Agentic Marketing: The AI Operating System for Modern CMOs

02:27:53.760 –> 02:28:02.790
Julia Nimchinski: Thank you so much, Mary. Thank you so much to all of our panelists, and we are transitioning to our next panel. Welcome to the show mark organ.

893
02:28:03.180 –> 02:28:05.399
Mark Organ: Delighted to be here once again.

894
02:28:05.820 –> 02:28:10.480
Julia Nimchinski: We’re delighted to interview what’s new.

895
02:28:11.820 –> 02:28:30.582
Mark Organ: What’s new? I mean, what’s I mean, it’s incredible. The speed at which things are moving in a gentic AI which is what we’re gonna talk about today. You know, I’ve been seeing now a number of companies who are, you know, replacing, you know, entire functions with with agents with

896
02:28:31.030 –> 02:28:33.339
Mark Organ: various levels of success.

897
02:28:33.874 –> 02:28:49.520
Mark Organ: But yeah, I mean, it really feels like marketing, especially sort of prospecting is ground 0 for for how AI is being used in in the, you know, in technology companies. So

898
02:28:49.970 –> 02:28:51.800
Mark Organ: it’s an exciting time.

899
02:28:53.020 –> 02:29:01.800
Julia Nimchinski: Definitely so excited for this. Let’s do a quick introduction. So we have leading Cxos technologists. Cmos

900
02:29:02.070 –> 02:29:04.710
Julia Nimchinski: Amanda. Welcome to the show again.

901
02:29:05.120 –> 02:29:07.780
Amanda Kahlow: Hi! Thanks for having me happy to be here.

902
02:29:09.520 –> 02:29:15.089
Julia Nimchinski: Happy to feature you. Kingston, let’s start with you.

903
02:29:16.040 –> 02:29:21.300
Kingston Duffie: Hi! I’m Kingston Duffy! I’m the founder of Navoo. This is my

904
02:29:21.680 –> 02:29:35.080
Kingston Duffie: 4th startup in Silicon Valley. So I’ve been around the block a few times. AI really looks like a big one. You know my, my career in Silicon Valley started with the Internet back in the nineties. And this one feels just as big.

905
02:29:38.410 –> 02:29:39.730
Julia Nimchinski: Amazing mark.

906
02:29:39.950 –> 02:29:41.000
Julia Nimchinski: It’s her show.

907
02:29:41.880 –> 02:30:01.009
Mark Organ: It’s awesome, that’s great. Well, you know, that one thing that’s really cool about this panel is that we do have people who have been through at least a couple of different revolutions. You know, in terms of marketing Tech and Silicon Valley. Me personally, I I founded Eloqua back in late 1999, one of the 1st

908
02:30:02.828 –> 02:30:09.279
Mark Organ: I guess hosted multi-tenant status companies out there found just a few months after salesforce.com.

909
02:30:09.380 –> 02:30:12.399
Mark Organ: And I’ve seen lots of interesting things in my time.

910
02:30:12.839 –> 02:30:28.389
Mark Organ: And it is interesting to see, you know how the AI revolution in marketing, how it’s how things that are similar to what’s come before, but also some things that are really, really, radically different. And I definitely want to explore

911
02:30:28.778 –> 02:30:43.530
Mark Organ: that today, I guess we’ll have people introduce themselves when when they answer their question. Maybe let’s just start by saying, What is the gentic marketing? And you know, how is it different from what has come before?

912
02:30:44.110 –> 02:30:49.340
Mark Organ: Love to hear from? Maybe, Chris, maybe. Why don’t you jump in on this one.

913
02:30:49.600 –> 02:30:57.540
Chris Hicken: Yeah, I mean, maybe we’ll start with, yeah. Agentic. AI, I guess, for people who are joining for the 1st session, we maybe define

914
02:30:57.700 –> 02:31:19.280
Chris Hicken: what we mean by agentic AI, which is, it’s kind of the precursor to Agi. So we started with AI. Now we have a gentic AI, and we’re on the path to Agi, and we can fake some Agi stuff with with AI agents. And so whereas now we might say something like, Write me a subject line. You might say that to the AI

915
02:31:19.580 –> 02:31:45.129
Chris Hicken: with a gentic AI, you might say something like run a reactivation campaign for users who ghosted us in the last 90 days, and when your AI is set up correctly it will pull the right audience. It will draft the email flow. It’ll test subject lines. It’ll send messages. It’ll monitor performance, and then it’ll report back to you how it’s done. So that’s the transition we’re making from. From

916
02:31:45.560 –> 02:31:49.090
Chris Hicken: you know what we had before, which is like chat, gpt into into agents.

917
02:31:51.530 –> 02:31:54.890
Mark Organ: Right. Someone else gave a counterpoint on that.

918
02:31:55.480 –> 02:32:21.349
Amanda Kahlow: Yeah, I’ll jump in. So yeah, quick intro for myself, Amanda Calo, CEO, and founder of one mind. I was the former CEO and founder of sixcents. I like to say I started sixcents to find buyers, and one mind now to close them, so I have unfinished business for the b 2 b go to market space. When I think about where we are today with going agentic. Everybody’s talking about agentic.

919
02:32:21.370 –> 02:32:36.939
Amanda Kahlow: you know. Kind of a little bit of a spicy take. But I feel like agentic is yesterday’s news in a lot of worlds. It’s the simple, task-based workflows. And Chris, as you mentioned, really the where we’re going is to the place of Agi. And what is the definition of Agi? If you talk to Sam Altman, it can replace

920
02:32:36.940 –> 02:33:01.939
Amanda Kahlow: workers and do as well across a specific job as a human. I think we all think of Agi as having the Eq. Plus the IQ. As well, and the emotional intelligence and the relationship building. And there’s other pieces that you know, we’re not 100% there yet. But in the context of sales and marketing, is it? Can it carry on a conversation, engage buyers, meet them when they’re active and ready, and give them what they’re looking for, to drive

921
02:33:01.940 –> 02:33:20.439
Amanda Kahlow: your the end goal, which is pipeline and close one business and doing it in a delightful way. And I think that’s like, ultimately we are at Agi and Gtm. If we can do it in a way that really serves buyers. And if we serve buyers, we’re going to serve our own, our own needs of growth and efficiency.

922
02:33:21.530 –> 02:33:29.169
Mark Organ: Yeah, no, I think that’s a great point. That’s been a touchstone for me in my career. In terms of you know, great marketing actually delivers a great buyer. Experience.

923
02:33:29.643 –> 02:33:44.779
Mark Organ: And I don’t know. Is, is AI really doing that yet? Are AI agents actually delivering a superior buyer? Experience that I. Personally, I feel the answer is, no, but I would love to hear, maybe someone who believes that yes, and a no

924
02:33:45.261 –> 02:33:49.240
Mark Organ: on. Are we actually delivering great experience, yet with AI agents.

925
02:33:50.200 –> 02:33:51.679
Amanda Kahlow: Oh, I’m a fuck! Yes, sorry.

926
02:33:53.230 –> 02:33:58.760
Zack Holland: But and it’s, you know, through our superhumans. But I’ll leave it at that, and it’s far better than what a human can do.

927
02:33:58.760 –> 02:34:18.159
Chris Hicken: Well, Amanda and I agreed ahead of time that we would intentionally disagree with each other on things to make it interesting. So out of the box. AI is terrible at having conversations with people, the way that a human would it actually is truly terrible. You know, in our platform we’ve had to do an incredible amount of

928
02:34:18.440 –> 02:34:39.249
Chris Hicken: put in place an incredible number of controls to force the AI to act in a specific way that feels more natural the way that we would interact with each other as humans. And so you can definitely get it better. I still wouldn’t say that it’s an excellent replacement for a human when you’re doing kind of AI conversations.

929
02:34:39.250 –> 02:35:00.740
Chris Hicken: you could say that it’s very good. I think you can get it to very good at this stage, but also with very good. You’ve made the AI. In some ways you’ve taken away some of the creativity and flexibility of the AI, and you’ve had to put in place a lot of constraints that tell the AI exactly how it can behave in certain situations, and so it feels to me it feels more like

930
02:35:00.740 –> 02:35:10.579
Chris Hicken: traditional programming than it is letting the AI do its own thing creatively. So that’s my take at least my experience in building my AI Startup.

931
02:35:10.950 –> 02:35:14.477
Kingston Duffie: Yeah, I I promote it a different way, which is

932
02:35:15.780 –> 02:35:24.250
Kingston Duffie: in the either. Or if you think my job is to replace a human, I agree entirely. But if you think about marketing.

933
02:35:24.350 –> 02:35:28.360
Kingston Duffie: marketing’s job is to talk to the masses, and if you have

934
02:35:28.710 –> 02:35:32.309
Kingston Duffie: tens or hundreds of thousands of visitors coming to your website.

935
02:35:32.430 –> 02:35:42.380
Kingston Duffie: And what you’re giving them today is a thousand page user’s Manual for your company. That’s a bad experience. And the question is, can AI do a better job

936
02:35:42.670 –> 02:35:47.309
Kingston Duffie: than that for the buyer, and I think the answer is absolutely yes.

