Zapier wants to be the AI hub where all your agents connect

Note: Transcripts are automatically generated from episode audio, and are not fully corrected for spelling, grammar, and formatting.

RANA EL KALIOUBY: I think it’s really hard to be the CEO of a software company right now. What’s your experience been?

WADE FOSTER: I had another CEO friend say, “10 out of 10 fun and 10 out of 10 anxiety right now.”

EL KALIOUBY: Okay.

FOSTER: I was like, “Yeah, that kind of resonates.” It is not hard to find individuals who have, quote-unquote, 10X’d their productivity with AI. It is much harder to look around and find the companies that are 10X as productive because of this. Why is that? We have the technology, but we don’t know how to build the organizations around it quite yet. There are going to have to be a lot of changes, not just with adopting the technology, but also rethinking how these organizations work from the ground up.

EL KALIOUBY: That’s Wade Foster, the CEO of Zapier. Fifteen years ago, he co-founded Zapier on a simple idea: to link thousands of apps that were never designed to talk to each other so users could work seamlessly between them. Now, Zapier is doing that for AI agents, and that has changed everything for the company. I wanted to talk to Wade because he saw an opportunity in AI, scaled AI across his entire organization, and totally redesigned his product, thinking about AI from the ground up. I’m Rana El Kaliouby, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution. Hi, Wade. Welcome to Pioneers of AI.

FOSTER: Thanks for having me.

EL KALIOUBY: I’m so excited to have you on. I’m a Zapier customer.

FOSTER: Thank you.

Copy LinkHow Zapier began with a simple pain point

EL KALIOUBY: We use Zapier to integrate all our different AI agents. We’re going to talk about that. But before we go there, I would love to hear the origin story of Zapier.

Did you always know you wanted to be an entrepreneur?

FOSTER: I did not. I grew up in Central Missouri, Jefferson City. It’s the capital of the state. My exposure to jobs was that you could either work for the state or be a doctor, a teacher, or a police officer. As a young kid, I’m like, “Man, the number of jobs that exist is very, very small.” My worldview was quite small. I didn’t really get exposed to entrepreneurship until much later.

What really was the wake-up call for me was the tail end of university, when the financial crisis hit. While I was a good student, it didn’t matter. No one was hiring. No one was taking jobs. I think that was when I realized no one’s waiting to give a handout to me. Even if you do a good job, you have to make your own path.

I was fortunate enough to find a local entrepreneur who had a small software shop in Columbia, Missouri, and I got hired as an intern there. That’s where I fell in love. I was like, “Building software is really fun.” I started to have that itch where I’m like, “Man, I think I would really like to start my own company someday.”

This is 2011. SaaS is on the rise. That’s the year Stripe launched, Twilio launched, things like that. If you went to the forums of these companies, you would see a very common type of customer request. It would always be, “When will you integrate with X?” It was followed by a bunch of customers saying, “Plus one,” “Me too,” “Yes, I want that,” and eventually a product manager saying, “Hey, everybody, thank you for the feedback. Seems like an interesting idea. We’ll take a look.” If you’ve ever worked at one of these companies, you know that’s code for, “Yeah, probably not going to happen.”

EL KALIOUBY: Right.

FOSTER: Brian sees this and is like, “I think we can make a nice, simple UI so you don’t have to know how to use the APIs and connect this stuff.” In my day job, I’m working with the Marketo API doing email marketing, and I’m a bad engineer, so I’m having a tough time.

Anyway, we’ve got this local hackathon coming up, Startup Weekend. We go, we team up with Mike, we hole up in a garage in the back of this building all weekend, and come out with a prototype. It worked. It demoed well, so we ended up winning the hackathon.

EL KALIOUBY: Amazing.

FOSTER: Then after the weekend, it was like, “Well, what do we want to do with this thing?” We’re in Central Missouri. It’s not like there are a lot of investors in the area just jonesing to put dollars toward a bunch of recently graduated college kids. So we were like, “Okay, we want to keep working on this. We’re having fun.” We said, “We’ll do it nights and weekends,” kept our day jobs, and met up after work. We’d snag some Hy-Vee Chinese, which is the local grocery store, get some takeout, and then get to work.

