Fireside Chat with Wade Foster & Jonathan Kvarfordt — The Rise of Agentic Distribution

00:00:06.240 –> 00:00:08.170
Julia Nimchinski: And we are live!

2
00:00:08.620 –> 00:00:17.159
Julia Nimchinski: Welcome to the Agentic Distribution Summit. Over the next 3 days, we’ll explore how Argentic Distribution is redefining GTM.

3
00:00:17.550 –> 00:00:35.989
Julia Nimchinski: And adjusting distribution is an emerging architecture of GTM where AI agents act as both the interface and the operator. We’re excited to hear from leading CXOs, analysts, VCs, and AI-native founders who are rethinking distribution.

4
00:00:35.990 –> 00:00:43.139
Julia Nimchinski: From AI First Principles. And, before we kick things off, just follow along the conversation.

5
00:00:43.290 –> 00:00:49.870
Julia Nimchinski: in our HSC Slack, and don’t forget to check out the sponsor on the right corner of your screen.

6
00:00:50.750 –> 00:01:01.989
Julia Nimchinski: And welcome, Wade Foster, one and only the CEO of Zapier, and Jonathan Crawford, head of GTM Growth at Momentum.

7
00:01:02.470 –> 00:01:06.990
Julia Nimchinski: The Rise of Agent Distribution. Excited to host you. How have you been?

8
00:01:08.040 –> 00:01:09.420
Wade Foster: Great, thanks for having me, y’all.

9
00:01:11.070 –> 00:01:16.459
Jonathan Kvarfordt: Yeah, super excited to be here, and I get to pick on Wade’s brain. I’m, like, feeling all giddy already, this is gonna be fun.

10
00:01:17.280 –> 00:01:18.960
Julia Nimchinski: Awesome! Let’s get into it.

11
00:01:20.530 –> 00:01:21.320
Jonathan Kvarfordt: Let’s do it.

12
00:01:21.550 –> 00:01:29.610
Jonathan Kvarfordt: Do we need quick intros real fast? Not that Wade needs introductions, but I’d love to hear from you, Wade. Like, obviously you’re the CEO of Zapier, but give us the… I’d love to hear a little bit more of your background.

13
00:01:30.150 –> 00:01:39.779
Wade Foster: Yeah, Zapier founded the company in 2011, so 14 years ago now, just out of school, so I’d only been in the workforce for a year. I was doing…

14
00:01:40.050 –> 00:01:51.149
Wade Foster: Whole bunch of, like, oddball internet stuff at the time. Founded the company to solve integrations, so we made it easy to connect stuff. But in 2016, the company really…

15
00:01:51.680 –> 00:02:09.629
Wade Foster: sort of came out of its shell with our workflow product, being able to chain these integrations all together to solve workflow. And it turns out, workflow is really valuable in the age of AI. And so, you know, in the last couple years, we have been spending a lot of time

16
00:02:09.870 –> 00:02:14.040
Wade Foster: building AI capabilities into our core product, but alongside that.

17
00:02:14.200 –> 00:02:20.549
Wade Foster: We’re dogfooding this stuff just as much as anybody is, using our own product, using other products, to figure out how to…

18
00:02:21.280 –> 00:02:22.900
Wade Foster: grow, at the end of the day.

19
00:02:24.220 –> 00:02:40.649
Jonathan Kvarfordt: Yeah, I love it, and I’m obviously the head of growth and momentum, which we do a lot of, kind of, automations, and a little bit different use case than Zapier is, but, very similar in a lot of ways, aligned in a lot of ways, too. We actually use Zapier internally for a lot of stuff, because, like, Momentum’s not the hero-be-all, like, Zapier, so we love and use Zapier internally.

20
00:02:40.650 –> 00:02:44.529
Jonathan Kvarfordt: And, then I do the GTMEA Academy, where I teach

21
00:02:44.530 –> 00:03:02.419
Jonathan Kvarfordt: 10,000 plus people in the last 3 years, how to use AI, different stuff, and, like, one of the first places I have them go to is Zapier, with AI agents, saying, if you want to know an easy, very user-friendly way to learn how to use AI agents, Zapier’s, like, the number one place they send people to, which is awesome. So, I’m really excited to go into it with you. Well, tell me this way.

22
00:03:02.590 –> 00:03:21.169
Jonathan Kvarfordt: So I think that, I know you guys came out with the announcement, I watched your video, which you kind of talked about the need for orchestration. I think people get really confused between, like, the basics of, like, okay, we understand what automation is, how would you determine the difference between just plain automations, where Zapier has been living for such a long time, and now where it’s going to orchestration? How do you define that?

23
00:03:21.870 –> 00:03:26.500
Wade Foster: Well, yeah, I think, it is confusing, because in the past.

24
00:03:26.500 –> 00:03:26.950
Jonathan Kvarfordt: Yes.

25
00:03:26.950 –> 00:03:44.069
Wade Foster: you know, automation was purely deterministic. And so, you know, this is kind of your classic if-this, then that, else this type logic, where you just sort of chain a series of actions, and in our case, they’re API calls, together to produce an outcome.

26
00:03:44.180 –> 00:03:48.700
Wade Foster: And, you know, if you’re looking at, an agentic world.