937
02:35:49.060 –> 02:35:49.820
Chris Hicken: That’s a good point.

938
02:35:49.820 –> 02:36:18.110
Amanda Kahlow: I couldn’t agree with that more. It’s bringing content to life it’s having. This is why the conversational side works so well is that especially of a deep, complex products. You have to navigate the whole site and try to find your way. That is atrocious for a buyer. Right? So just meet me where I am and have a conversation. And, Chris, you’re right. I agree with you that you do have to put it on strong guardrails and out of the box trying to do it yourself. Good luck, like really hard to do and do well and not hallucinate. Yeah, but I totally agree with that.

939
02:36:18.110 –> 02:36:36.190
Chris Hicken: I guess Kingston I come from. So I built a company called User Testing, which is all about really deeply understanding the user’s experience. And so I guess I’m probably a little bit more sensitive to the experience that we deliver. And so you probably could say like, Yes, AI does a better job than just dumping someone onto website and letting them figure it out on their own.

940
02:36:36.240 –> 02:36:54.760
Chris Hicken: It helps them find information. But I guess I’m sensitive to the quality of the experience. Like, if we if you deliver an experience that’s just okay, or a user has to ask AI several times to do something, they’re just going to give up. And so, at least from my perspective, the quality of the experience still has to be very, very high.

941
02:36:54.970 –> 02:36:59.600
Kingston Duffie: Yeah, so don’t want to dominate here. But I do want to say

942
02:36:59.920 –> 02:37:26.159
Kingston Duffie: there’s a lot of hype in the AI space right now. And yet what is the danger is that we are going to miss? You know we’re going to throw out the baby with the bathwater. The AI is solving a lot of really big problems right now. And if we set the bar too high, if we set the expectations so that we’re at an Agi level way before we’re ready. We’re just going to be leaving value on the table. So my feeling is.

943
02:37:26.190 –> 02:37:39.719
Kingston Duffie: let’s keep moving that bar up and move it up aggressively. But where we are right now, I think we’re at a place where we’re, you know, in our own product. What we’re seeing is people are coming in and using an AI on a website.

944
02:37:39.950 –> 02:37:40.180
Mark Organ: Okay.

945
02:37:40.180 –> 02:37:50.184
Kingston Duffie: Just having, like, I’m getting the answers. I actually had it to the questions I actually had. That can make a huge, positive difference. It doesn’t need to be

946
02:37:50.890 –> 02:37:58.900
Kingston Duffie: a full replacement for the entire sales and marketing team that’s behind that. I still think

947
02:37:59.210 –> 02:38:09.030
Kingston Duffie: AI is very early in the creativity side of the marketing business, but it’s very much there in terms of the content delivery side of the business.

948
02:38:09.930 –> 02:38:39.450
Zack Holland: Yeah, I mean, I would just to pop in. I completely agree. My name is Zach Holland. I’m the founder of a company called Avery AI, where we have what we call the AI marketing manager. So an agent to take care of a lot of the things you use agents for. So on the strategy and content creation side. But then a marketplace of human experts to fill the gaps where the AI can actually match to the right person. So obviously, my, my standpoint is that there’s still hundreds of use cases where you still need human expertise, and we will for a lot longer than a lot of companies advertise the idea of like

949
02:38:39.450 –> 02:38:52.090
Zack Holland: Agi replacing marketing. But I also think marketing is a really big, broad term when we talk about marketing. Yes, AI is better at analyzing a massive amount of information and spitting the right things back already than a human.

950
02:38:52.130 –> 02:38:53.010
Zack Holland: But

951
02:38:53.270 –> 02:39:19.420
Zack Holland: under the marketing umbrella you have creative brand direction you have. You have media buying. You have all these places where agents are still kind of dropping the ball, and it’s definitely, rapidly improving. But I think if we’re talking about first, st AI was to replace a task and then agentic AI is to replace more of a workflow. Then Agi is to replace a role. I think we we have a significant amount of time until you can really pull people out of roles that weren’t super basic. To begin with.

952
02:39:20.260 –> 02:39:21.010
Zack Holland: Amanda.

953
02:39:23.646 –> 02:39:24.870
Chris Hicken: I think it. Yeah.

954
02:39:24.870 –> 02:39:25.210
Mark Organ: And then.

955
02:39:26.190 –> 02:39:40.069
Mark Organ: yeah, it it does. It does feel like we’re still where we’re narrowly defining tasks. You know, in in order in order for the technology to really work, at least so far. But things are. Things are changing quickly.

956
02:39:40.740 –> 02:39:45.970
Chris Hicken: Megan Megan, you’re surprisingly silent on this topic. I think you have a strong opinion here.

957
02:39:45.970 –> 02:40:14.150
Meagen Eisenberg: I mean, I certainly do. I’ve always been an advocate for sophisticated tech stacks for marketing. I think it’s when you understand what’s out there and you use it. You have a 1st mover advantage on it. You will definitely build more pipeline. You’ll attract the buyers. You need to go after you scale your team? Certainly in startups and late stage startups. And now in public companies, we have deployed 16 technologies that are AI specific technologies

958
02:40:14.150 –> 02:40:18.600
Meagen Eisenberg: across our marketing functions, and we’re continuing to do proof of concepts.

959
02:40:18.600 –> 02:40:41.930
Meagen Eisenberg: And so I would say, there’s not a function in marketing that can’t be using it. And I certainly am hiring people who are comfortable with it, and if you don’t embrace it, you really can’t stay on the team. I would say so, and I have a 200 person team, and I don’t necessarily want to make that bigger. I want to make it more efficient. I want to compete. So I actually think these tools are amazing in what they’re doing already.

960
02:40:41.930 –> 02:41:06.719
Meagen Eisenberg: And I’m fortunate that I’m at Samsara and we’re embracing these tools across the company, across engineering and go to market. And we’re seeing some great early results. So I think it works. I think it does personalization where we want it. I think it scales. I think that. Yes, you have to. You have to understand it, and and some of it you have to train, and you have to learn, but also just create

961
02:41:06.720 –> 02:41:10.699
Meagen Eisenberg: an environment for your team to to play with it. Because also, it’s just fun.

962
02:41:11.450 –> 02:41:21.820
Mark Organ: Right? It’s a good segue. But what’s what’s the what’s 1 task that you wish that AI could solve tomorrow that today it just doesn’t cut it.

963
02:41:24.620 –> 02:41:46.105
Jennifer Rapp: I can chime in there. Hi, guys, I’m Jen rap. I am Cmo at a creative company. So I can really go into like creative side of AI. Here. I’m Cmo at Super side, and we are an AI powered creative service. I also come from Klaviyo and ran marketing at Doordash so very varied. Background.

964
02:41:46.740 –> 02:42:01.250
Jennifer Rapp: I think the the area that I want to employ AI, and maybe some of you on this call are, I would love to hear. If you are, I cannot for the life of me figure out how to bring AI in to enhance marketing attribution.

965
02:42:01.370 –> 02:42:02.440
Jennifer Rapp: Can any of you.

966
02:42:03.090 –> 02:42:19.690
Jennifer Rapp: Are you able to say like, Look AI is telling me that this lead came from this, you know, had these 6 touch points, and this is why they resonated and do more of that. I think to me that would be a huge AI win. I haven’t figured out.

967
02:42:20.980 –> 02:42:24.479
Amanda Kahlow: I think that’s what Chris Golak’s new company is doing, but I don’t know much about it.

968
02:42:24.480 –> 02:42:40.100
Zack Holland: Yeah. And then, like on the consumer side, you have like companies like north beam that have been pushing the the boundaries there for a while. Presci and AI just roll out their own AI model all around like AI enabled mixed media modeling. So there’s some interesting things in the space. I mean, I think the most interesting thing is

969
02:42:40.421 –> 02:43:05.769
Zack Holland: kind of like, Megan was saying right now, everybody’s deploying 16 different AI tools. What I what I find really interesting is like, when will we get an agent system that actually does start to approach Agi, that can use these 16 tools for me to accomplish the larger goal, which is more sales, spending, less money, or or whatever that is, and where it’s more of like the orchestration between agentic tools that you you can get pulled out of otherwise it just feels like.

970
02:43:05.770 –> 02:43:24.759
Zack Holland: Oftentimes there were a lot of Martech tools before now. There’s a lot of martek tools that have AI in them, and they’re better. But it still feels like you spend all day jumping around between 16 tools. And then you got to go, demo 3 more, and oftentimes you wish there was a you know, that AI could start to actually pull me out of the need to be in all of these siloed tools, and actually bring things together.

971
02:43:25.230 –> 02:43:29.739
Kingston Duffie: You know, I’m my background is really the engineering side of things, and

972
02:43:30.140 –> 02:43:33.860
Kingston Duffie: whenever I hear orchestration I have nightmares over

973
02:43:34.020 –> 02:43:41.100
Kingston Duffie: big orchestration systems that are just in a constant state of chaos. What seems like a lot better approach.