Copy LinkThe first customer proved the idea worked

EL KALIOUBY: At what point did you realize, “Oh my God, we’re onto something interesting here. We need to quit”?

FOSTER: I had this gig at this local software company that was doing some natural language processing. In fact, it was a professor trying to automatically grade student papers, which now —

EL KALIOUBY: Okay.

FOSTER: Seems quaint.

EL KALIOUBY: Right.

FOSTER: It’s like ChatGPT could do that. But in 2011, it was like, “How? How do you do this?” I remember trying to sell this product, and even though the product was slick, well designed, and good engineering, I just couldn’t get anyone to buy this thing. I thought I was bad at my job. That was kind of my recent history: “Man, selling software is hard.”

Then with Zapier, I remember the first customer we had, a guy named Andrew Warner. I’d emailed him. He ran a podcast, Mixergy, at the time. I’d come across his name in a forum where he was looking for a PayPal-Highrise integration. I emailed him and said, “Hey, you still looking for one of these?” And he said, “No, but did you build a plug-in for it or something?”

I said, “Well, we can sort of hook up PayPal and Highrise.” Then I did some snooping and found what other tools he used. I realized he used Wufoo, AWeber, and a few other things, and said, “We could hook that stuff up too.” He pretty much bit hook, line, and sinker and was like, “Oh, I need something to take my Wufoo contacts and put them in my newsletter that runs on AWeber.” I was like, “Okay, great. We do that.” And I was like, “Brian, Mike, build the Wufoo-AWeber thing. We’ve got to go.”

So we give him access to the thing, and the next thing I remember is he’s like, “Wade, this looks cool, but I can’t get anything set up. Can you get on Skype and show me how to do it?” I’m dating myself with all these product names.

EL KALIOUBY: I’ve used Skype, yeah.

FOSTER: Yeah.

EL KALIOUBY: It’s OK.

FOSTER: I’m coaching him through setting all this stuff up, and every step of the way the product just barely works. It is rough. We had been working on it for only a few weeks at that point in time, so it was not very good yet. But we get to the end, the Zap is turned on, and we go to test it. He fills out his form on Wufoo, hits submit, and we refresh the webpage to see whether there’s an email address in there. Sure enough, there’s the email address.

His reaction is like, “Oh, this is incredible, Wade. How much money do I owe you? This is going to save me hours every single day.” That was when I knew. I was like, “Holy cow, this product kind of stinks. It barely works, and yet his reaction is, ‘How much do I owe you?’” In the past, I was working on this quote-unquote slick software, the software that’s good, well designed, and all this other stuff, and I couldn’t get a taker.

I was like, “Man, if we could just get this product to not totally stink, I think we’re onto something.” It was early December when this happened, and by January, I’d quit the day job and was like, “All right, let’s go figure out how to go faster on this thing.”

Copy LinkWhy GPT-4 triggered a company-wide reset

EL KALIOUBY: I’ll be right back with more of my conversation with Wade Foster. I want you to take us back to this moment in 2023 when basically Zapier’s relationship with AI changed. ChatGPT-4 had just launched, and you called for an internal code red.

What did you see about ChatGPT-4? You saw both a concern and an opportunity, and what did you do with that?

FOSTER: Yeah, so we basically noticed a couple of things with GPT-4. First, it came out roughly six months, I think, after 3.5. Second, the capability improvement was obvious. Talking to GPT-4 was just a lot smarter than 3.5. And the cost had come down meaningfully. So it was one of those things where it felt like, what happens when 4.5 comes out? What happens when 5 comes out? Does that happen six months from now? What happens if it happens four months from now? Are they going to launch these things faster and faster? Are they going to get better and better? Are they going to get cheaper and cheaper? Even if one of those things is true, this is very impactful to the whole industry and to what we work on. If all three of those are true, then this is the most important technology that has maybe ever been invented, certainly since the internet, certainly in my active career.