27
00:03:48.840 –> 00:03:56.009
Wade Foster: It doesn’t quite work that way. Instead, you’re giving an AI a task, and then you’re letting the AI decide

28
00:03:56.150 –> 00:04:13.109
Wade Foster: what should it go do? They decide what to go figure out, almost like you would delegate to another human and say, hey, can you go figure this thing out? They may not do it the way you prescribe, but you’re hopefully getting the outcome. Now, what orchestration is, is it turns out, in an agentic world.

29
00:04:13.320 –> 00:04:20.360
Wade Foster: it’s actually really hard to delegate that context to an AI and expect good outcomes.

30
00:04:20.360 –> 00:04:35.889
Wade Foster: And so what’s really working today is the more you can actually chain, or provide an AI with context, and start to chain actions along, you actually get much better outcomes. And so the irony is this deterministic workflow,

31
00:04:35.890 –> 00:04:50.219
Wade Foster: is actually really, really valuable when you can add AI to it piecemeal. In very narrow cases, you tend to get a lot higher reliability, a lot higher consistency, and it opens up a whole new world of use cases. So you’ve sort of got, like, deterministic

32
00:04:50.220 –> 00:05:07.009
Wade Foster: flows on one side, inference flows on the other, and then this, like, big blurry middle of, like, how much are you using AI, how much little are you using AI, and it turns out the middle, there’s a ton of value there, even though the far… the AI side of the house is what’s kind of, like, sexy and gets all the headlines.

33
00:05:07.890 –> 00:05:31.690
Jonathan Kvarfordt: Yeah, I know I sent you a bunch of prep questions, but I’m gonna go down so different paths, so I know you’re gonna be game for some good stuff, but just curious, we can start with some meaty stuff. Like, what are the things that you’re doing internally with Zapier for Zapier that you’re, like, thinking, hey, this is cool, and it may not apply to anybody else, but, like, how do you go about thinking… let’s start with this, actually. How do you go about syncing with your team? Where do you slap in Zapier for your own workflows or agentic automations?

34
00:05:31.690 –> 00:05:32.060
Wade Foster: Yeah.

35
00:05:32.060 –> 00:05:35.280
Jonathan Kvarfordt: What’s the funnest, coolest, most interesting thing you guys have done so far?

36
00:05:35.710 –> 00:05:53.819
Wade Foster: Well, so, I think we use Zapier and AI all over our go-to-market process. I think the most obvious place that we have put AI, which I think if you’re not doing this yet, this is, like, the first thing you should do. Like, there’s no excuse to not do this, which is, sales prep.

37
00:05:54.000 –> 00:06:04.220
Wade Foster: You know, if your sales reps are getting on a call with a prospect, and they don’t have a brief that combines all the context that the internet knows about them, what you have in your CRM, what you have in your support desk.

38
00:06:04.840 –> 00:06:09.279
Wade Foster: you need to stop and do that right now. It’s one of the easiest things you can set up and build today.

39
00:06:09.280 –> 00:06:34.279
Wade Foster: It really just requires triggering on, you know, calendar invites. If, you know, go look at my calendar invites for the day, go do deep research on this prospect, pull what you have out of CRM, pull what you have out of your, support ticket, and then send me an email, or send me a Slack message, or generate a PDF and deliver it to me, so that those sales reps walk in equipped with actual information. It’s one of the most basic, simple use cases, but very, very, very high value.

40
00:06:34.280 –> 00:06:35.610
Wade Foster: you.

41
00:06:35.720 –> 00:06:47.809
Wade Foster: So that’s, like, on one end of the spectrum, what’s really easy to get started. Then, I think if you want to look at, kind of, like, the other end of the spectrum on stuff that I think is really interesting, but we’re still figuring out, is…

42
00:06:48.290 –> 00:07:02.789
Wade Foster: you know, it turns out one of the most powerful assets that companies have for ranking well in things like ChatGPT, or Claude, or Gemini, or these tools, is your help desk, your knowledge base.

43
00:07:03.020 –> 00:07:18.330
Wade Foster: folks are asking these long-tail sets of questions. And, you know, if you’re like most businesses, your knowledge base, your help to docs, they’re probably not, like, well-kept, up. Partly because in the past, like, they just didn’t get a lot of…

44
00:07:18.440 –> 00:07:20.670
Wade Foster: Traffic, a lot of,

45
00:07:21.150 –> 00:07:30.930
Wade Foster: value, and so unless you were crazy disciplined about doing a good job with this, I don’t think most companies really invested a lot in it. Right. But nowadays, you can actually build…

46
00:07:30.980 –> 00:07:46.570
Wade Foster: tools that make it a lot easier to have a much richer help desk. So, for example, you can take every ticket that you’re answering in email or chat, and say, hey, anytime a customer is asking us a question, I want to take the

47
00:07:46.570 –> 00:08:01.329
Wade Foster: answer that the support rep gave, and I want it to go generate a help doc. Now, you probably need to still have a human in the loop there to verify, do we already have a help doc for this or not? Is it accurately stripping out PII and things like this?