974
02:43:41.320 –> 02:44:06.320
Kingston Duffie: as we sort of re-engineer for the marketing organization for, and the whole go to market organization is that we should be thinking about data rather than tools. We should be thinking about which of these tools is throwing off useful data. And how do these tools consume that data? And you know, there’s lots of good technologies for data islands to do this stuff. Our tool, for example, is sitting out on the website

975
02:44:06.320 –> 02:44:14.739
Kingston Duffie: having conversations with people coming to the website who are high in the funnel. And we throw off really really interesting data, you know, which is basically.

976
02:44:14.740 –> 02:44:31.390
Kingston Duffie: what are the what we’re looking inside. We’re like an X-ray machine looking inside the mind of the top of the funnel. And that data is super valuable. But a lot of our customers don’t know how to use it, you know. At most they’re going to sit and look at a feed of it like it’s an Instagram feed.

977
02:44:31.390 –> 02:44:49.659
Kingston Duffie: And we would love to see other innovators saying, Oh, if you can give me access to that data. Look what we could do for you. So I’m hoping to see a lot of innovation of where independent companies are starting to find ways in which to share. AI produced and AI consumed data.

978
02:44:50.960 –> 02:45:14.100
Amanda Kahlow: I love that Kingston and I would like Double click there to say, the other place. I think people are getting this wrong like across using these AI tools. And what they’re thinking about is they’re thinking about replacing their existing systems and how we do things today. And I think we have to really realize that this is a revolution. This is something that new. And so a lot of the tools that are spinning up are making what we do today is a point solution. Better

979
02:45:14.100 –> 02:45:39.059
Amanda Kahlow: versus if I have like starting from 1st principles, like, if I’m starting from beginning, and I have this tool at my disposal. What would I do with it? And so, like, you know, in our specific case, like when and to no fault of our amazing customers. But they’re thinking about. Let’s put a superhuman to replace your inbound. Let’s put a superhuman to replace a sales engineer. Well, and let’s put a superhuman to replace a Csm. We’d like to think about it as why not just have a superhuman

980
02:45:39.060 –> 02:45:55.020
Amanda Kahlow: to do? Go to market from the beginning of the funnel to the end of the funnel, have memory through all the conversation. Look up all what’s happening in Crm, and it’s at that one role versus because humans have limitations. So like, as we were saying earlier, where is the bar in AI versus a human?

981
02:45:55.020 –> 02:46:19.989
Amanda Kahlow: For better or worse, we actually have a much higher bar for AI than we do for humans like you think about a salesperson, and you ask me if my superhuman hallucinates I’m like, have you listened to your salespeople right on calls like they hallucinate, and they hallucinate nefariously like they know they’re doing it to get the deal as soon as the superhuman does it once they never do it again. So I challenge people to look outside of the box of the way things are working today, and if we had a clean slate. What would we do?

982
02:46:20.930 –> 02:46:25.130
Zack Holland: Yeah, and I think that’s an interesting like decision for the future is like, do we build.

983
02:46:25.170 –> 02:46:39.389
Zack Holland: yeah, humans have limitations. It’s like, do we build to unlock their limitations and make them superpower, you know. Give them superpowers, or do we build to replace them with superhumans, which I think is just kind of like a core difference. And I think in, you know, on the marketing side, we’ve underserved

984
02:46:39.390 –> 02:47:01.340
Zack Holland: users where in engineering, they’ve gotten these these products quickly. That do make an engineer a superhero, you know they really do. 10 x what they can do. And all we’ve really been doing on the marketing side is rolling out a lot of tech that, like bogs, users down with more things to do in more of these like really point solutions. So it’s like, obviously super interested in, like how you can take a marketer and make them a 10 x marketer by taking those limitations away.

985
02:47:02.240 –> 02:47:06.799
Chris Hicken: Well, Sam Altman recently said that he thinks Agi will be available in

986
02:47:07.860 –> 02:47:12.270
Chris Hicken: a few 1,000 days is what is so so available.

987
02:47:12.270 –> 02:47:16.110
Chris Hicken: he so well. No, no, Agi is not. No, no.

988
02:47:16.110 –> 02:47:19.209
Amanda Kahlow: His definition of replacing work. So like, he wanted that.

989
02:47:19.210 –> 02:47:21.190
Amanda Kahlow: Okay and Microsoft contract, anyway.

990
02:47:21.190 –> 02:47:44.860
Chris Hicken: Okay. But if you go by the kind of the standard definition of Agi, we’re probably in his mind. We’re like 10 years out, really from that being available. So what we’re talking about is, how do we make AI agents more and more capable to replace more and more tasks and feel more like Agi while we’re waiting for Agi to become available. And so when you say like, what what do you wish? AI can do better.

991
02:47:45.010 –> 02:48:03.469
Chris Hicken: I mean. It’s it’s, you know, if I think about like, you know, what’s the definition of Agi, I mean, I want it to have better memory. I want agents to be able to collaborate better together. I want AI to be able to pursue a goal across multiple tasks. I want the AI to be able to learn from outcomes

992
02:48:03.780 –> 02:48:12.530
Chris Hicken: today. Those are very unsophisticated, even in AI agentic AI workflows. And so I’m hoping that those things improve in the next

993
02:48:12.930 –> 02:48:22.200
Chris Hicken: year or 2, so that we can really take advantage of them for their workflows. And maybe Amanda really replace, you know, like a whole, an entire role or an entire function.

994
02:48:22.840 –> 02:48:25.419
Amanda Kahlow: Yeah, I got you on the memory and the workflow.

995
02:48:25.620 –> 02:48:26.020
Chris Hicken: Yeah.

996
02:48:26.020 –> 02:48:29.150
Zack Holland: Yeah, a lot of that stuff is is here now, for sure.

997
02:48:29.650 –> 02:48:58.089
Mark Organ: Yeah, I think the big thing for me what’s missing are agents that think like scientists, that I create hypotheses and go out and test it and learn. And that’s I think that’s 1 of the things that’s sort of both special about humans. And I look forward to see more hypothesis, creation and testing experimentation and insight development would be pretty amazing. I don’t know if you’ve seen anything do that yet. I’ve not seen anything do that. Well.

998
02:48:58.616 –> 02:49:03.199
Mark Organ: and apparently there’s some limitations, even on the math, in terms of getting that done. Well.

999
02:49:03.200 –> 02:49:13.989
Kingston Duffie: You know, the guardrails takes on a whole new meaning when you get into the world of marketing and sales, because, of course, there’s guardrails, which is, you know, don’t talk about

1000
02:49:14.250 –> 02:49:26.880
Kingston Duffie: stuff that’s not suitable for the workplace, but there’s guardrails which is just. Don’t destroy my brand. You know that telling a story in a way that just isn’t the way you want it told can be

1001
02:49:26.940 –> 02:49:45.869
Kingston Duffie: super destructive. So you know, I think we’re we spend a lot of time really thinking about how to do guardrails and what we’ve learned so far is, boy. The real simple basics make a big difference, such as here’s a whole set of questions not to answer, boy, you know, with that kind of stuff.

1002
02:49:46.070 –> 02:49:46.650
Amanda Kahlow: In cash.

1003
02:49:46.650 –> 02:49:48.930
Kingston Duffie: Really really go the wrong direction.

1004
02:49:49.370 –> 02:49:53.530
Zack Holland: Yeah, and or or why for this phase right now, of where we are.

1005
02:49:54.000 –> 02:50:18.420
Zack Holland: pacemakers who can guide AI products in the right way are becoming so valuable right where it’s like. There’s so much of that minutia where you might not trust just an agent system to go off and start testing things with your brand in public. But somebody really experienced with a lot of AI powers underneath them. You would trust that person to go off and test more things. And with and without you there managing those guardrails day to day as a busy, you know, business owner.

1006
02:50:18.420 –> 02:50:31.190
Zack Holland: Right? So it’s like, there’s that that interesting role that’s emerging of like what senior engineers are doing with cursor and things like that. But on the marketing and sales side of that kind of like tastemaker guide, whether that’s in brand or on ads, or whatever that thing is.

1007
02:50:32.163 –> 02:50:34.829
Chris Hicken: Oh, sorry, Megan, go ahead!

1008
02:50:35.320 –> 02:50:54.760
Meagen Eisenberg: Oh, I was just going to say, when you’re talking about really, how you prompt, and the guide rails that you’re giving it. It reminds me in school, when you had the assignment where you had to tell your teacher how to make a peanut butter and jelly sandwich, and I don’t know if any of you had to do that where it was a very step by step, instruction, and so you would say, you know.