We weren’t exactly sure what to do with that yet, but we were like, “We need people to pay attention internally.” And we never called the code red. We didn’t even know what a code red was. No one had any sort of, “When Wade calls a code red, here are the steps that you take.” It was just like, “Code red,” and people were like, “OK, I hear that that’s important. What do we do?”

So we were figuring out what to do, and I remember teeing up a handful of things that we were like, “OK, this is what I want to change in terms of how we operate.” A lot of it was like, “Here’s how the product needs to change. Here’s how operations need to change. Here’s a list of stuff that’s really important.” When I look back at that time, though, I would say there was really one thing that stood out as being more effective than anything else: We paused the company for an entire week and ran a hackathon. It didn’t matter if you were in engineering or not.

EL KALIOUBY: That’s awesome.

FOSTER: It was like, “Everyone’s going to put your hand on a keyboard, and you’re going to build something with AI.” And at the end of that week, we went from about 10 percent of people using AI as part of their daily job to over 50 percent.

EL KALIOUBY: Amazing.

FOSTER: And as a result, more and more of the company started to build an intuition around what this technology is, what it’s capable of, how we might build in tandem with the technology for our customers, and how we can use it ourselves to improve our own operations. And I think that company pause, where it was like, “Hey, we need to go do some sense-making together,” I’ve since seen be productive for almost any other company that does one of these. They all cite the hands-on learning time as being way more impactful than the CEO memo or the AI committee or any of that kind of stuff.

Copy LinkHow Zapier became a hub for AI agents

EL KALIOUBY: So a few things came out of that hackathon, one of which, I imagine, is this new product strategy. The reason why we use Zapier is, as I said, we use Claude Code, and we have a whole bunch of AI agents doing work on our behalf. But to really have these agents work for you, you need to connect them to your other tools.

We wanted to connect Claude to our email and our calendar and Airtable and PitchBook and all these other platforms, and that led us to Zapier. So you’ve become the AI connector, or the AI integration hub. How did that idea come about, and what did it take to do that? And I would love for you to also explain what an MCP is, because that’s basically how agents talk to each other.

FOSTER: Yeah. I think some of this was a little bit fortuitous, right? In 2011, we set out to just integrate stuff. We were going to make integrations really easy for people. So our integration story is 15 years old. We’ve just gotten really good at integrating any tool that you have on the internet, far better than anyone else.

Pretty quickly, we started to notice, and so did many others, that these AI models are incredibly smart on a bunch of textbook information. They’re trained on the internet. So if you want to ask it, “Hey, how should I think about a marketing campaign?” or “How should I think about this strategy?” or “How should I think about building a product like this?” it will give you a very generally good answer. But it kind of leaves you wanting. You’re like, “Well, I don’t want a generally good answer. I want something that’s great for the environment that I operate in.”

What a lot of folks started to notice is that if you gave it that context, if you let it see your CRM data, if you let it listen to your closed-won and closed-lost Gong calls, if you gave it access to your email, your Slack, all this stuff, and then started asking questions and building automations, you got very specific and very good answers, stuff that makes you start to go, “That’s as good as I’d get from a good teammate internally,” or maybe better in some cases because it’s able to read over all this stuff that most humans can’t and hold it in context.

So that was where it was like, “Oh, Zapier really has a role here,” because we can bring the integrations into these model capabilities and help you orchestrate the workflows, the agents, however you want to build those out. So we just doubled down on that.

In the past, the integrations were all done through APIs. Now there’s this new thing, MCP, as you called it. It stands for model context protocol. Effectively, it’s just a way for agents to talk to other agents. I think for most folks, it doesn’t matter to know much more than that. But the nice thing about it is it allows you to talk in natural language to an agent or to something else, and then behind the scenes, the AI knows how to convert that into structured requests, aka an API call or something like that.