48
00:08:01.670 –> 00:08:14.530
Wade Foster: You can do something pretty impressive. Similarly, you can do the same thing for Gong calls. So if you’re on calls with your prospects all the time, and your prospects are answering questions, hey, do you all do this? Do you all do that? We’ll take the transcript.

49
00:08:14.640 –> 00:08:29.170
Wade Foster: from that transcript, extract all the commonly asked questions, generate a help doc for it. And so now you’re just building up this very rich catalog of long-tail questions that your customers are asking in a pretty simple way.

50
00:08:29.170 –> 00:08:39.759
Wade Foster: And all of those things are now starting to get indexed by Google, of course, but also these LLMs. And in the LLMs, I find that humans tend to ask these, like, deeper questions.

51
00:08:39.760 –> 00:08:43.980
Wade Foster: than you probably were happening on Google. And so, you know, that’s the kind of…

52
00:08:43.990 –> 00:08:45.419
Wade Foster: You know, if you start to get…

53
00:08:45.550 –> 00:09:04.770
Wade Foster: go down this rabbit hole, you start to uncover, like, all these nooks and crannies that have maybe been overlooked, or not properly, you know, like, addressed in the past. Not because there wasn’t value there, but because it was really hard, and now there’s actually some tools to make it easier to do this stuff, and it turns out if it’s easier to do it, it’s worth actually investing in.

54
00:09:05.730 –> 00:09:22.329
Jonathan Kvarfordt: Do you feel like, going off that use case, that there seems to be kind of a trend around empowering from the bottoms up of, like, giving individuals, you know, tools like Zapier, whatever the case might be, to develop their own workflows, and then more going the top-down, because obviously leaders have a different view of what’s happening in the organization. So.

55
00:09:22.380 –> 00:09:35.139
Jonathan Kvarfordt: What’s your approach? When you have companies come in and say, okay, Wade, we have all this stuff, we could then automate all these bajillion little tasks we could do. Where does the revenue lift come from? Does it come from the bottoms up and saving time? Is it coming from top down? Like, what do you see as the difference?

56
00:09:35.140 –> 00:09:53.979
Wade Foster: Well, I think it’s both, but what I recommend to folks is, I do think one of the first places to start is to do hackathons, to do bootcamps, to start to empower your entire organization to put their hands on keyboards, to figure out, like, what this technology is actually capable of. And you start to get, then.

57
00:09:53.990 –> 00:10:09.369
Wade Foster: momentum around this. Your employees start to get excited about it. The fear of AI starts to dissipate, and the excitement starts to slip in. And that’s a really important, like, cultural moment to get over, because the, you know, mainstream press would have you think, AI’s coming for my job.

58
00:10:09.370 –> 00:10:25.839
Wade Foster: But if you lean in and learn this stuff, I promise you, your skills are gonna be valuable. Every company I know wants to hire people who are good at AI right now. Like, that is one of the most valuable things, and so once you get your people into it, you can start to get a lot of, like, really interesting ideas bubbling up, bottoms up.

59
00:10:26.030 –> 00:10:30.859
Wade Foster: Now, that, I think of as raising the floor inside your organization.

60
00:10:31.140 –> 00:10:37.440
Wade Foster: Right. Once you have that, I do think that’s where it’s very helpful for, you know, leadership teams.

61
00:10:37.840 –> 00:10:53.089
Wade Foster: executive teams to come in and sponsor, like, a handful of, like, high-impact use cases. And the place to look is the places that are really difficult for a, you know, individual in the organization to go make it happen. So, for example.

62
00:10:53.090 –> 00:10:59.739
Wade Foster: One of the themes that’s talked a lot about in AI is the combination of jobs. There’s certain skills that are getting squished together.

63
00:10:59.750 –> 00:11:14.609
Wade Foster: And that’s a really difficult thing for, you know, an IC to just say, hey, I’m, you know, I wanna… I wanna be both a designer and a marketer at the same time, or I wanna be both an engineer and a product manager at the same time, and I wanna squish the job families here inside this company.

64
00:11:14.610 –> 00:11:28.589
Wade Foster: Like, that’s a weird thing for an IC to figure out, but it’s also a painful thing to decide to do. And that does require a leader in the company saying, hey, I see an opportunity here for us to do something that really

65
00:11:28.650 –> 00:11:44.819
Wade Foster: upends how we’ve done stuff in the past, but we have enough examples. You know, there was a hackathon project that showed what was possible here. We have an individual inside the company that shows that this is possible, and we believe that we can shift from how we have been doing things into the past to this new way. So here’s how we’re gonna do it.

66
00:11:44.820 –> 00:11:58.439
Wade Foster: And, you know, it does take a leader to say, I’m gonna sponsor that, and I’m gonna take… I’m gonna go through the pain to get to this better spot in the future. And so that’s where I think the tops-down stuff can be really, really valuable, is… is trying to go solve those types of problems.

67
00:11:58.830 –> 00:12:09.810
Jonathan Kvarfordt: Totally. Do you feel like there’s… in my experience talking with people with AI, both leaders and individuals, a lot of times it’s some… it’s the unsexy work of, like, documenting the process, or…

68
00:12:09.980 –> 00:12:28.220
Jonathan Kvarfordt: Understanding the ICP, things that AI can assist you with, but it’s not going to create without some sort of human element. Once you have that kind of, again, the unsexy, unglorified work, then AI has really, really powerful, it’s like, almost like a slingshot. You can pull it back, do all this hard work, you don’t feel like you’re going forward, and you release it, and you just fly it really quickly.