1009
02:50:54.760 –> 02:51:17.469
Meagen Eisenberg: get some peanut butter. Instead of getting peanut butter with a knife which was there, they would just stick their hand in the peanut butter, and then it was like, Get a piece of bread instead of a slice of bread. You got a little ripped up piece, and so you kind of went through this funny exercise of being very literal and specific, exactly what you wanted to do. And certainly I’ve experienced that in playing with Chat Gpt and others where I tried to build the

1010
02:51:17.470 –> 02:51:41.250
Meagen Eisenberg: doll and the doll thing meme that went around, and I was like, grab my laptop and then walk forward. But I didn’t say exactly what shelf it was on what it looked like, and it grabbed my sunglasses, and then it morphed into this really weird laptop and moved forward. But I just think there’s a lot around, not only the guardrails, but also just prompt engineering. And how we’re we’re training ourselves to be more literal. It also reminds me of

1011
02:51:41.250 –> 02:51:49.370
Meagen Eisenberg: computer science classes. When you’re writing code, you’re just you have to be so specific in it. So I don’t know. I find it a lot of fun.

1012
02:51:50.050 –> 02:52:01.179
Chris Hicken: And I want to make sure people don’t miss the nugget that Amanda just dropped for the audience when you’re trying to control Openai, and it’s not listening to what you want it to do. If you put it into caps.

1013
02:52:01.370 –> 02:52:09.949
Chris Hicken: it actually does start to listen to you. So that’s a Little Nugget there for prompt engineering put things in caps that you really really really need the AI to follow.

1014
02:52:11.360 –> 02:52:14.009
Chris Hicken: Thanks. That’s a big insight. I didn’t know that

1015
02:52:14.010 –> 02:52:17.350
Chris Hicken: I didn’t. I thought I thought that was, yeah, thanks, Chris.

1016
02:52:17.350 –> 02:52:23.719
Chris Hicken: No, seriously. I’ve used that trick on several of our prompts, and it’s the only way I can get Openai to actually listen.

1017
02:52:24.570 –> 02:52:25.650
Amanda Kahlow: And put it up. Front.

1018
02:52:25.650 –> 02:52:27.170
Meagen Eisenberg: You yell at it.

1019
02:52:27.465 –> 02:52:31.309
Chris Hicken: Earlier is better. Yeah, earlier is better and bold is better or sorry.

1020
02:52:31.310 –> 02:52:32.979
Zack Holland: Exactly like some employees.

1021
02:52:33.936 –> 02:52:34.909
Chris Hicken: Yeah, okay.

1022
02:52:34.910 –> 02:53:03.130
Amanda Kahlow: And do it often, repeat it often early and often in all caps. So when you you know, I mean, it sounds so basic. But yeah, some of these things are important, and that’s what you know, Kingston, as you were saying, people. Or maybe it was you, Chris, earlier, that you try to build it yourself. It’s just not going to work if you don’t know all these little nuanced tricks right to make it happen. And you know the balance between fine tuning and large prompts and context windows, and doing rag and like, when to use which and what works and what actually will deliver the ultimate experience is.

1023
02:53:03.130 –> 02:53:07.729
Amanda Kahlow: you know, it’s a it’s an art right now, like, and it’s it’s a beautiful art. If you ask me.

1024
02:53:08.140 –> 02:53:18.720
Kingston Duffie: Yeah, our system. It took us a long time to realize we were just giving it too many instructions. We cut our instructions by about 70%. And it got a lot better.

1025
02:53:18.860 –> 02:53:21.649
Kingston Duffie: It just had to be the right 30%.

1026
02:53:21.940 –> 02:53:23.650
Chris Hicken: Exactly. That’s right.

1027
02:53:25.400 –> 02:53:47.600
Mark Organ: Well, let’s let’s talk a little bit. I mean, there’s there is a lot of a lot of hype around the I work for. I don’t know where you guys think we are on the hype curve. It really feels like we’re kind of close to the top. But love to hear from you like what is the most overhyped aspect of agentic marketing? And but where’s the real lasting value that people are missing.

1028
02:53:47.600 –> 02:54:03.409
Mark Organ: that when we get to the bottom of the hypervent things start to level out again, kind of like the segway which is now, you know, a standard for tours and warehouses, you know where’s the real lasting value with respect to agentic marketing.

1029
02:54:03.710 –> 02:54:06.637
Zack Holland: Did you just compare Agentic Marketing’s hype to the Segway.

1030
02:54:08.300 –> 02:54:10.909
Mark Organ: That’s my favorite example of that hyper.

1031
02:54:11.390 –> 02:54:19.650
Mark Organ: But no, but you know what though, chances are, this technology is going to follow something like that maybe maybe be a little more impactful than the Segway.

1032
02:54:19.870 –> 02:54:20.839
Zack Holland: Yeah. I’m glad I did.

1033
02:54:20.840 –> 02:54:35.359
Mark Organ: But but but yeah, there are probably some aspects today where it is overhyped and not going to deliver everything that people think. And then there’s gonna be something that’s going to be utterly transformational forever, and marketing departments will never be the same again.

1034
02:54:36.330 –> 02:54:59.050
Chris Hicken: I guess I don’t really care where we are on the hype curve, I mean in terms of adoption. We’re really early on the adoption of agents for workflows. I mean almost nobody that I know uses them in their personal life. I mean the people that I know that are using them are, they’re building it into products. So it’s these are engineers or product managers with deep, you know, coding expertise. And so I actually think that. And I’ve told this to everybody at my company.

1035
02:54:59.050 –> 02:55:07.270
Chris Hicken: That Agentic AI is not a nice to have skill. It’s a must have skill that dramatically improves the quality

1036
02:55:07.270 –> 02:55:20.589
Chris Hicken: of the quality and throughput of your work. If you use it well and incorporate it into your daily workflows. The way that you might open up Chat Gpt to ask a question today. And so, yes, I think you know, Hype.

1037
02:55:20.600 –> 02:55:35.529
Chris Hicken: sure, I’m okay with the hype. But you know, I think we need to get people to start using it. And right now it feels daunting like, where do I even get started? I hope, Mark, we get to that question of how do we actually get started with Agentic? AI, because I think a lot of people want to use it. They just don’t know

1038
02:55:35.820 –> 02:55:40.660
Chris Hicken: what to do next. Like I have. I have my chat gpt open now what I do from here.

1039
02:55:42.130 –> 02:55:42.480
Jennifer Rapp: I can.

1040
02:55:42.480 –> 02:56:09.830
Kingston Duffie: Yeah. My impression is that we got the heart cart out in front of the horse a little bit, in the sense that in order to have an orchestration of things. You’ve got to have things, and I think we’re still early on. The things have like our our thing, I would think is one of the things I don’t think we’re in the agentic market at all. Yet what we’re what we are as a point level solution that we would love to be part of an agentic orchestration.

1041
02:56:10.020 –> 02:56:15.929
Kingston Duffie: But until people start to figure out where you can bolt things together, there isn’t enough value yet.

1042
02:56:15.930 –> 02:56:19.640
Chris Hicken: What do you mean by things, Kingston? What do you mean by things I don’t understand that.

1043
02:56:19.810 –> 02:56:23.820
Kingston Duffie: Well call them tools if you want. I think of them as

1044
02:56:23.990 –> 02:56:28.650
Kingston Duffie: as things that have data in and data out, because I’m an engineer. But

1045
02:56:29.220 –> 02:56:50.569
Kingston Duffie: think of them as if I’m going to orchestrate something. I’ve got to orchestrate tools. So some of those may be existing tools. I want to go drive a Google ad campaign, or I want to go, you know, change my Cms or whatever it happens to be. Those may be conventional tools, or those may be AI agents themselves. I don’t think that matters in the sense that what you’re doing is

1046
02:56:50.570 –> 02:56:59.860
Kingston Duffie: orchestrating. And as an agentic. AI. What you’re saying is the orchestration itself is going to be driven by an AI.

1047
02:56:59.980 –> 02:57:28.430
Kingston Duffie: What I’m trying to say is, aside from the conventional tools. We don’t have a lot of tools here that are really amenable to an AI driving them. I think that’s that’s my perception. Right now, if you ask an AI to drive a Google ad campaign. Good luck with that. I think that that’s a very, very complex thing that requires some combination of creativity and expertise and knowledge. And I don’t think we’re quite ready for that yet.

1048
02:57:28.550 –> 02:57:33.379
Kingston Duffie: But our tool, for example, is dealing directly with

1049
02:57:33.480 –> 02:57:47.849
Kingston Duffie: an audience on the website, and is throwing off all this data about how people are using the website. And I can imagine some very, very simple flows that come from that where we say, Oh, if we’re seeing this

1050
02:57:48.000 –> 02:57:54.930
Kingston Duffie: set of questions getting asked that should drive this content generation that should drive this ad campaign.

1051
02:57:55.080 –> 02:58:00.289
Kingston Duffie: Some baby steps in that direction are going to be very, very valuable to marketing teams.

1052
02:58:02.900 –> 02:58:04.769
Chris Hicken: Amanda, you must know that.