EL KALIOUBY: Maybe we can give an example here. So if I’m talking to my chief of staff, Blue, we call it Blue, on Claude, and I’m like, “Can you add this new potential investment to the Airtable? Here’s the name of the founder. Here’s the website of the thingy.” Or actually, even better, “Go find details on the founder. Go find the website,” blah, blah, blah, “and add it to the CRM,” right? That then gets converted to a language that Airtable can understand, and that’s you guys, or that’s the MCP, I guess?

FOSTER: Exactly. Basically, it takes this unstructured request, which is like, “Hey, do this thing and do it this way,” et cetera, which, when you’re talking to another human, another human knows, “Oh, I know exactly what you mean.” But computers need very structured requests to be able to do those well. So what an MCP does is it takes that human language and converts it into something that a computer knows how to do, but it allows it to talk to other agents that are not part of its core system.

Tools is what a lot of people call this stuff. With the Zapier MCP, you can talk to all of your tools. You can talk to all your apps. So you can say, “Hey, pull out the five biggest customers we have out of the CRM. Draft an email inviting them to this event.” You can do that sort of natural-language type of stuff.

But you can also go a step further. That’s where you are talking to it. You can also say, “Hey, now every time this happens, I want you to build an agent that does this workflow for me.” And it can go, “Oh, great. I’ll make sure to set up an agent for you, or a workflow that does that,” so you don’t always have to be reminding it, “Well, hey, it happened again. Can you please do it again?” or, “It happened another time. Can you please do it again?” It can just always be listening and paying attention to the environment and then taking actions for you.

That’s where I think a lot of folks started with AI as a chatbot. Now we’re starting to get a lot of AI as an assistant, talking to these things. I think the next era is AI truly taking action, delegating it to these agents, these bots, these workflows, et cetera. And that’s what I’m really excited about, where I think Zapier can be a huge unlock for individuals and companies.

EL KALIOUBY: Can you talk through the business model? What does it look like?

FOSTER: Yeah. For Zapier, it’s pretty simple. It’s a subscription with usage-based credits associated with it. I think a lot of this is pretty standard in the AI era. I think what’s very different about these AI products is, in the past, when you were working with software, you would have an interface where the human still has to do all the work. So I’m still the person in there writing the email, logging, doing the data entry, all that sort of stuff.

I think with these AI tools, with things like Zapier, the value is that it can do the work for you. So it is closer to having this entity that does work on your behalf when anything pops up. And so it sort of makes sense for it to be closer to this credit-based style system where it’s like, “Hey, every time this event happens,” it’s doing real work for me.

EL KALIOUBY: Is the right way to see Zapier as the middle layer or the middleman? And if so, is this a vulnerable position to be in? I’ll give an example. I did mention that we are using you guys.

In one particular instance, we wanted to integrate PitchBook. We use PitchBook a lot. We’re investors, right? We couldn’t find their MCP on Zapier, so we asked PitchBook, and they were like, “No, no, no. You can just use our MCP directly.” And that made me wonder: What is the role that Zapier plays, and how do you future-proof yourself so that you’re not just taken out of the equation?

FOSTER: Yeah. So I think there are a couple of ways that folks should set up their AI stack, and I think getting this right is strategically important to companies. Right now, I see a lot of companies and organizations going all in on one AI stack. Maybe it’s, “Hey, I’m a Claude shop,” or maybe it’s, “I’m a ChatGPT Codex shop,” or maybe, “I’m a Gemini shop,” or maybe, “I’m a Copilot shop,” et cetera.

The savvy companies are not doing this, and the reason why is that the frontier is changing super-duper fast. So just because one of these labs has the best model at a particular job at any given moment in time does not mean that they will always have it. They’re constantly leapfrogging each other. Their models are good at different tasks and different capabilities. Plus, you’ve got open source not too far behind, which is almost nearly as good as these frontier labs at a fraction of the cost.

So any savvy company is saying, “OK, if the best capabilities are constantly changing, I need to build an AI stack that allows me to rotate the best tools for the best jobs.” Where we think Zapier is really valuable in building this stack is that you want to hook all those connections into Zapier. And now, anytime you want to swap in and out between these things, it’s as simple as pointing your connections to a new agent, a new different place.