69
00:12:28.220 –> 00:12:38.509
Jonathan Kvarfordt: Do you see that a lot with yourself, like, with Zapier, people saying, hey, let’s make… give us all this pipeline. You’re like, great, let’s slow down before we go to pipeline, and let’s work on some basics. How do you kind of help people through that?

70
00:12:38.510 –> 00:12:56.019
Wade Foster: Yeah, I definitely think that’s a really common challenge here. Everyone wants to just jump to the finish line of, like, I want an agent that’s just gonna grow my social media, or I want an agent that’s gonna drive around Pipeline, or I want an agent that’s gonna do this. And, that’s not really where agents are at today.

71
00:12:56.230 –> 00:13:15.209
Wade Foster: you know, I think what’s really helpful is what you just described, is, okay, I need to break this problem down into smaller pieces. And so, the people who are really good at this stuff tend to be strong systems thinkers. They’re thinking through, okay, I need this bit of data, I need this bit of context, you know, step one, step two, step three, here’s how this all should work.

72
00:13:15.690 –> 00:13:26.829
Wade Foster: okay, this data’s a little messy, I gotta go figure out how to clean it up first so the AI can take advantage of it. So they’re doing that sort of step, and then figuring out how to orchestrate it step by step.

73
00:13:27.050 –> 00:13:31.980
Wade Foster: And I think the reality is, it’s often easier to start with

74
00:13:31.980 –> 00:13:51.869
Wade Foster: traditional automation, traditional workflows, and then say, hey, we’re gonna go, add AI to it. So, for example, a really common workflow for Zapier in the past has been, I want to take a lead from, say, maybe, like, Facebook. I’m doing Facebook advertising, I get a lead there, and I want to save it to a CRM. That has forever been, like, a really popular use case on Zapier.

75
00:13:51.890 –> 00:13:59.649
Wade Foster: Well, it turns out, if you want to get started with AI, a really great way to change that workflow is to say, hey, let’s take that lead from Facebook.

76
00:13:59.750 –> 00:14:07.509
Wade Foster: But let’s also now go do enrichment on it. You know, in the past, you might hit an enrichment API. Now, you might say, hey, I want an AI to do deep research on it.

77
00:14:07.600 –> 00:14:10.310
Wade Foster: Okay, great. Now let’s save it to the CRM.

78
00:14:10.380 –> 00:14:26.720
Wade Foster: And all of a sudden, one small change, like, really tiny change, you’re not building a complicated agent or anything like that, but the value you got from that one simple automation has gone up massively, because you have all this extra context on that lead that’s coming from Facebook, and you never had to ask the person.

79
00:14:26.720 –> 00:14:30.979
Wade Foster: And so those are the types of ways that you can get started.

80
00:14:30.980 –> 00:14:44.780
Wade Foster: And once you start doing those little tiny steps, you’ll find it’s much easier to build to the more sophisticated things. And you also might find that, like, hey, we actually don’t need the sophisticated things. Like, these, like, traditional workflows actually work pretty good for us.

81
00:14:45.850 –> 00:14:55.519
Jonathan Kvarfordt: I think part of the balance is a lot of people are seeing that there is that place where it’s not… you don’t always need an AI agent, sometimes it’s just a simple automation. But I think once you see the power of, like.

82
00:14:55.590 –> 00:15:13.049
Jonathan Kvarfordt: The difference between saying, okay, send me an email every time a lead comes in, versus orchestration layer, which is qualify it, deep research it, route it to the right person, personalize it, all these steps happening at once in conjunction, so it’s not in siloed conversations, but the AI has an understanding of what’s happening everywhere. That’s where I think the power comes in, that…

83
00:15:13.050 –> 00:15:20.969
Jonathan Kvarfordt: I think people should get an idea of, like, oh, like, once you get your hands on it, like, this is really cool. It’s a lot of fun stuff you can do that saves, obviously, time, but then the quality goes up, too.

84
00:15:20.980 –> 00:15:22.160
Jonathan Kvarfordt: You feel like that way, too?

85
00:15:22.460 –> 00:15:29.880
Wade Foster: Well, yeah, and it’s… it’s… instead of thinking almost like a master agent, it’s like you’ve got all these tiny agents, where it’s like, I’ve got my…

86
00:15:29.880 –> 00:15:30.230
Jonathan Kvarfordt: Right.

87
00:15:30.230 –> 00:15:43.619
Wade Foster: research agent, I’ve got my QA agent, I’ve got my email writing agent, I’ve got my… like, each of them does a tiny, tiny little step, and then, you know, you can plug those into workflows, or you can create tools and give those tools to an agent that now can call them

88
00:15:43.620 –> 00:15:57.450
Wade Foster: and do something interesting with them. Either can work, you’re gonna get different pros and cons based on, sort of, how you go assemble that. But, you know, I do think that the power is when you make them small, and make them very good at a particular task.