1053
02:58:04.770 –> 02:58:32.170
Amanda Kahlow: Well, I have a thesis, and it has nothing to do with what I’m working on that I was at an event last week, and I heard somebody talking about Icp, we still need to really define our Icp right? And I kind of sat there and I was scratching my head because I was like, Wait, do we like we’re not the one to solve this, by the way. So I think there’s got there are other like creative tools and things that can do this. But if you can actually target everyone and you can have hyper personalization, one to one based on their needs.

1054
02:58:32.240 –> 02:58:57.070
Amanda Kahlow: Now, I think there is some. I definitely good uses of Icp on the product side, like, okay, I can still only well, maybe like products can build themselves right over time. But this is where the hype mark to mark to your earlier question about hype. I think we’re not there yet from a product perspective. But and we’re probably not there yet from a marketing perspective. But imagine to have that dynamic content based on what I learned like, okay, I’m having a conversation with you, and I learned you have this particular pain point.

1055
02:58:57.190 –> 02:59:25.959
Amanda Kahlow: And you’re in this industry in this vertical. And you’re using this product like boom. I should hit you with the exact message in whether it be in the form of like a video that came from one of our customers, or whether it be in the form of a conversation, or a white paper, or whatever it is to have dynamic, personalized, real time content. I think that’s like the ultimate goal, and then it throws away all these old principles of like gotta. We only can do so much. So let’s get focused on what our core Icp is, and so I don’t know.

1056
02:59:25.960 –> 02:59:41.369
Zack Holland: I think when most people talk about Icp, though, they’re talking about for their actual product. So if you sell an it solution for manufacturers. You have an Icp naturally, that you can’t just run dynamic personalization for everybody in the world and get the right messaging out in front of the right people, right? I mean it. You’re right.

1057
02:59:41.370 –> 02:59:42.000
Zack Holland: No like

1058
02:59:42.000 –> 02:59:53.099
Zack Holland: it’d be awesome to have like. And that’s what we’re already getting to is like hyper personalization, where we can get our message in front of the right people. But who those right people are for certain products isn’t just everybody right?

1059
02:59:53.350 –> 02:59:57.010
Amanda Kahlow: Fair when you have limitations of what your product can do. But like I mean.

1060
02:59:57.010 –> 02:59:58.499
Zack Holland: Which most products do.

1061
02:59:59.080 –> 03:00:15.879
Amanda Kahlow: I mean. Take, for example, like I mean, you look at like a seat like Hubspot, like they’ve like in in salesforce. They both kind of like salesforce. Say, the Enterprise Hubspot did. Smb. They focused on their lane. They were going after small businesses so they could get hyper focused, and then the marketing followed along like they could serve anyone.

1062
03:00:15.880 –> 03:00:24.490
Zack Holland: Those are 2 super horizontal products. But like a lot of products, are mostly vertical. Right? And so you have to have icps, right? Those aren’t gonna go anytime soon.

1063
03:00:24.940 –> 03:00:49.369
Amanda Kahlow: Well, but I think I mean, like I look at like even ourselves. I’ll take myself for an example like we can sell like I had a roofing company come in today and to use our superhumans. That is not my Icp. But shit. They were looking about to spend 6 figures with us after a 30 min meeting. And I’m like, Yeah, we might need to look at like we can service anyone. So I think, as these projects, you know, it depends like on. That’s what I started with. The earlier comment about the product can’t scale to like

1064
03:00:49.390 –> 03:01:02.710
Amanda Kahlow: that. But marketing has been defined with defining the market. This chief market officer versus the chief marketing officer. And if it is defining the market and the product can meet multiple by multiple serve multiple masters.

1065
03:01:02.730 –> 03:01:09.119
Amanda Kahlow: Do we really need to do this anymore? And does it like turn that whole concept on its head? I think that we’re about to go into a new world.

1066
03:01:10.590 –> 03:01:15.549
Chris Hicken: So Kingston just to to. I guess, Kingston, you know I think you’re you’re right

1067
03:01:15.660 –> 03:01:25.090
Chris Hicken: that you need the things to be ready for orchestration, but in some senses you can cheat a bit in the interim. And what I mean by that is.

1068
03:01:25.090 –> 03:01:46.409
Chris Hicken: there are tools like, for example, Zapier, that can move data from one system to another system or from one database to another database. So you can start to consolidate data that you need for accomplishing a task already today and start running and use that as your tool. Basically, the database is your tool, like, for example, let’s say, a product marketer wants to

1069
03:01:46.410 –> 03:01:50.469
Chris Hicken: do automated competitive intelligence.

1070
03:01:50.470 –> 03:02:17.259
Chris Hicken: Right in Gmail, you subscribe to emails about your competitors. You have zaps that will send that data into a spreadsheet. And then you can have another agent that looks at that spreadsheet once a week and summarizes what it’s discovered, and you have a 3rd agent that looks over the last 6 months and and tries to compile the trends. And then you have a 3rd one, a 4th agent that will post those updates into a shared slack channel, for example.

1071
03:02:17.260 –> 03:02:23.239
Chris Hicken: So I do think there’s, you know, it’s not perfect. But I think there are like small things that you can do today using

1072
03:02:23.430 –> 03:02:32.470
Chris Hicken: tools like Zap Zapier is one of my favorites, but you can use tools like Zapier to accomplish a lot of marketing activities without the the tools being ready.

1073
03:02:32.650 –> 03:02:34.270
Zack Holland: And we and we have Mcps.

1074
03:02:34.500 –> 03:02:35.740
Chris Hicken: And you have. Ncp. You’re right. You have M.

1075
03:02:35.988 –> 03:02:53.890
Zack Holland: And so like, I mean, you’re getting. We’re getting closer to a place to a world where you know, agents can connect to anything, whether it’s an AI native app or or more of a legacy one quickly and easily. I mean, the Hubspot rolled out their Mcp today. That’s a that’s a pretty big change for agents right there, right? So we’re definitely not super far away from a world where

1076
03:02:54.140 –> 03:02:59.099
Zack Holland: an agent can can browse the right things, get the right data and then start to orchestrate you in the right direction.

1077
03:02:59.100 –> 03:03:03.800
Chris Hicken: Actually Zach, for people who don’t know, since Mcp is still relatively new. Do you want to just explain it?

1078
03:03:04.908 –> 03:03:25.220
Zack Holland: Yeah, definitely. So Mcp just stands for model context protocols. And it’s pretty much just a a language. Now that allows that make sure that an application can be easily navigated and understood by AI agents. Right? And so when a tool like Hubspot rolls out their Mcp. It means that now, through that Mcp, like what we used to have to build an in-house Api for

1079
03:03:25.220 –> 03:03:48.380
Zack Holland: an agent can, without having to build this really long, expensive integration can now speak to Hubspot. Right? And more and more tools are rolling out. Mcps, quicker and quicker. Google rolled out an entire language for how Mcp should be written. And so you have to think that things like Google ads and things like that are very close to to agents being able to browse and to be able to navigate in the right ways like a like a human would.

1080
03:03:48.490 –> 03:03:50.430
Kingston Duffie: And so good.

1081
03:03:50.800 –> 03:03:54.099
Kingston Duffie: I just wanted to say that in in our experience.

1082
03:03:54.280 –> 03:04:23.409
Kingston Duffie: you know, if you talk about Enterprise scale, you know. Let’s let’s talk about marketing teams with hundreds of people in them as opposed to ones at Enterprise scale. You’re going to want complex orchestration. You’re dealing with lots of moving parts in a full organization. And you’re trying to, you know, make big decisions that have to happen fast as you move down to Sme, where you’re dealing with a marketing team. That’s often 2 or 3 people at most.

1083
03:04:23.440 –> 03:04:29.309
Kingston Duffie: There’s no way they’re going to deal with the complexity that you’re talking about, right that the idea of building

1084
03:04:29.370 –> 03:04:49.989
Kingston Duffie: a functioning workflow of any degree of complexity is is way beyond that. But, on the other hand, innovation comes in through the Sme much faster than it does at the Enterprise level. So I think for agentic marketing agent AI marketing to really take off someone’s going to have to go figure out.

1085
03:04:50.130 –> 03:05:19.070
Kingston Duffie: how do we make that so simple that the smaller companies can actually start to use it? Because they’re going to be the ones that really innovate. And the yes, a few enterprises will invest heavily, and they’re going to take 5 years to do it and eventually have a really big, complicated orchestrated system. But I think the sweet spot there is still coming in through Sme and figuring out how do we make small, nimble companies able to use the tooling that’s really going to make some exciting things happen.

03:05:19.070 –> 03:05:19.770
Chris Hicken: Yeah.

1087
03:05:19.770 –> 03:05:34.930
Mark Organ: Right or nail the and nail the change management from what I’m seeing, it’s user experience and change management and getting the human so that they can adapt is seems like the biggest challenge that I’ve seen in companies. But.

1088
03:05:35.460 –> 03:05:35.850
Zack Holland: When you got.

1089
03:05:35.850 –> 03:05:36.670
Mark Organ: Just think of that!