That means you’re not beholden to, anytime you want to try a new setup, going and resetting all those connections up in a separate place. The second thing you really get out of it is a safe place to observe and govern all these agents. So if you’re an organization, you probably have a lot of activity going on, and you want to be able to see what’s going in and out of these tools and when and where. If you’ve got a bunch of fragmented MCP setups all over the place across your entire workforce, you’ve got vulnerabilities hanging out all over the place that are eventually going to bite you.

The last thing that I think is really important for organizations is that you want to make sure these things are running efficiently. Right now, deploying an agent is kind of hard. A lot of folks just have them running locally on their laptops. It’s like you can’t ever shut your laptop lid anymore. Well, if you are able to deploy the workflow, deploy the agent to Zapier, there are two things you get. One, you get a place where you can host it online, so you don’t have to have a spare laptop running or anything like that. But two, the other thing that we’re doing to help you is turning those workflows into deterministic workflows.

Now, why is this important? If it’s running agentically full time, you’re burning tokens left and right, and tokens are very expensive, whereas a workflow runs on code, and that code is reliable, low-cost, and trustworthy. So any place where you can be turning those agents into a deterministic workflow, you probably should. It’s really figuring out how we can be AI-first and build this sort of agentic infrastructure that can run in an automated way and scale.

Copy LinkWhy individual AI gains do not scale

EL KALIOUBY: Yeah, and I want to move us into that. In addition to seeing this opportunity on the product side for Zapier, like becoming the AI agent integration engine, if you like, or hub, you also saw an opportunity to reimagine how work is done within Zapier. And you have this framework or philosophy of moving from individual AI to institutional AI. So tell us more about that.

FOSTER: Yeah.

EL KALIOUBY: Because I feel a lot of organizations have individuals using AI, but it’s not institutional. It’s a hot mess.

FOSTER: Well, 100 percent. I think we’ve all felt this, where you look inside your organizations and it is not hard to find individuals that have, quote unquote, “10x their productivity with AI.” It is much harder to look around and find the companies that are 10x productive because of this, where their revenue is 10x’ed or their costs are a tenth of the price, or whatever.

So why is that? Just because that person got a 10x improvement doesn’t necessarily mean that 10x improvement flowed through the entire system and meant, “OK, great, now we’re generating 10x more customers.” So if you really want to 10x the organization’s productivity, I think you have to step back and say, “OK, if my goal is to generate 10x the number of customers, what’s the equation for my company that creates customers?”

If you want to be reductive about it, there are really two things that generate customers. It’s, I’ve got a product, so it’s the R&D department, like we built something, we made something. And then it’s sales and marketing. We sold the product. So those two things add together to equal customer.

When you’re a very, very small company, almost all the effort the company is putting in, 100 percent of it, is about those two things. But as companies get bigger and bigger, there’s all this coordination that kicks in. So there’s all this other work that sort of happens inside a company that’s supporting generating and building products and selling and marketing, but it’s not directly doing those things.

So when you think about, OK, if I really want to change the equation here, how do I get 10x the productivity for a company, I think you’ve got to be hitting one of those three spots in the organization. How am I 10x’ing the efficiency of my go-to-market engine? How am I 10x’ing either the quality or the speed at which I’m generating new products? Or how am I making the coordination costs of delivering those things go down by a tenth so that I can redistribute the time and effort into those two buckets?

I don’t see a lot of companies yet at that stage where they’re thinking through this in that same sort of scientific way. I think a lot of it is much more haphazard, where it’s like, “Hey, you individuals, we got you some AI tools. Build some cool, crazy stuff that’s going to make us more productive.” So I think we’re in this stage where we have the technology, but we don’t know how to build the organizations around it quite yet. There are going to have to be a lot of changes that happen, not just with adopting the technology, but also rethinking how these organizations work from the ground up.

Copy LinkHow Zapier is rebuilding work around AI

EL KALIOUBY: Don’t go anywhere. I’ll be right back. How are you approaching this internally? What changes have you made? How are you upskilling people? What does your org chart look like?