89
00:15:57.940 –> 00:15:58.530
Jonathan Kvarfordt: Yeah.

90
00:15:58.940 –> 00:16:03.219
Jonathan Kvarfordt: I’m gonna shift gears just to Hera, because I’m curious from your lens as, obviously a leader.

91
00:16:03.320 –> 00:16:16.469
Jonathan Kvarfordt: Back in April is when Shopify CEO came out and said, like, hey, we’re not gonna hire anybody unless you can tell me that AI can’t do it. And then, you know, Mark Benioff of Salesforce has said, like, they’ve automated 50% of their workflows, Duolingo’s come out, like, there’s all this…

92
00:16:17.160 –> 00:16:30.530
Jonathan Kvarfordt: talk, and then I would say undoed, like, not meaning to have this happen, but there’s pressure on people to say, all these other companies are doing this amazing, cool AI stuff, now I gotta do it, right? How are you thinking about, or how do you recommend people to think about

93
00:16:30.680 –> 00:16:37.930
Jonathan Kvarfordt: where they are today to get to that level without feeling overwhelmed, so they can feel like they are getting to an AI-powered

94
00:16:38.060 –> 00:16:38.700
Jonathan Kvarfordt: Reality.

95
00:16:38.700 –> 00:16:49.830
Wade Foster: Well, yeah, so we went through this in 2023. I Code Red internally, which was sort of our similar version of, you know, Toby’s memo, saying, like, hey, we gotta go live on our AI future.

96
00:16:49.940 –> 00:16:53.139
Wade Foster: Now, I’ll be the first to say, like, this was…

97
00:16:53.740 –> 00:17:01.250
Wade Foster: not necessarily, like, a popular decision at the time, internally. In fact, I got called sensational, alarmist, etc.

98
00:17:01.370 –> 00:17:02.380
Wade Foster: But…

99
00:17:02.900 –> 00:17:15.040
Wade Foster: One of the things that was really powerful that we did alongside that, one of the first things, and this is one of the things that really worked, was saying, hey, we’re gonna do a hackathon for everyone in the company. Everyone’s gonna put their hands on this technology and start to use it.

100
00:17:15.200 –> 00:17:31.130
Wade Foster: And what we saw was, pre and post that hackathon, we went from, you know, around 10% of people using AI daily, to now north of 50% using AI daily. So just one hackathon right at the beginning, all of a sudden there was this huge growth rate in adoption.

101
00:17:31.420 –> 00:17:37.410
Wade Foster: Right. Then, over the course of a year, we repeated that hackathon, two other times.

102
00:17:37.610 –> 00:17:52.620
Wade Foster: And then alongside that, as part of our all-hands, we always did show and tell. Show-off use cases. We found people from every part of the org, not just engineers, we found, you know, people in the people org, people in the marketing org, people in the sales org, people in accounting, just to show off. What have you built this week? What have you built this week?

103
00:17:52.630 –> 00:17:58.049
Wade Foster: And all that effort created a lot of knowledge sharing, a lot of bottoms-up understanding, and a lot of…

104
00:17:58.420 –> 00:18:02.150
Wade Foster: Just, like, peer-to-peer support on, like, how do you go figure this stuff out?

105
00:18:02.360 –> 00:18:06.909
Wade Foster: And, you know, after about a year, our AI adoption went from

106
00:18:06.940 –> 00:18:29.339
Wade Foster: you know, about 10% to 90%. So, you know, at the beginning, you were weird if you were using AI. Now, it’s the opposite. You’re kind of weird if you’re not. It’s like, wait, what are you doing here, you know? Right, right. And so, that, to me, is how you really make the shift, is you gotta invest in folks. And I think one of the biggest places I see CEOs and other leaders falling down is they’re mimicking

107
00:18:29.340 –> 00:18:33.830
Wade Foster: Toby and others with a memo. They’re saying, hey, we need to go be AI first.

108
00:18:33.830 –> 00:18:36.719
Wade Foster: Which, I’m all for that, I think that’s a good move.

109
00:18:36.720 –> 00:18:54.989
Wade Foster: But then they stop there. They’re just sort of like, everyone just go figures it out. And then what happens is the delegation trap. It gets pushed down to the executive layer, and the executive layer’s, well, shoot, I’m not exactly sure what to do. I know what I’ll do. I’ll delegate it to somebody on my team to figure out AI. And then that person’s like, well, shoot, I’m not really sure what to do, I’ll delegate it to somebody on my team.

110
00:18:54.990 –> 00:19:12.799
Wade Foster: it’s down to the end of the organization, and everyone’s just kind of trying to figure it out on their own. And, you know, some people do, and a lot of people don’t. A lot of people are like, well, do I figure out the AI thing, or do I try and hit my OKR for the month? And so you’re just sort of caught in this, like, whirlwind trap, where I think it’s way more valuable if

111
00:19:12.800 –> 00:19:24.620
Wade Foster: the CEO and the leadership team say, nope, we’re gonna put some effort behind this, and you know what? We’re gonna do it too. We’re gonna put our hands on the keyboard, we’re gonna learn this stuff too, and it just makes it a lot easier for the organization.