1090
03:05:36.990 –> 03:06:04.530
Zack Holland: I think people are getting pushed to adapt. Now it’s it’s nice to sell an AI product because you get people who are like my boss told me I have to start using AI products. And so I’m here to use this. And so it’s nice to be on the other side of that. But just to really quick on what Kingston said, I think it’s totally true. I do think that even at the Enterprise side marketing teams are getting smaller and smaller right? Like marketing teams are a 5th of the size that they used to be. So they are more nimble than they used to be. And it’s also that you can get in and work with individual teams within a large enterprise organization. So

1091
03:06:04.530 –> 03:06:09.170
Zack Holland: the marketing initiatives of an enterprise team or an events team within a large enterprise system

1092
03:06:09.170 –> 03:06:22.009
Zack Holland: prove it out on a team function before you start to scale out into into enterprises. But I completely agree that if you can make things simple for that, for the Sme and and for the you know, the the 4 or 5 person marketing team. It’s a good way to prove things out for scaled up.

1093
03:06:22.410 –> 03:06:52.039
Chris Hicken: Yeah, I think the last, you know, before we move on from Mcp. And I didn’t read Hubspot’s press release. But I’m not so. I’m not sure about this, but one of the promises of Mcp is, for example, anyone at the organization could go into the Hubspot slack channel and ask questions of their own hubspot database like, who did we, you know? Which customers did we acquire last week, or what’s the size of my biggest account, or for the you know. Ibm account. What have they? You know? How many times have they renewed, and what’s their current licensing amount? So

1094
03:06:52.040 –> 03:07:03.100
Chris Hicken: in theory, I don’t know if Hubspot deployed it this way, but in theory you would be able to ask your own database questions like this and democratize it for the Sme in a slack channel. For example.

1095
03:07:04.240 –> 03:07:16.499
Zack Holland: Yeah, that’s that’s exactly what it is. Or you could use a platform like Avery where you could talk to your AI marketing manager, and you could talk to through the Mcps, Hubspot, Google, Meta, and all of it within the context of what you’re actually doing in your program.

1096
03:07:19.300 –> 03:07:38.219
Kingston Duffie: Yeah, I think there’s an interesting future where there’s been a lot of presumption whether you’re in the sales organization, the marketing organization that the center of your organization is the Crm. And I think that the Crm may get moved to the side. We may see something else trumpet in the middle. Hubspot and Salesforce, of course, would love to

1097
03:07:38.480 –> 03:07:47.820
Kingston Duffie: keep that center spot. But AI is such a disruption. I think there’s room for somebody to come in and potentially, really change.

1098
03:07:47.970 –> 03:07:50.709
Kingston Duffie: You know what’s at the center of that hub and spokes.

1099
03:07:50.710 –> 03:07:58.820
Zack Holland: Amanda with y’all’s sales superhumans. How much is the Crm that, like core piece of the infrastructure.

1100
03:07:58.990 –> 03:08:01.310
Amanda Kahlow: Of bringing the Crm data into the superhuman.

1101
03:08:01.310 –> 03:08:08.799
Zack Holland: Like, how important is the Crm in the future? Pretty much, you know. Does it remain this like vital piece, or does it start to become less important.

1102
03:08:09.230 –> 03:08:15.300
Amanda Kahlow: I think the data is vital. But does it need to sit in a structured system of record that.

1103
03:08:15.300 –> 03:08:19.669
Zack Holland: Yeah. Does it sit in Hubspot? Does it sit in salesforce, or does that? Does that entity start to to change.

1104
03:08:19.670 –> 03:08:23.640
Amanda Kahlow: One of those is one of our largest customers. So I I have to say, yes, alright.

1105
03:08:25.965 –> 03:08:27.980
Amanda Kahlow: So I am a big fan.

1106
03:08:28.614 –> 03:08:34.949
Amanda Kahlow: Yeah. So yes, it still sits there. But you know, at the end of the day it’s like we need. We need the data.

1107
03:08:34.950 –> 03:08:42.199
Amanda Kahlow: And we need to have like somebody said earlier. I don’t remember who was saying this. I think it was Kingston, but about memory, right? So, remembering.

1108
03:08:42.200 –> 03:09:06.560
Amanda Kahlow: pulling out every piece of information about the opportunity, about the account, about where they are in their sales cycle, or what products they have, what we want to upsell, but also like, hey? The superhuman talked to somebody yesterday I talked to Cmo. And now I’m talking to the Cro. Can I bring those conversations together like we literally one of the most impressive things that we’ve done as a company that I just am so excited about is, 3 weeks ago a Cro talk to our superhuman for 90 min

1109
03:09:06.860 –> 03:09:11.680
Amanda Kahlow: over the weekend when no salesperson was going to get on the phone, and then then he passed it to the Cmo.

1110
03:09:11.780 –> 03:09:17.539
Amanda Kahlow: This is of a publicly traded company. We had a 30 min call and close a hundred K. Deal in a matter of weeks.

1111
03:09:17.540 –> 03:09:41.979
Amanda Kahlow: So she did all the selling she scoped. She did the use case when the salesperson wasn’t available, but it was being written back to your question. The information is written back to our Crm. So that she could pull that information in, and she had memory of her past conversation so she could have the context and pick up from where she left off with the Cro when she talked to the Cmo. To say, Hey, I actually know you have these pain points, and you’re trying to go into these markets. And this is what you’re doing today.

1112
03:09:41.980 –> 03:09:51.029
Amanda Kahlow: and could just go versus starting all over fresh again. So I think that’s what’s really critical in these. And it’s, you know, not the easiest thing in the world to build

1113
03:09:51.306 –> 03:09:55.169
Amanda Kahlow: but we’re excited about it. Sorry I didn’t mean to make that a pitch.

1114
03:09:55.510 –> 03:09:56.979
Zack Holland: Yeah, no, it’s all. It’s awesome.

1115
03:09:57.010 –> 03:09:58.249
Mark Organ: I can’t help myself.

1116
03:09:58.250 –> 03:10:02.810
Mark Organ: It wouldn’t be you on a panel if you didn’t do a pitch, so we you didn’t disappoint.

1117
03:10:02.810 –> 03:10:04.340
Amanda Kahlow: Can’t help myself. I just get some.

1118
03:10:04.340 –> 03:10:04.810
Mark Organ: I know.

1119
03:10:07.280 –> 03:10:18.475
Mark Organ: Yeah. But actually, it’s a good segue, though in terms of what the marketing department of the future looks like. Like, what do the humans actually do, I mean, do we even have a Cmo in the future?

1120
03:10:18.770 –> 03:10:19.360
Amanda Kahlow: Yes, Megan.

1121
03:10:19.360 –> 03:10:20.670
Mark Organ: With, yeah.

1122
03:10:20.670 –> 03:10:37.700
Mark Organ: what do the people? What do the people do have? What is the how does the staffing look like, you know, 10 years from now, for example, 5 years now, 10 years. Now, what does the marketing organization of the future look like? Actually, Megan would be a good person to talk to Job Megan 10 years.

1123
03:10:37.700 –> 03:11:01.710
Meagen Eisenberg: Yeah, I mean, I think you’re a leader of agents, right? Whether you’re you know. Maybe it’s the Cao chief agent, officer or something. But I think you are. You’re not going to go up in marketing unless you can manage agents. I do think there’s some element of understanding, you know. Recently they were talking about Vcs will never go away, because

1124
03:11:01.710 –> 03:11:16.679
Meagen Eisenberg: there’s some sort of like intuition and and sense of markets and themes and different things. I think there’ll still be an element of creativity and leadership. But to your point of needing large teams. No, and I think

1125
03:11:16.680 –> 03:11:41.569
Meagen Eisenberg: AI native companies don’t have marketing at all right. A lot of these startups. They come to me for advisory and none of them. They don’t have anyone in marketing yet, and they’re getting 100 million 200 million. We have one that just 300 million. Right? So it’s definitely changing. I think we won’t be logging into software. I think we’ll have just a console. Right. We’ll have Gemini, or we’ll have perplexity, or you know there’ll be tools that

1126
03:11:41.570 –> 03:12:05.449
Meagen Eisenberg: I just go to that. And I start asking questions or start prompting it to go do something. And maybe at some point I’m not even typing. I’m just talking back and forth with my computer, asking the questions and then making decisions or assessing the decision. Do you want to do this? Yes, go do that. No, you know, don’t do that. And so I certainly think we’re changing and evolving for sure.

1127
03:12:07.010 –> 03:12:10.930
Chris Hicken: The Cmo is a really weird job. It’s the

1128
03:12:11.090 –> 03:12:16.659
Chris Hicken: the probably the on the executive team. It’s probably the hardest job to fill. And and what I mean by that is

1129
03:12:17.750 –> 03:12:30.869
Chris Hicken: the Cmo. Is responsible for a crazy number of different things. The Cmo is responsible for brand and creativity and creative product, marketing, performance, marketing, customer, marketing, Comms and Pr

1130
03:12:31.250 –> 03:12:37.510
Chris Hicken: content, social media events. And each each of those functions has it

1131
03:12:37.610 –> 03:13:01.530
Chris Hicken: has a specialized person who you can’t like, you can’t just move your product marketer into performance marketing. And they’re going to perform. So it’s almost like each team within marketing is its own specialized unit with non-transferable skills between each other. So I agree with Megan that the Cmo. Is really the ultimately is the orchestrator of a team of Ais. But, Megan.

