of something. Go execute, go do something for me. They also want AI to think about analyzing some data. Tell me, surface some anomaly for me. Summarize a bunch of data and extract the key points so I don’t have to go write a bunch of SQL queries to figure out what’s actually going on. And then the last one being, help me make a decision for execution, for personalization. Decide for me what’s going on on the fly. Historically, we’ve always thought about, and rightfully so, sort of the first use case was always content generation, sort of create. But if you think about the, well, help me create, help me do, help me analyze, and help me execute, it’s multifaceted in terms of what it is. So we’re definitely going way beyond just creation into the sort of four aspects. And I’m sure there’s even more aspects of that, of the overall workflow.
Greg Kihlstrom: And so within that as well, you know, there’s a lot of opportunities. There’s also some concerns that, you know, brand managers, marketers, creatives, all of the above want to make sure that this stuff that’s automated and created is also compliant, you know, brand compliance and consistency. I think that’s another argument of why you don’t copy and paste stuff out of a million different apps into each other and stuff. I think it, That can hurt that as well, but you know, how do you look at brand compliance? And you know when we’re talking about these AI agents, there is a little bit of there’s some autonomy there So, how do they stay brand compliant?
Kevin Li: So so that’s actually really interesting because AI agents can actually help brand compliance. When people think about autonomous, it doesn’t mean complete freedom. I want to really clarify that. Autonomous doesn’t mean complete freedom. Autonomous means they can do something without supervision. That’s actually a very big delineation. That’s huge. It’s like if they repeatedly do something to your expectation without supervision, that’s still autonomous. Whereas I think what people are fearful of is going rogue. That’s the Terminator scenario. Exactly. But I think that delineation is really interesting because AI agents, I think to the contrary, can actually help with brand compliance. Because what is brand compliance? It’s someone looking at something to decide whether or not it passes the test. And so if you’re able to train an AI agent to do that, that’s actually going to help brand compliance. It may actually reduce sort of human-based errors to do some of these things. And obviously, brand compliance is a legal compliance, compliance in general, right? What must you say? Do you have to write disclaimers? Do you have to write accessibility things, et cetera? But each of those things over time are going to be sort of perfected by agents. And so that delineation between autonomous and going rogue are two separate concepts. Autonomous doesn’t mean necessarily that.
Greg Kihlstrom: You touched a little bit on helping marketers make better decisions, getting better results, but I’d like to go back to that a little bit as well. How do you look at AI agents supporting both strategic support as well as practical supports, especially in, we’re not just talking about a single channel, we’re talking about omni-channel, we’re talking about personalization, all of those kinds of things. So how do they support in the complex marketing world?
Kevin Li: Yeah, I think that’s then one of the advantages of sort of us providing AI agents as part of Optimizely 1. So the core principles of Opal is it works across Optimizely 1. So it doesn’t matter where you are. I’m sure you’ve seen our sort of architecture of plan, create, globalize, store, et cetera, personalize, experiment, analyze. So our intent is it doesn’t matter where you are within that overall marketing lifecycle, marketing workflow, you can use Opal. And the key thing is, it’s not sort of just a use case driven within the specific task. So if you’re in planning, like, well, what if you wanted to generate some variations for experimentation? You can do that. If you’re in publishing, but what if you want to actually sort of look at analytics from previous results to see, hey, is this even a good idea to publish or should we make some tweaks to it, et cetera? So the ability to actually traverse across the activities, that’s sort of a core strategic pillar. The second thing that’s really strategic for us is the ability for, like I mentioned earlier, for customers to choose to use their specific data that’s already in our system to help them go do something. So a good example being. If you’re in our content marketing platform and we see that you’ve started a campaign every single year around a certain time, maybe it’s Black Friday, maybe it’s Christmas, maybe it’s back to school, etc. The AI should be smart enough to say, hey, I’ve seen you do something, let me proactively suggest something. So making use of your data. in a right way. And then the last one is actually sort of traversing. It shouldn’t just be creating. Can you help me decide? Can you help me analyze? Can you help me go do something and execute? So those are the three core things. And that’s really kind of our guiding principles, if you will, for somebody obviously on safety, governance, compliance, et cetera.