FOSTER: There are places where it’s changed, but I would say it still looks pretty similar to what it has in the past. I don’t think it’s going to remain that way. But the three key AI transformation efforts that we have going on map to those areas pretty cleanly.

We have a software factory that’s working on building the products of the future. So that’s instead of engineers sitting down and writing the code, that’s building the machinery that writes the code, and engineers are now working on the factory and saying, “Hey, how do we get this set up so this code can run in a loop?”

EL KALIOUBY: That’s super cool.

FOSTER: We’ve got these go-to-market agents that are doing the same thing. A lot of that is working at our small-business funnel, where they never had sales deployed to help them in the first place because we couldn’t afford to do it. But now, with agents, you can put something in the loop there, and we’re trying to figure out, OK, how does that change the go-to-market engine?

And then we have this workflow for the whole company that, internally, we call kind of hive-company-type stuff, AI native, or whatever word. We don’t really have a great word for it yet, but the idea is we want to make sure that all of our systems are legible to the AI, and it can operate the company. That means when you go into a meeting, the agenda is crafted by the AI. When you’re looking at action items, the action items are generated by the AI. Some of those action items are taken by the AI. So you’re doing a lot more of this work where, in the past, you’d have humans shuttling information from one department to the next and making sure this stakeholder is connected to that stakeholder and all that kind of stuff.

A lot of that now is being encoded into these AI systems so that we can eliminate the human tasks around these operational coordination tasks that companies have to face.

EL KALIOUBY: Give us an example of a task that was not automatable before and now is doable because of AI.

FOSTER: Let’s talk meeting prep. I’ll give an example for me. I take a lot of calls with customers, and the types of calls I take with customers are often pretty open-ended, pretty exploratory. Who knows if it’s going to lead to a bigger deal or not? I’m often just trying to learn, trying to figure some stuff out.

Pre-AI, there were a couple of ways I could prep for that call. One, I could go ask the account rep, “Hey, could you give me a brief and really get me up to speed on everything going on with this account?” If I ask that, I would be pulling the account team away from active deals, active prospects, things that could actually put money on the board, for them to help me out with this more speculative thing I’m asking about. They want to do a good job of it, so they’ll put real time and effort into it. So that was one way I could do it.

Two, I could just take it on and do it myself. OK, maybe I go do that same effort they were doing. Maybe I put as much effort into it. I probably don’t, to be honest.

With AI, you can put these agents to work doing really good prep for any of this stuff, and that’s an example of a place where a task that should have been done well before just couldn’t really be done well enough. Not really. But an AI can do it pretty dang well. So all of a sudden, we’ve all got meeting-prep agents for pretty much everything we do, when nobody was really doing that before.

So you can see that in just about any profession, any place. There are all sorts of places where tasks were simply not being done because they were too hard, they were too expensive, or we were not able to do them at a high enough quality or with high enough consistency. And it turns out with an agent, you’re able to actually tackle that thing, and now the humans can do bigger, better jobs on top of that.

Copy LinkManaging AI costs while scaling adoption

EL KALIOUBY: Yeah. I want to talk about the cost of AI, and you kind of alluded to it earlier. Everyone’s token-maxing right now, so AI at scale can become quite expensive. We’ve seen companies like Microsoft cancel their cloud code licenses because of costs, and other companies are spending about $7,500 per employee on tokens. How are you thinking about cost as you are encouraging and leaning into everybody using AI at the organization?

FOSTER: Yeah. Look, at some point, any cost that a company spends has to translate into ROI. That’s how we think about any cost. None of us show up to our jobs and are allowed to spend more money than the company earns.

EL KALIOUBY: Right.

FOSTER: You do have to justify it at a certain point. I think with AI, a lot of companies in the last year said, “Hey, we’re in this experimental phase, and so we’re going to be willing to overlook the ROI equation for the time being because we need to go figure this stuff out. We’re going to spend a lot of money. Some of this stuff is going to work. Some of this stuff isn’t going to work, and at the end of this whole exercise, we’re going to figure out how to be a little smarter about that.”