112
00:19:24.960 –> 00:19:34.769
Wade Foster: To make the shift together as a team, versus sort of, like, one person has to go figure this whole thing out, or, like, you know, the sprawling tendrils of an organization have to figure it out.

113
00:19:35.590 –> 00:19:55.279
Jonathan Kvarfordt: Yeah, there’s… there’s this world where I think that in those type of memos of saying, go figure it out, or else, you know, it’s like, okay, how am I gonna do this? Like, there’s so much noise, and so much different platforms that do different things, and so figuring out, okay, do I do it as a Salesforce agent? Do I do a Glean agent? Do I do a Zapier? Like, what do I do? Each platform does things differently, so it’s like…

114
00:19:55.280 –> 00:20:00.739
Jonathan Kvarfordt: having two things. One, a leader that is clear like yourself, and then giving people the space to play without

115
00:20:01.050 –> 00:20:14.440
Jonathan Kvarfordt: fear of getting fired, like, just go play on a simple, basic workflow, it’s not gonna break anything, but that gives you a taste of what’s happening, I think is really important. It’s just that a lot of people, like you said, do the memo, and that’s as far as it goes, which is sad to see that.

116
00:20:14.440 –> 00:20:14.810
Wade Foster: Yeah.

117
00:20:15.370 –> 00:20:15.950
Jonathan Kvarfordt: If I can.

118
00:20:15.950 –> 00:20:26.820
Wade Foster: I think the show and tell part, I think, is also really powerful. Like, show and tell creates a lot of knowledge sharing, because now everyone is seeing, like, oh, what did you do? How did you do that? Like, what tool did you use?

119
00:20:26.930 –> 00:20:36.749
Wade Foster: And, like, that’s perfectly understandable, like, you know, it… I think all of us feel a certain amount of pressure, and, like, we’re behind, but reality is, like, this stuff is still crazy new. And so…

120
00:20:36.750 –> 00:20:38.120
Jonathan Kvarfordt: It’s really helpful.

121
00:20:38.120 –> 00:20:42.699
Wade Foster: to just see what other people are using, and what other people have figured out. It’s not like there’s…

122
00:20:43.260 –> 00:20:47.550
Wade Foster: Totally. A huge amount of, like, training resources and material for this. There’s a lot of just…

123
00:20:47.560 –> 00:20:54.139
Jonathan Kvarfordt: trial and error and discovery, and so that knowledge sharing, that show and tell aspect, I think, is really helpful right now.

124
00:20:54.710 –> 00:21:00.580
Jonathan Kvarfordt: In the Academy, the sessions that have been the most successful with any AI tech has been the ones where

125
00:21:00.650 –> 00:21:20.289
Jonathan Kvarfordt: someone showed something, and they’re like, okay, how’d you do that? And then we break down the prompt and the workflow and stuff, and then people are like, oh, then he starts to connect the dots, which is awesome. I’m gonna ask a question for Reggie, but before I do that, curious for you, since 23, what has been the business impact of doing that strategy of that stuff, and now everyone’s implementing AI, but what are you seeing business-wise as a result from that?

126
00:21:20.910 –> 00:21:36.659
Wade Foster: Well, you know, I think, obviously, internally, we have AI, like, 90-plus percent of the staff is using AI. In fact, I think it’s 97% of our staff is using AI daily, 100% uses it, you know, weekly or monthly. So, you know, you can imagine just, like, all the, like.

127
00:21:36.690 –> 00:21:52.279
Wade Foster: places inside the company where we’re reaping the productivity benefits that even I’m just not even aware of. And then, of course, there’s things like what I described, the, you know, daily sales briefs, the help doc use case, you know, the code generation, all those things are happening.

128
00:21:52.280 –> 00:22:04.670
Wade Foster: Then on the product side of the house, you’re seeing the product just get a lot better at embedding those capabilities in for our customers to use. The end result there is we’ve done something like 350 million AI tasks for folks.

129
00:22:04.670 –> 00:22:13.429
Wade Foster: The revenue for folks using AI is growing, like, 300% year over year. So there’s, like, a very real impact on

130
00:22:13.430 –> 00:22:20.350
Wade Foster: The business, because customers are now able to benefit from our ability to learn, and then encode that stuff in the product for them.

131
00:22:21.540 –> 00:22:39.799
Jonathan Kvarfordt: Yeah, and I can… I’m gonna put a plug in for your own tech for a second, but you guys have done a really amazing job of documenting the AI agent templates, which is really nice, because then people can go in and use the templates, which is so nice, because when you don’t know when to start, you’re like, just go to the template, pick one, just run it through, and make it work in your workflow, which is awesome, you know? Yeah, show and tell.

132
00:22:40.060 –> 00:22:53.980
Jonathan Kvarfordt: Show and tell. So Reggie, Jose on the chat, sorry if I didn’t say your name right, asked the question, if campaign execution could be fully automated, where should human sales leaders focus to create strategic advantage instead of operational drag? What do you think?

133
00:22:54.210 –> 00:22:55.790
Jonathan Kvarfordt: with campaign execution.

