After the SaaSpocalypse, with Wade Foster

You did a great job, Elio. I'm stoked to be here.

Yeah, we had the whole team out in LA and funny enough, the night we got in, the hotel we were staying at was hosting the Grammys. So, like you know, hear all these like tech people walk in in like, you know, hoodies and, you know, stuff like that while all the like Hollywood elite are like dressed to the nines. Oh, yeah. Like the best looking people you've ever seen are just like, you know, walking through this thing and it's like, this is a different vibe than I'm used to.

I didn't anybody?

Mhm.

So we founded the company in 2011. And the insight that we had was you would go to the help forums of the early SaaS companies. So this is like Dropbox and Evernote and Salesforce and QuickBooks and things like that. They all had these public community forums at the time and customers would submit feature requests, ask for help, share tips and tricks, all that sort of thing. And there was one type of post that was common across all of them. And it was, 'When will you integrate with X?' And when you'd click into those threads, you would see the same pattern. It would be customer after customer saying, 'Me too. Yes, please. I want this plus one.' That kind of thing. And then eventually a product manager would come in and say, 'Hey, thanks for the suggestion. We'll get back to you on that.' And so we saw that and thought, hm, I think we can build a product focused on integrating these tools. We'll put a UI on top of the APIs and that way people don't have to understand how these technical underpinnings work. Now, from that, it wasn't a huge leap of logical conclusion to suggest that well, one of the things we ought to be doing is sending those people to our site some way. And if we were in say the Evernote help forum and they were wanting to find like an Evernote Dropbox integration, well, it sure would be nice if we sent them to a page that was for Evernote and Dropbox. And so we thought, hm, we ought to have a big old directory that lists out all the different ways in which you can integrate with Zapier. And so we just started building the site. So every new integration we add would create landing pages for all the new combinations we would have and you know we drop into the forums and say hey you can go build an integration with the APIs directly if you're looking for something that maybe could do this off the shelf we're working on a project Zapier you might check it out and let us know what you think and that ended up also being a really good way to drive traffic through search. Nobody was searching for like there wasn't huge volumes of people searching for any particular integration for Evernote, Dropbox, but people searching for any particular type of integration turned out to be quite large. And so pretty soon we started to see that this was working.

I mean over the course of 3 months.

Yes. But search isn't this thing that's like up and to the right right away. It's this like very steady steady improving channel. And so we just kind of saw it week over week was like, you know, week two was better than week one and week three was better than week two and week four was better than week three and it just kept compounding and compounding and compounding. And you know, if you do that for a long period of time, any amount of compounding adds up to a big number. And so that was kind of the original the like product and the PLG engine. They worked like a you know a hand and a glove.

Well, we never stopped, but definitely the scale is heavily favored towards the apps themselves. So, we built probably the first 50 ourselves and then quickly realized to get to any sort of scale, we probably will need a community to sort of help us go about this. And so we built a developer platform and we can go into the details why but it took off and as a result...

Usually was okay. HubSpot built the HubSpot integration.

ActiveCampaign built the ActiveCampaign integration. Slack built the Slack integration. So on and so forth and that happened to the tune of you know over 8,000 different apps at this point in time. And so yeah, we still do build and maintain some ourselves, but it's a very small number of the overall fraction community and you know, we're maintaining the ones you would think. It's like the Google integration and stuff like that.

Yeah, it would have been about then. Yep.

Truth be told, we should have done it sooner. We had the data sooner, but for a whole host of reasons, we just hesitated. But yeah, I think the thing that we saw that we should have jumped on sooner was we were attracting a lot of larger companies who had they valued the product, but they had slightly different categories of needs. Now, a lot of it's kind of the normal things you would expect around, you know, governance, it's admin controls and permissions and things like that, certain compliance needs, certain regulatory needs. We should have said, let's go build not just a sales team, but also a product team that's going to be dedicated to building out a specific package of Zapier for enterprise that's dedicated to that set of users. And so like the signals are just market pull, market demand. People were using the product. They loved it, but they had reasons why they're like, we can't really use it the way that we want to go use it, but we want to use it. You just need to help us with, you know, ABC thing. And so after sort of getting knocked above the head, you know, many many times, we finally woke up and we're like, 'Oh, I guess we probably ought to be doing this.'

Well I think the companies that I have seen grow the fastest and the most efficient the product and the go-to-market are kind of one and the same. They fit together like so perfectly well. Maybe the product has like virality baked into it and so just in using the product you're naturally sharing it with others and it grows because of that. In the case of Zapier it was the more apps we had the more landing pages it spun up and it worked well on search. You know, the interesting thing I think is that the founders that seem to have the best shot at building these are both students of what makes products great, but they also really understand customer acquisition and they're up to speed on how like modern platforms work. They understand, you know, maybe it's like how AI is surfacing recommendations or how like the YouTube algorithm or the TikTok algorithm works. So they understand sort of like mechanically like how those platforms drive traffic and drive awareness and drive impressions to their audiences and somehow they're able to like build a product and the go-to-market that again sort of has that like hand-in-glove feel to it and that's where there's like really just like it almost feels magical when you sort of do those things in tandem.