Greg Kihlstrom: Yeah, yeah. One quick follow up to that. I mean, I think one of the bigger gaps that I see out there with marketers is the time to do the analysis and get the insights. And, you know, so like what you’re touching on here of it being easy to switch back and forth to get the insights so that you can make better decisions instead of, I think, When it’s a lot of manual work, you just get stuck in this, like, OK, we’ve got to launch the next campaign. We don’t have time to look at what happened last year, last month, whatever. So that kind of speed to insights and creation, it seems like the all-in-one aspect of this is really powerful.
Kevin Li: That’s the practical side, which is why we’re also seeing the 500% increase in adoption. It’s the, how do we bring AI and the benefits of agents and all of that to the fingertips of the marketer as they’re doing their job, not make marketers go learn something else completely separate and then figure out how it all ties together and then worry about, hey, someone from legal not gonna be not okay with this, et cetera. So that’s really the thing. It’s like, bring the tools to the humans so they can get more work done.
Greg Kihlstrom: Yeah, yeah. So before we wrap up here, I wanted to touch on just two other things quickly. Lots of big announcements this week, but Optimizely recently announced the acquisition of NetSpring. I’m going to have Vijay, the founder and CEO on the show as well, but I wanted to get your take in your purview as product strategy and all. How does this fit into the Optimizely strategy and how do you think it’s going to be able to benefit what brands are able to achieve?
Kevin Li: Yeah, for us, and this is a super exciting announcement for us, I think it’ll pay off huge strategic dividends for us going into the future. So there were two things that we looked at when we were thinking about this space. So the first one is the exponential growth of data warehouses in general. So we’ve been talking about the single source of truth for data for Yeah, it’s been a while. It’s been a while, right? But I think now enterprises have finally gotten their act together. They have enterprise data architecture. They have unified data dictionaries. The maturity of data management and governance is night and day versus five years ago, 10 years ago. And you also see the growth of data warehouses in general. Snowflake, BigQuery, Amazon, Microsoft, et cetera. Everybody who’s somebody in tech has a warehouse offering. And enterprise is starting to put sort of data there. The second thing is, what data are they actually putting in there? It’s actually the most sort of consequential data that a company actually has. It’s revenue data. If you’re e-commerce, it’s where you track your return rate. If you’re a financial services firm, it’s where you actually, how many people started a credit card application? How many people got rejected? How many people got approved? How much credit utilization are they using? All of that data sits there and guess what? They don’t ever want that data to ever leave their own data warehouse. And so, actually very similar to kind of what I just talked about with AI agents, which is let’s bring AI agents to the marketer. NetSpring really means, give us the means to connect what we’re doing on the optimization side. So, take experimentation for instance, and connect it to those specific business metrics that exist inside of a customer’s data warehouse. So, how do we tie it into that? And so, when we kind of think about that, historically when you do optimization, Maybe it’s a simple A-B test, just as an example. You try A, you try B. The metric you’re tying to is oftentimes the observable metric. So for e-commerce, a very simple example being, did you actually add to cart? Because that’s what the experiment can see. Does the CFO care about Add to Cart? Yes. I’m not going to be like, no, they don’t. But what do they actually really care about is return-adjusted revenue. How many people bought something and kept it and didn’t actually return? So if you follow e-commerce, returns is super expensive, tons of waste, tons of fraud, lots of issues, right? So if you’re able to run an experiment and say, oh, we actually drove up return-adjusted revenue, That metric is inside of the warehouse, so how do we connect it? So that’s essentially our sort of first horizon on this. And in longer term, we see sort of this architectural shift. Last year, we sort of announced the launch of SaaS CMS. We said, hey, people want to go at CMS from platform of service to software service. On the analytics side of it, we’re also seeing big trend of saying people want to start thinking about moving from hosted analytics on the vendor side to putting in an analytics and visualization layer on the warehouse directly because all that data is already there. Why make two to three copies of the same data? Why have data move back and forth? Why take the risk of sensitive data leaving your warehouse? So that’s essentially what we’re thinking about for the longer term on that second horizon.