1132
03:13:01.610 –> 03:13:16.209
Chris Hicken: I’m curious what you think about this in general. When I’ve hired Cmos they tend to have of those things that I listed. They tend to have one area that they grew up in. So they tend to be very good in brand or very good in performance marketing. But in the other areas.

1133
03:13:16.400 –> 03:13:39.160
Chris Hicken: naturally, they’re less strong. So they don’t actually have strong expertise. Not always, but generally they don’t have strong personal experience and expertise across all the different functions. And so I guess the question is, you know, is the Cmo. How much value can the Cmo add as the orchestrator of of these different agents, when even they themselves don’t have

1134
03:13:39.440 –> 03:13:44.619
Chris Hicken: deep experience in all the areas that they manage. What do you think about that?

1135
03:13:44.620 –> 03:14:13.950
Meagen Eisenberg: I mean, I guess I could just ask right, like, what’s my list of things I should be doing in product marketing right now, or what should I be doing with Brand? And there’ll be all these knowledge bases I can access, I think. Cmos the best Cmos. Yes, they come in, and maybe they have a product marketing background or a growth background. But as soon as you’re a Cmo, you start to experience all the other functions. Certainly at Mongodb. I’d never done Comms or Pr. And brand and trial by fire. I was in there doing that, and I think

1136
03:14:13.970 –> 03:14:39.040
Meagen Eisenberg: the smart ones learn to hire really? Well. So you’re hiring someone who’s who needs to hire well and can do their research or pull the information they need. So I still think you can lead. And actually, maybe you need less of a background in anything. You just have to be smart enough to know what to ask, where to go understand the business, and maybe all the way back to just being a really good learner.

1137
03:14:39.040 –> 03:14:59.180
Meagen Eisenberg: Right? We always talk about what do you look for when you’re hiring continual learners and people that are smart that can connect the dots and understand. And so I think you’re going to be. You know, the 10 X marketers are going to be the ones that learn quickly, try things can hire where they need to hire, and are good at prompt

1138
03:14:59.400 –> 03:15:02.909
Meagen Eisenberg: writing prompts or knowing what to ask. They’re curious.

1139
03:15:03.560 –> 03:15:24.059
Chris Hicken: So reading between the lines, I think what I’m hearing you say is, and I would agree with this, that a Cmo. Is more likely to have seen excellence delivered in each of those areas that they oversee. So they’re going to be one of the best judges of quality. And when you say they know how to hire for the right people. I think they would also be able to assess the output of an AI

1140
03:15:24.180 –> 03:15:28.079
Chris Hicken: to make sure that it is delivering quality at at a level of excellence.

1141
03:15:28.350 –> 03:15:29.869
Meagen Eisenberg: Yes. Yeah. Sure.

1142
03:15:30.580 –> 03:15:56.899
Zack Holland: Like senior leadership, right to be a great CEO. You don’t have to have deep experience in every single thing that every single person under you does you have to understand what excellence looks like, and be able to pull that out to drive it, which is the exact same thing we have to do from our AI employees right? And and in the future it’s just going to be only more important that, like senior tastemaking leadership is just going to be the main value that separates people who are 100 x users of AI and people that are 2 x.

1143
03:15:57.460 –> 03:16:05.059
Kingston Duffie: I also wanted to make a point that when I look at marketing, you know, I’ve hired a bunch of marketing guys in startups over the years. And

1144
03:16:05.560 –> 03:16:07.470
Kingston Duffie: you know I think of it as 3 things.

1145
03:16:07.630 –> 03:16:12.939
Kingston Duffie: There’s creativity. There’s psychology, and there’s execution.

1146
03:16:13.180 –> 03:16:40.190
Kingston Duffie: You know. The creativity is thinking outside the box really, genuinely thinking outside the box. Why is, how can I make something happen that someone else hasn’t thought of before? Psychology is really getting inside the heads of the of my customers mainly to see the world from their perspective and execution is just all that meat and potatoes stuff that has to get done, you know. Like, how do you make a good advertising campaign work? How do you? You know all that stuff? My feeling is that

1147
03:16:40.561 –> 03:16:45.060
Kingston Duffie: there’s a lot of hype around the 1st 2 in AI. But there is

1148
03:16:45.130 –> 03:16:47.699
Kingston Duffie: real opportunity on the 3rd one

1149
03:16:47.710 –> 03:17:15.740
Kingston Duffie: that that what you were saying, Chris, I think, is exactly right. There’s just a lot of complicated things to do in marketing today that don’t require a huge amount of creativity or deep psychology, but they do need complicated execution. And that’s the low hanging fruit for AI. And I think that the companies that find the simple ways in which I can bring AI to bear and actually solve a problem. I have. Those are the guys who are going to get the big.

1150
03:17:15.810 –> 03:17:17.900
Kingston Duffie: the big returns in the short term.

1151
03:17:19.890 –> 03:17:22.270
Mark Organ: Yeah. Good point. That’s great.

1152
03:17:23.950 –> 03:17:52.899
Mark Organ: is there? Is there any role that you can see yourself adding in either in your organization or your customers organization because of AI. We talked a lot about cutting down someone talking to the chief agent officer, for example. That’s pretty interesting idea, like, maybe maybe that is one type of role is that somebody’s just in charge of of developing and fine tuning and even killing agents in the future. Love to get your thoughts on new roles in the future.

1153
03:17:53.620 –> 03:18:16.590
Amanda Kahlow: Yeah, I mean to this a similar point. We’ll have it. We’re working on a Chief AI officer which is not just working on product, but also across the whole organization like that needs to be, it needs to roll up across. And so that cross functionally, so it’s not just in product. It’s not just in marketing, not just in sales. But how are we doing this? And being as most effective and efficient as possible? Across the org.

1154
03:18:16.970 –> 03:18:19.830
Zack Holland: Like an AI’s are, yeah. Makes sense.

1155
03:18:19.830 –> 03:18:20.439
Kingston Duffie: I think just.

1156
03:18:20.440 –> 03:18:39.329
Zack Holland: You know, making sure that you’re building, and that’s something that’s safe and and building in the future that’s actually protecting. And I think, like that. That kind of role of like somebody who is obsessing about, you know, like the future of the product, and and how that will affect humans, and the kind of the governance of all of it. It’s like an interesting role that you wouldn’t see a couple of years ago. But I’ve started to see people hire now.

1157
03:18:39.600 –> 03:19:03.479
Kingston Duffie: So a lot of teams, of course, have marketing Ops. But I think there’s going to be marketing development. You know. I think that you’re getting to a level of sophistication expectations where you’re going to want people whose expertise is how to develop these systems that we’re talking about, which is not the same skill set as how do I manage a database or a Crm? It’s much more about the creativity of

1158
03:19:03.600 –> 03:19:15.769
Kingston Duffie: you know how to do, how to. And is that prompt engineering? I don’t really care. It’s really about. Can they go make something great out of the new tooling that’s available.

1159
03:19:17.180 –> 03:19:18.559
Chris Hicken: So for us.

1160
03:19:19.070 –> 03:19:22.570
Chris Hicken: Well, we never got around to. How do you get started with Agentic AI.

1161
03:19:22.570 –> 03:19:23.470
Mark Organ: So.

1162
03:19:23.470 –> 03:19:37.610
Chris Hicken: The simplest thing for most people is going to be to go get something like Langchain or flow wise, which are tools that help you set up your own agents. There are Youtube videos that you could watch within

1163
03:19:37.700 –> 03:19:59.799
Chris Hicken: 30 min. You’ll have your 1st agent set up and running on your own very, very simple to get started. So that’s FLOW. ISE. That’s flow ice, and Langchain, LANG. CHAI, n, so you can, you can change. You can check out those tools, watch Youtube videos to get started. But for us, we’re hiring. I’m interviewing later

1164
03:20:00.110 –> 03:20:21.150
Chris Hicken: later today and tomorrow, people, 2 folks that will be effectively our internal AI agentic AI employees. So they’re going to help us set up workflows to help us streamline our own internal processes, and they’re also. So that’s 1 of their responsibilities. The other one is to teach the team how to use

1165
03:20:21.150 –> 03:20:33.139
Chris Hicken: agents more effectively, and to use these workflow tools. And then the last thing they’ll be responsible for is to your point. Kingston, creating tools that can plug into the ais

1166
03:20:33.140 –> 03:20:39.829
Chris Hicken: for the purposes of accomplishing tasks, especially ones that are shared between between teams. So

1167
03:20:40.000 –> 03:20:45.259
Chris Hicken: knock on wood. We’ll have that person hired in a in a couple of weeks. But yeah, that’s that’s who we’re we are adding.