I think that’s what we’re seeing right now. There was a lot of token-maxing going on in the name of experimentation, in the name of figuring a lot of this stuff out. It feels like we’re kind of getting to the end of that era, where companies are saying, “Okay, we saw a lot of that stuff. There was a lot of goodness here, and there was a lot of waste.” On the waste side, how do we step back and think through a smarter way to go about this?

As a result, we’ll see a whole host of solutions. Maybe employees get token budgets, where you’re only allowed to spend a certain amount. Maybe certain workflows are shifted from being agentic to deterministic. I think this is where Zapier plays. It’s like, “Hey, we can help you take some of these expensive workflows that maybe aren’t generating a great ROI right now, but if you ran them at half the cost or a 10th of the cost, they actually would be something that you’d be happy to keep on.”

So I think there’s a mix of organizational solutions and technology solutions that can help people figure out which of these things gets to continue on because it is justifying its spend, versus which of these things needs a little tighter scrutiny because it’s just waste. The meme is like, “Hey, we’re asking Fable 5 to do this basic task for us.” It just doesn’t make sense. You don’t need that kind of power to do a basic task.

Copy LinkWhy AI makes software leadership thrilling

EL KALIOUBY: I think it’s really hard to be the CEO of a software company right now because you’re at a moment where a company’s valuation could swing on a headline, in this whole idea of SaaS is dead and the SaaSpocalypse moment in time. So what are your views on that? We did have Aaron Levie, the CEO of Box, on the show, and he talked about how he had to jump back into founder mode. What’s your experience been?

FOSTER: It’s interesting to hear that was Aaron’s response. I had another CEO friend say, “10 out of 10 fun and 10 out of 10 anxiety right now.”

EL KALIOUBY: Okay.

FOSTER: I kind of chuckled at that. I was like, yeah, that kind of resonates. Where it is fun for founders is this is a period of intense creative disruption. You’ve been gifted this incredible technology. You get to rethink everything from first principles, and that’s just so exciting if you’re entrepreneurial-oriented, creative-oriented, et cetera. So that’s the part that is exceptionally fun and cool.

There have been times in Zapier’s history where things have been much more stable, less chaotic, et cetera, but also more boring.

EL KALIOUBY: Yeah, yeah, yeah.

FOSTER: This time, you can say a lot about this time period, but boring it is not. That’s where you wake up pretty excited about it. Now, of course, there are a bunch of challenges. We’re fortunate enough that we don’t have a ton of outside investors, so we don’t have the same scrutiny that a public company might have that’s trying to figure out how to navigate this era.

But we do have customers, and we have employees that are stakeholders and shareholders, in some ways, in what we’re doing. So there is a lot of work you have to do to make sure that you’re painting a clear picture of the future when you yourself are trying to figure out what exactly this is going to look like. You kind of know the direction, but you don’t know the final form factor yet. That’s a challenge.

You’ve obviously got a massive media storm around this stuff. Information comes fast, and it’s hard for the employee base to always discern fact from fiction. Not everything out there is factually correct. There’s a lot of disinformation. There are a lot of people pitching their own agendas. You’re really just trying to settle people and reorient them and say, “Hey, here are the customers we’re serving. If we make sure to deliver the best services we can in this, this, and this dimension, that is only going to lead to good things for us.”

Those are the parts that end up being really challenging: the communications part of the job, being able to rally people around a change, and then ultimately, when something new emerges or you find that your hypothesis is wrong, being willing to call it quickly and then get everybody to go, “Oh, not that way. More this way.” Then getting them to keep doing that over and over again can be jarring. But I do think that is part of the fun, too.

EL KALIOUBY: 10 out of 10 fun, 10 out of 10 angst or anxiety, I think, encapsulates and summarizes pretty well everybody’s experience with AI. Wade, thank you so much for joining us on the show. That was great.

FOSTER: Awesome. Thank you for having me.