134
00:22:55.990 –> 00:22:59.879
Wade Foster: If campaign execution, you know, I think…

135
00:23:00.540 –> 00:23:03.109
Wade Foster: I think the place where,

136
00:23:04.140 –> 00:23:22.270
Wade Foster: you know, humans are really still involved is thinking through, like, well, what exactly are you executing in that campaign? There still is a lot of steps involved that you have to think through around, like, what do we actually want to get out of this? And then the other place where there’s a lot of work is then auditing. Hey, is this actually…

137
00:23:22.800 –> 00:23:32.329
Wade Foster: being automated in a way that is delivering value for us. And that, I think, is one of the things that folks quickly realize when they start to deploy these AIs, is that

138
00:23:32.330 –> 00:23:35.819
Wade Foster: Yes, it’s fully automated, but the audit part of it

139
00:23:35.820 –> 00:23:58.569
Wade Foster: takes a lot of work. There’s a lot of just, like, waking up, and, you know, first thing you do is go look at the logs, look at the leads that came in, look at everything that happened, be like, did it work? Did it work? Yep, okay, that one’s working, that one’s working. That’s not quite right. And then, of course, these models, they’re constantly changing, and so there’s a lot of work to do to be like, okay, can we make this prompt better? Can we make it better in that way? Oh, the model changed some way, so, like, do we have to make some tweaks?

140
00:23:58.570 –> 00:24:07.339
Wade Foster: on it, so there still is a lot of just, like, in-the-details, executional bits. And so the people that I think are excelling the most

141
00:24:07.340 –> 00:24:24.050
Wade Foster: are folks that do have that strategic mindset around, like, what is a great campaign? What, like, do I understand my customers? Do I understand human psychology, but that can also translate that into these, like, tactical, step-by-step bits that really, make a campaign hum at the end of the day.

142
00:24:24.990 –> 00:24:35.129
Jonathan Kvarfordt: I have a friend of mine who texted me, I’m gonna ask her a question that she asked me to ask you. So, you guys put out, I guess you could say a maturity model of different, positions, titles, of

143
00:24:35.470 –> 00:24:38.539
Jonathan Kvarfordt: You know? And so, I’m curious, in general.

144
00:24:38.860 –> 00:24:51.459
Jonathan Kvarfordt: what do you see as the key workflows? Because, like, we talked about before how people are saying how AI is going to replace humans and all this other shenanigans, right? When you’re hiring for your team internally with Zapier, because you guys are obviously so automation and AI forward.

145
00:24:51.650 –> 00:25:02.340
Jonathan Kvarfordt: what are the key things that’s human-based? Obviously, understanding automation is AI is one thing, but what do you look for saying, I need someone to have this to fit in and make sure they can use Zapier in the best way possible? What’s the human stuff you need?

146
00:25:03.020 –> 00:25:10.339
Wade Foster: I mean, it’s… it’s still, like, a lot of the similar things that you’re looking for in the past. Like, you’re looking for somebody who…

147
00:25:10.730 –> 00:25:20.939
Wade Foster: is a problem solver, you know, is close to the craft, is fresh on the tools. Yeah, I think one of the most challenging things right now is

148
00:25:21.620 –> 00:25:24.010
Wade Foster: Particularly when you’re trying to hire leaders.

149
00:25:24.600 –> 00:25:30.090
Wade Foster: Is a lot of leaders get… Abstracted away from the craft.

150
00:25:30.480 –> 00:25:37.079
Wade Foster: And the craft has changed a lot, and is changing a lot. And so you’re really just trying to understand

151
00:25:37.460 –> 00:25:48.989
Wade Foster: Have they stayed fresh on this? And so that’s where it’s just really helpful to ask them questions like, what are you playing around with right now? Can you show me what you’re building? And the people that are actually building things.

152
00:25:49.200 –> 00:26:05.979
Wade Foster: they usually get crazy excited about it. They’re like, oh, I gotta show you what I made yesterday, like, this is so cool, like, let me show this. And they can’t help it. And then there’s… the folks you can tell who aren’t doing that, they start to get a little nervous. You’re like, hey, can you share your screen and just show me a thing? And they’re like, well, you know…

153
00:26:05.980 –> 00:26:06.680
Jonathan Kvarfordt: You’re like, oh!

154
00:26:06.680 –> 00:26:14.100
Wade Foster: I don’t think you’re actually trying this stuff. And so, that’s where I think it’s just really helpful. You’re just trying to just…

155
00:26:14.380 –> 00:26:23.589
Wade Foster: sense out, like, who’s got curiosity? Who’s solving problems? Who’s got good ideas, for how these things work? And that’s usually just… that’s usually enough to stand out right now.

156
00:26:24.820 –> 00:26:44.719
Jonathan Kvarfordt: I know we have, 2-3 minutes left, so I’m gonna ask a large question. I’m not sure if you have time to answer it, but obviously a lot of people on the line are go-to-market leaders, like, you’re talking marketers, rev ops, CROs, etc. And a lot of people are asking the question around… I hear this from people’s boards all the time, around, okay, you need to use AI to

157
00:26:44.720 –> 00:26:52.949
Jonathan Kvarfordt: not hire anybody else, add in revenue, or add in pipeline, reduce costs, blah blah blah blah. So these massive type of requests that people have.

