Scott Brinker (3:16)

Well, I actually, I find in the content production pipeline, this is one of the use cases where people are getting pretty good at actually harnessing AI effectively. Now, this doesn't mean like you turn it over to an AI agent and it does the complete end to end, like, oh, concept, concept and production and distribution analysis. And yeah, I went off and I got an espresso and I came back and it was all done. No, but like, each step along those ways we are seeing people productively use AI to like, okay, well, help me brainstorm when I want to do this or help me evaluate this. Okay, all right, now I want to create this thing. Can you take this piece of content and help break it up into different components for, you know, different channels where I may deploy it? Can you help me with the analysis of, you know, customer engagement around it? You know, again, these are all things that marketers could do before, but they were generally very time consuming. And we're seeing a fair number of content production use cases where the human still has to be the one driving each step of that, you know, but the time frame in which they can get those things accomplished has shrunk considerably. So that's a real world win, I guess. The other one I'd say is the chatbots on websites. Now, we've had chatbots on websites for years. We can all agree they've largely sucked this year. They started to get good and they've started to get good for two reasons. One is they're now powered by these LLMs that are actually very good at conversational interaction. And the second thing is companies are starting to get smart about attaching these chatbots, particularly for customer service use cases, to the right data on the back end, their knowledge bases, their ticket histories, their customer profiles. So that it's not unusual now to hear companies that are deploying these, seeing customer interactions with these chatbots getting a 60%, 70% resolution, you know, again, if it doesn't get resolved, you have to make it really easy to get to a human rep who can resolve it. But still, you know, for those cases where it can, like self serve, get the customer the answer they want, get them on their way quickly, and be very efficient for the company. You know, again, this isn't rocket science, but those are two pretty good use cases.

Podcast Host / Interviewer (5:32)

Let me give you an n of 1. I hate the chatbots because, yes, they are lowering costs and they're able to execute 75% of queries successfully and then eventually they bounce to a human. I personally find the experience of talking to an LLM, A, it's very clear when you're talking to an LLM and B, it is very frustrating for the end consumer. It might be more profitable. I'm not sure how good it is for the long term relationship. So as we talk about 2026, I'm curious to see if the chatbot experience is something that gets better or starts to deteriorate better brand relationships.

Scott Brinker (6:11)

No, it's a great thing. You know, I guess for me it probably depends on what the very specific use case is. And I say this as a person, that the most frustrating experience for me is sitting on hold listening to people tell me on a recorded voice how important I am to them. And like, you know, you are 74 in the queue. So to the degree questions I can get answered by a chat bot right away, I'm serving the camp of like, you know, I'll take that path first.

Podcast Host / Interviewer (6:36)

Yeah, and please, we want to give you a survey at the end so you can tell us how much we stink. Yes, I agree that the automated phone processing, I'd actually rather talk to a bot, I think, than type to it. Okay, so you mentioned the content production workflows are getting pretty good. The chat bots are getting all customer facing experiences essentially. Are there any other verticals where you're seeing or you Predict that in 2026 we will see better use of agentic AI?

Scott Brinker (7:10)

I think the back office use of analysis of data. Right now actually a lot of these AI engines are very capable of doing a good job with that. The reason they're not used very much for that is because we aren't connecting the right data to them or the data we have is surprise, surprise. This is a theme that's been in martech. Since there's been martech, a lot of our data is just not very good, you know, but this has become one more reason for marketers to invest in. Like, you know, we really got to get our data layer right because everything we want to do and AI is just sort of the latest incarnation of this. Everything we want to do is kind of predicated on having good data and being able to get that data to the right place at the right time. I'm not saying that's easy, but at the end of the day, that really is the mission if you want to harness what this technology makes possible.

Podcast Host / Interviewer (8:02)

So if you looked back at 2025 and had to summarize in one sentence. I'll even give you a paragraph. What is the big takeaway from the adoption of AI? What did we learn this year that will be applied moving forward?

Scott Brinker (8:19)

Hmm. So many different things we could pick on that. I would say this. Um, we are, we are coming to grips with the fact that AI is very, this sort of generative AI, this generation of AI is very different than the kind of automation we've had in the past. You know, we've had automation and marketing for, you know, decade, decade and a half. But it's been what they call deterministic. You know, it's you, you explicitly configure if this, then that and this other thing is great. It's very reliable, predictable, explainable. It also tends to be very fragile. We tend to create all sorts of complex exceptions and whatnot and quickly can get a mess. On the other hand, we're starting to see, as we experiment with incorporating these LLMs into our workflows and our automations, it's almost like the inverse trade offs. Okay, LLMs are not as predictable as that goes. They have variants to them. They are what we call non deterministic, you know, in how they're going to respond. And that has challenges, but it also has a set of benefits that actually these LLMs are better at adapting. You can throw things at them that would have broken one of those deterministic workflows of I'm expecting X, you didn't give me X, what do I do with it? You can throw more things at an LLM. You can throw things at it that are going to change and the LLM can adapt to it. And so where things are getting really interesting is you're seeing marketers here in 2025 and I think we'll see a lot more of this in 2026. They're starting to combine the two. They're like, okay, we're going to have a workflow and that overall workflow is going to be deterministic and very well structured because that's what we need to run our operations as a business. But we're starting to incorporate the use of LLMs with specific steps along that workflow because it allows us to do things with like unstructured data or you know, generation of media or things like this that were impossible to do with those deterministic workflows. We've started to experiment with that in 2025. I think we'll see a lot more of that and start to move it more into production in 2020.

Podcast Host / Interviewer (10:31)

6 My takeaway from this as we look back at 2025, it was the year where we moved beyond essentially content marketing into conversational marketing, and now we're making the migration into analysis using LLMs. And as you said, we have had marketing automation for 15 years or so, but now we're able to move beyond deterministic rule sets to have something that is a little bit more related to judgment. And that wraps up this episode of the martech Podcast. Thanks again to Scott Brinker, the godfather of the Martech industry, for joining us. If you'd like to contact Scott, you could find a link to his LinkedIn profile in our show notes or on martechpod.com you can always visit his website, which is chief martech.com it's chief M A R T E C dot com if you haven't subscribed yet and you want a regular stream of marketing and technology knowledge in your podcast feed, hit the subscribe button in your podcast app or on YouTube and we'll be back in your feed every other week. All right, that's it for today, but until next time, my advice is to just focus on keeping your customers happy.

Scott Brinker (11:49)

Foreign.

Benjamin Shapiro (12:02)

Thanks for listening to the Martech Podcast and I hear everything. Production Looking to launch or scale a podcast like this one for your brand? Then visit iheareverything.com.