1168
03:20:47.240 –> 03:20:48.920
Mark Organ: Anyone else or someone you’re adding.

1169
03:20:50.650 –> 03:20:51.870
Zack Holland: I mean, if you’re.

1170
03:20:51.870 –> 03:20:57.589
Zack Holland: it’s an everyday marketer, and you’re looking to use agentic marketing to to improve your workflows. 100 x what you’re putting out.

1171
03:20:57.710 –> 03:21:25.830
Zack Holland: You have to come and try Avery out same thing with Amanda superhumans on the sales side. But I think, of course, like there’s amazing tool, Lindy and respell, or other tool, like great like work, like workflow tools that you can start to build like Zapier, like agentic workflows. But I think, if, like all that is still like, I don’t have the time. I don’t have the bandwidth. All this stuff scares me. I think like using, you know, using a tool like Avery. That kind of like makes things more simple and straightforward for you on the strategy development and the content creation side. It’d be a good place to get started.

1172
03:21:27.430 –> 03:21:28.140
Mark Organ: Not exactly.

1173
03:21:28.820 –> 03:21:58.069
Amanda Kahlow: Your question. But one thing that we’re doing to like encourage people to start using more AI like it’s not official yet, but I am working with the board to come up with. If you can replace yourself, we’ll forward. Rest your equity right so, and every time you want to ask me for a head count. I’m asking you like, okay, so have you built a superhuman to figure out if you can do that before you get your head count. Tell me there’s a limitation with with agents or with our superhumans to get that job done, and if there isn’t, if you can’t do it, then we’ll talk about like additional headcount.

1174
03:21:58.070 –> 03:22:10.780
Amanda Kahlow: But I love incentivizing people to like really think about their own work and great, and you get it forward vested, and you can stay at the company because we’ll give you another job. But all that equity is forward vested. And now we’ll give you new equity for your next job.

1175
03:22:11.180 –> 03:22:12.430
Zack Holland: Has anyone done it?

1176
03:22:12.590 –> 03:22:14.110
Amanda Kahlow: I’m working on it. I haven’t gotten.

1177
03:22:14.270 –> 03:22:17.270
Zack Holland: That’s awesome. I love that I love the initiative. It’s really cool.

1178
03:22:17.270 –> 03:22:28.869
Amanda Kahlow: I mean, it has to be fully replaced like they’re gonna take on. And there are certain jobs within our company that we can absolutely do it. So I’m like great. And I look at the scheme of things like I’ll give out more. There’s always more equity going around right so like.

1179
03:22:28.870 –> 03:22:29.400
Chris Hicken: And.

1180
03:22:29.400 –> 03:22:43.939
Amanda Kahlow: You can always work on that, and then I can make everyone else make a lot of money in the process. And so I’m all, for like trying to find ways to, you know, give employees incentives to be their best, and help the company grow and scale, and be most as efficient as possible.

1181
03:22:45.330 –> 03:23:11.489
Mark Organ: Do you have a question from the from the chat around government and regulation looks like the EU nannies are are trying to do something in the AI act. To regulate things. What are you guys seeing in terms of the regulation front, and how that might impact your ability to to innovate and just innovation and marketing. You know, AI gen tech stuff in general.

1182
03:23:12.280 –> 03:23:22.490
Zack Holland: I think, like an interest. Just an interesting side point on this is that marketing inherently today, before we’re we have AI’s marketing to ais, you’re marketing to humans. And humans have an amazing way of

1183
03:23:22.700 –> 03:23:31.730
Zack Holland: keeping things within the realm of, you know, humans hate things that look. Obviously AI done. There’s there’s like there’s almost like a governance of the customer already

1184
03:23:32.190 –> 03:23:48.180
Zack Holland: in marketing that keeps you from kind of like going too far, like pushing things in a weird direction, I guess, and so I think, unlike some other AI industries which can get really touchy in marketing, you can be a little bit more fun, and also, like that customer who is like responding to the marketing, tends to keep things kind of like within a realm of

1185
03:23:48.791 –> 03:23:55.319
Zack Holland: being somewhat governanced by itself, so I think it requires a little bit less than like some of the other industries just out of the gate.

1186
03:23:57.120 –> 03:23:59.359
Mark Organ: Yeah, any other thoughts on regulation, for

1187
03:23:59.610 –> 03:24:10.169
Mark Organ: before we get it before we get into a speed round, one more thought on regulation. I mean, I’m sure the EU will find a way to mess things up because they always do.

1188
03:24:11.567 –> 03:24:12.969
Mark Organ: It’s like that.

1189
03:24:14.230 –> 03:24:24.100
Chris Hicken: But you know, like a smart AI can infer a lot about you, even if you haven’t said something specifically. So you know, there’s some questions there about how do you

1190
03:24:24.280 –> 03:24:27.830
Chris Hicken: treat the data that you’ve inferred.

1191
03:24:28.310 –> 03:24:37.230
Chris Hicken: you know, like medical conditions, or where you live, or what your job is? How do you protect that information which AI will get from you over time?

1192
03:24:40.110 –> 03:24:50.979
Mark Organ: Yeah, awesome. All right. Quick. Speed around the top of the hour. What’s the best new AI tool that you have that you use in your own personal workflow that you’ve adopted in the last 2 months.

1193
03:24:51.310 –> 03:24:52.360
Mark Organ: Chris, go.

1194
03:24:53.990 –> 03:25:14.789
Chris Hicken: Okay. Best tool for us has been clickup. It’s a tool that combines all of your tasks, chat documents, everything into one place to your point, Kingston. All the data is in one place, and when you have all the data in one place, AI is extremely helpful for figuring out for automation, but also for finding stuff

1195
03:25:14.980 –> 03:25:24.489
Chris Hicken: for making sure that you get, you know, updates pushed to to you automatically to make sure that you know what’s going on at the company. Phenomenal phenomenal tool.

1196
03:25:24.490 –> 03:25:27.570
Mark Organ: Awesome. Alright! That’s awesome. Zack!

1197
03:25:28.265 –> 03:25:51.890
Zack Holland: Been a huge fan of a new platform called Flora on the creative engineering side. So people who are saying like trying to develop ad copy instead of having to build something for image generation, pull it off, stick it in Photoshop and go ahead and send it to your designer. You can build custom workflows in Flora to make some really beautiful stuff using a ton of different models. Their branding is also incredible, and I’m a brand nerd, and so been a big fan of Flora recently.

1198
03:25:51.930 –> 03:25:54.129
Mark Organ: All right. Let’s check it out. Kickstin.

1199
03:25:54.740 –> 03:25:58.189
Kingston Duffie: Well, I might surprise you, but I I

1200
03:25:58.360 –> 03:26:04.960
Kingston Duffie: I’ve come back to chat, Gpt itself and realized that I was only using

1201
03:26:05.350 –> 03:26:09.970
Kingston Duffie: 5% of its capability. Like, I think, if you really start

1202
03:26:10.610 –> 03:26:14.160
Kingston Duffie: diving deep, you realize how how far that tool can go.

1203
03:26:15.250 –> 03:26:16.730
Mark Organ: Awesome. Megan.

1204
03:26:17.130 –> 03:26:36.950
Meagen Eisenberg: Yeah, I’m a fan of Senja, SENJ. A. It allows you to crowdsource across sources and collections of information and then display it on a website. You could go to samsara.com forward, slash upgrade! If you scroll down, you’ll see all these different cool quotes coming in that have been sourced across different sources.

1205
03:26:38.620 –> 03:26:40.760
Mark Organ: That’s great. Did we get everybody.

1206
03:26:41.380 –> 03:26:42.149
Zack Holland: And get Amanda.

1207
03:26:42.450 –> 03:26:42.949
Amanda Kahlow: You didn’t get.

1208
03:26:42.950 –> 03:26:45.370
Mark Organ: Amanda. How could we, miss Amanda?

1209
03:26:45.900 –> 03:27:04.780
Amanda Kahlow: I’m okay to be missed. My answer is really boring, because I’m just gonna say that the most exciting thing I’ve been doing with AI is to manage my kids like school calendar, and all of their apps of like what the fuck do I need to do? And when I bring them a shoebox to school tomorrow, because I can’t read it all. So it just helps me give me the notifications of like, keep me on task.

1210
03:27:05.260 –> 03:27:18.670
Mark Organ: That is great, and my my favorite tool last couple months has been Grok. I love it. I think it’s incredibly powerful. Use it for all kinds of things, but that is the end of our end of our panel, Julia, back to you.

1211
03:27:19.330 –> 03:27:26.520
Julia Nimchinski: Thank you so much, Mark. Thank you. All of our panelists. Such an incredible session. We’re receiving a lot of messages in our slack

1212
03:27:26.760 –> 03:27:40.259
Julia Nimchinski: side. Note, we had to redistribute our just live stream. Some people are joining us on the website. Some people are joining us on Zoom to those of you who are in zoom. Please do contribute to the slack.