Yeah. I you know I think there's three founders of the three of us we had enough like shared experience. We'd all worked at this mortgage company that was really good at online marketing. And so we sort of all kind of honed our chops there, understood how like search algorithms work and all that sort of stuff, but we also had unique skill sets as well. So Brian was an exceptional product builder, backend engineer, Mike had like good frontend and like product instincts. And you know, I often felt like I had the least skills, but at least I would like hustle like nobody else. Like I was happy to do all the grunt work, all the customer support, all the marketing, like anything that needed to get done. I was happy to do. And so, you know, kind of between the three of us, it meant that there was no job in the company that we had to outsource and hire somebody else to get done. Like, we were willing to just do it with our own sweat equity. And that also is really valuable. I see a lot of founders, the first thing they have to do is they have to go raise money to hire people because they can't actually do the thing themselves. And so, you're already like starting the clock. Whereas you can kind of just go do it on your own. That means you buy yourself time to like figure out a whole bunch of stuff.

Yeah. Gosh. I mean, this is like a constant mental exercise to try and understand what parts of a business are truly valuable. We're having this interview at a time when everyone's pontificating about the SaaS apocalypse. All the stock prices are way down. You know, SaaS dead is kind of...

Yeah. Exactly. And while I think that is overblown, it didn't happen for no reason that those stock prices are way down. And it's because everyone is uncertain about the future. You know, in the past, you could look at those companies and say we feel like that recurring revenue is pretty predictable. And so as a result, like we feel comfortable investing at a certain price and we're happy to like let that go. Now people are saying, you know what, maybe that revenue isn't so predictable. And they don't know. And so that's what everybody is trying to ask of themselves inside their companies is to say what do we have that actually is durable over the long haul and what do we have that might be at risk and so that's the exercise we're going through is saying okay yes we have like a lot of integrations what particular part of this is valuable and I think the thing that we've realized is there's a lot of value still in our trigger and our authentication layer nobody allows you to run background agents like we do no one has like all the retry error management all that sort of stuff that's really strong and then the governance piece is really valuable in fact I think that's one of the things that gets even more valuable in the future where inside these organizations more and more people are delegating work to agents and those agents have access to tools and there's big questions around well what capabilities do you give those agents those agents can go move with autonomy really quickly and that means they can do a lot of powerful things very fast and are all those powerful things, things that you want them to do or things that you don't want them to do. And so, one of the things we've released is like Zapier MCP. And this is a really useful mechanism to grant access to tools to your AI, but do it behind sort of a gateway that governs that effectively versus just signing up to a bunch of random community MCP servers that have no security, no governance layer, none of those things. And yeah, hey, maybe it was easy to get it set up, but like who knows like if it's going to work the way you want to work. So, that's kind of a long-winded way of saying like all of us, I think we're, you know, trying to find the places that have the most durable value. And when I look into the future, I do think that there's a lot of value yet in background automation and in like proper governance of these tools. And I think those things just get more and more important the more advanced these AI systems get.

Well, agent has become a very overloaded term in the industry right now. I think it's used to communicate a whole bunch of different concepts. And so, you know, when folks are looking for an agent, you really just got to understand like, well, what is a job you're trying to do and what's the best way to achieve that? Sometimes that means they mean ChatGPT as the agent. Sometimes they mean a coding agent. Sometimes they actually mean a zap. Sometimes they mean like an actual like autonomous workflow like a Zapier agent. But you can't really answer that question without trying to understand like what is your actual job to be done and then let's build a system that achieves that goal within your sort of constraints like you know do you care about cost do you care about reliability do you care about efficiency like what is it the thing that you're trying to get out the other side of this and I think that's one of the biggest challenges you see organizations facing right now is there's a huge learning curve around this technology and there's a whole language and vocabulary that is new and people don't understand it yet and so you know sort of folks walk up and they're like, 'Hello, I heard the agents were here. Like, I would like some.' It's like, 'Okay, well, let's get into understand like what actually you're really looking for here.'

Well, you know I think humans definitely have a role in this work if you value quality. We've all experienced AI slop at this point in time which is really easy to generate and I think the difference between there's sort of two camps of people like folks that are sort of just showing up and prompting the tool to do the task and then not bothering to like evaluate it and see you know is this actually good or not and I'm not even talking about like does it hallucinate or not like it doesn't have to hallucinate to still generate stuff.

Yeah. But then there's the people that I find that are doing some of the best work with AI and they often are just very sophisticated in how they prompt it, the context they give it, and then when the output comes back in a way that they don't like, they often go back to the source and say, 'Okay, what context did I lay? What details can I get back into it?' And so they're really active participant in this loop. Now the interesting thing is for that person they're finding as the models get better and better and better they're able to do bigger better things themselves with it and the places in which the human intervenes it's changing along the way and so you know I do think you know sort of humans in the loop is going to continue but you know we're sort of that outer loop it's pushing us into like different areas yeah there's just more and more things where AI is able to just take on more of I guess the tedious work as we work through it. But the human still at least at this moment in time it's pretty valuable for assessing like is this good? Is this what I wanted? Particularly for like non-technical tasks.

Yeah once that recursive loop starts it's pretty wild and you can see it like when you use those models they are able to think for really long periods of time and operate for really long periods of time which is pretty impressive.