Greg Kihlstrom: Yeah, yeah, love that. Well, yeah, looking forward to seeing the trajectory there. And then last thing, the Google announcement, Google Gemini announcement. Do you mind talking a little bit about that?
Kevin Li: Yeah, for sure. So we’ve been a customer of Google for the longest time. So our experimentation platform sits on BigQuery. Specifically, we had a PR announcement about that. We’ve also been working closely with Google across the board. So when they announced the sunset of Google Optimize, which is a A-B testing solution, we were one of three partners selected. So we migrated a bunch of customers over. We have sort of good integrations there as well. And then obviously then when we looked at, you know, we’re not in the business of building our own LLMs, right? So when we go look at the marketplace, it’s going to be leveraging sort of someone else with that itself. So when we looked at sort of our overall tech, our overall architecture, And also the existing relation with Google, it’s kind of a no-brainer to sort of double down on specifically that relationship. So they’ve always done a lot with, you know, Vertex, Gemini, launching sort of new and new versions there. So that’s essentially sort of what’s also powering our AI agents underneath the hood.
Greg Kihlstrom: Great, great. So one last, well, two last questions for you here real quick. So just following back on AI agents, certainly lots of exciting things today and announcements this week and things like that, but where do you see this going? What should marketers keep an eye out in the maybe in the months to come?
Kevin Li: Yeah, I think the key thing is we’re still in the accelerating phase of innovation. So things are still speeding up and not slowing down. We’re sort of nowhere near this comfortable thing where people are like, oh yeah, we’re not in the CRM space, for example. And it’s like, yeah, we might want to track sales pipeline. This is still like, we’re in the early days. And so I think the important thing is really for marketing leaders to give it a try. That’s how you’ll learn about this. That’s how you’ll start to see and reap the benefits. And going back to that 500% adoption stat, that’s what we’re seeing too. It’s those customers are trying, and then they’re starting to see the subtle sort of bottoms up behavior change to say, I can actually get something, be more productive, do more with less, switch to less tabs, just very concretely, do less swivel chair type of things. I think that’s sort of the key thing because that’s the only way, I mean large organizations a lot of times it’s change management is hard. countless books on change management. So it is about really getting into this sort of AI mindset. And that’s in part why, you know, here as well, we have the marketer’s playbook for AI to try to get them to kind of think about it. And at various Opticons, we have a masterclass on AI. It’s like, hey, here’s how you got to think about prompt engineering, right? So really being more educated on this space, I think that’s the key thing.
Greg Kihlstrom: Yeah, great. Well, Kevin, it’s always great to talk with you. Last question here, I like to ask all my guests, what do you do to stay agile in your role and how do you find a way to do it consistently?
Kevin Li: I think for me, it’s really about staying curious. If you think about the NetSpring acquisition as sort of a good example, if you sort of look at that space, there’s really only four or five very early stage vendors who even do warehouse native analytics in particular. And so that’s not a widely covered sort of category itself, but staying curious, thinking about connecting the dots together, etc. I also have a decently large network of people who are like, hey, what do you think about this? What do you think about that? Probably spent too much time on LinkedIn. Same here. People posting things and being like, oh, that’s actually super interesting. Keeping it on a startup community, et cetera. I think that’s great. And then I think in my role, we just get so many almost like inbound M&A queries. It also actually really helps because a lot of times it’s like, oh, that’s interesting. Let’s at least take an initial call and we get to ask questions. how do they see their existing category, are there any systemic changes, etc. It’s just about keeping the pace and the frequency of information flow, really.