158
00:26:53.120 –> 00:27:04.010
Jonathan Kvarfordt: where do you tell people when they have this massive, ask from a board of saying, use AI and make sure you have a revenue? What do you say to people so they don’t feel like they have to move the ocean or boil the ocean?

159
00:27:05.760 –> 00:27:08.070
Wade Foster: Well… you know…

160
00:27:09.330 –> 00:27:18.829
Wade Foster: I think the place to start is just getting people using the tools. Like, you know, I’ve seen for ourselves, like, we’ve gone from…

161
00:27:19.170 –> 00:27:25.469
Wade Foster: you know, less than 10% of people using it to over 90% in a year. And…

162
00:27:26.570 –> 00:27:38.570
Wade Foster: you know, a lot of that was just… a lot of that wasn’t, like, grand ideas, and, like, big projects where it’s like, oh, we’re gonna, like, we have this goal, this ROI-driven number to go hit this. A lot of it was just, like, we just want people to start building stuff.

163
00:27:38.790 –> 00:27:44.959
Wade Foster: And creating space like hackathons, creating these luncheon, you know, these show-and-tell mechanisms.

164
00:27:45.290 –> 00:27:47.899
Wade Foster: And the creativity of her team starts to bubble up.

165
00:27:48.240 –> 00:28:01.869
Wade Foster: And pretty quickly, you can go from just giving people access to tools, giving people time and space to go build with this, to now start to identify, like, here’s 2 or 3 areas where there’s huge leverage.

166
00:28:01.910 –> 00:28:18.730
Wade Foster: And once you start to do that, then you can start to come back and have case studies that you can report back to your leadership teams, your boards, and say, okay, here’s where we think we actually can have, like, a really big impact, and we’re gonna put more muscle behind this. I think a lot of teams come in and they try and

167
00:28:19.150 –> 00:28:29.419
Wade Foster: they try and nail the return before they do the investment piece. And there’s just, like, a certain amount of experimentation that I find is so useful and so valuable early on.

168
00:28:29.740 –> 00:28:38.209
Wade Foster: And once you do that, then you start to actually have better ideas for, like, what it’s… how it’s gonna actually impact your business. So there is a little bit of leap of faith that has to happen in those early innings.

169
00:28:38.740 –> 00:28:39.450
Jonathan Kvarfordt: Nice.

170
00:28:40.890 –> 00:28:41.379
Julia Nimchinski: What in the middle?

171
00:28:41.380 –> 00:28:42.380
Jonathan Kvarfordt: Amazing.

172
00:28:42.380 –> 00:28:59.859
Julia Nimchinski: Yeah, thank you so much, Jonathan. Thank you, Wade. And lastly, we’re about to transition to one of the most interesting sessions here as well, from SEO to AIO. The definitions are endless, but, yeah, we are not hitting AGI, looks like.

173
00:28:59.880 –> 00:29:05.569
Julia Nimchinski: So, Jonathan Wade, what are your predictions? Where are we heading next in the GTML?

174
00:29:06.930 –> 00:29:09.940
Jonathan Kvarfordt: I’ll let Wade go first, he’s got a better X than I do.

175
00:29:09.940 –> 00:29:13.399
Wade Foster: I… where are we heading next? You know, I think…

176
00:29:14.020 –> 00:29:24.330
Wade Foster: The technology is already very strong and very economically useful, but I think the biggest challenge a lot of us are facing is this rollout period.

177
00:29:24.330 –> 00:29:38.459
Wade Foster: you know, the tech is here to do impressive things, but it is gonna just take a lot of time for teams to start to get good at using it, to deploying it, and so I think we’re in that sort of, like, slog period. But I suspect over the course of the next 12 months.

178
00:29:38.460 –> 00:29:49.320
Wade Foster: We’re gonna start to hear story and story and story from various companies who’ve started to figure out really impressive ways that their company just looks very different from using these tools.

179
00:29:50.660 –> 00:29:59.049
Jonathan Kvarfordt: Yeah, real quick, I’ll say, Zapier, you guys just put out, a really cool report on, the levels to get to orchestration. There’s 4 different levels you talk about, you know?

180
00:29:59.250 –> 00:30:14.920
Jonathan Kvarfordt: And I think organizations right now as a whole are very much at level 1, and like… like Wade said, I think over the next 12 months, we’re gonna see a lot of progression in people’s ability to actually move forward. That will also be combined with AI’s own ability. Like, we were all expecting GPT-5 to be the AGI version.

181
00:30:14.920 –> 00:30:31.310
Jonathan Kvarfordt: But it’s not quite there yet. I think even when we do have AGI, the slowest part of the process of getting it involved is going to be the humans, both from an organizational standpoint and process, because even if I had an AGI to do it, most organizations aren’t ready for an AGI to do anything anyways. So you gotta make sure that the org is ready and the people are ready.

182
00:30:32.970 –> 00:30:34.339
Julia Nimchinski: Thank you so much again.

183
00:30:34.500 –> 00:30:43.370
Julia Nimchinski: Thanks, Jonathan and Wade, and we are transitioning to our next session. Welcome, Kelly Hopping, the CMO of the Mendes, and Eli Schwartz.