Transcript: Revolutionizing Workflows with AI Agents | Jacob Bank | relay app

Host: Give an open source way to add integrations.

Host: We want our integrations to be really robust and really high quality.

Host: I think kind of a controversial take, but I think that's the worst of all possible worlds where you end up doing a ton of work to to implement them, and then you don't control your destiny if something in the integration doesn't doesn't work perfectly.

Host: And for us, integrations are so core to our business that like we just can't let we can't let that be in the hands of a third party.

Host: We need to we need to own the experience.

Host: We've been building from the back of the funnel to the front of the funnel.

Host: This is kind of based on my experience from my first company.

Host: So, in my first company, Time Fall, we were featured in the App Store on our launch week.

Host: We got like 300,000 downloads the first weekend and retained none of them.

Host: So, uh I was pretty scarred by that experience.

Host: And so, I decided that for the next company, I would rather have 10 rock solid, retained customers that find a ton of value in the product, and then figure out how to activate more into it, and then figure out how to bring more in top of funnel.

Host: And so that's the progressive stage.

Host: Obviously our YouTube channel is going to cross 10,000 views for the first time this month.

Host: It's like an order of magnitude fewer, but super high intent, and those convert really, really well.

Host: Like 25% of our paying customers come from YouTube because like if you're going to sit through a 45-minute tutorial of me building an AI agent, like you're you're you're high-etched.

Host: You really want to, you really want to use this product.

Host: All strategies don't work for all products, so there's always a product and the strategy fit and the channel fit, I think, for you because it's a horizontal product.

Host: Um, and because it enables so many customers.

Host: It's also a very demoable product and very outcome-oriented product where you can start with the outcome itself, and the outcome can be very attractive because it saves time.

Host: My guest today is Jacob Bank.

Host: Jacob is the founder and CEO of Relay.

Host: Relay is a startup that is building AI agents that will work for you.

Host: Um, he previously built products at Google in consumer shopping, G Suite, and Feriarius.

Host: He was also the founder of Time Fall, uh a new type of calendar that was acquired and integrated into Gmail and Google Calendar.

Host: In this conversation, we'll cover, you know, what are the different types of agents Relay is building, and how it's solving different customer problems.

Host: Uh how are agents evolving?

Host: Uh what is the potential of AI agents?

Host: And what are the challenges in creating a product in such a fast-moving space and being an AI first product.

Host: If this is first time listening to your startup project, don't forget to subscribe to us.

Host: Uh wherever you listen to the podcast, it will help us reach more audience.

Host: Jacob Bank, welcome to Startup Project Podcast.

Guest: Thank you for having me.

Guest: I'm excited for the conversation.

Host: Uh so let's get right into it.

Host: Uh what is Relay, and how did it get started?

Guest: Relay.app is a platform to build AI agents, and the journey has been a bit winding to get here.

Guest: Um, I started the company in 2021.

Guest: Uh my background, by the way, is in academic research in AI and then building productivity tools.

Guest: And so when I started the company in 2021, we had two founding premises.

Guest: Um, the first founding premise was that we all use a lot of tools to get our work done, and that the builders of those individual tools spend a lot of time figuring out how to make experiences within those tools better, but probably not enough time figuring out how those tools would interact with the others in your ecosystem.

Guest: For example, I used to be the product lead for Gmail, and we would agonize about every single pixel when you were archiving an email or replying to an email or starring an email.

Guest: But then if you said, hey, I really need to get data out of Gmail into Notion and then also Salesforce and then send a slack message we say, ah, sorry, good luck, use the API.

Guest: Um, and when we when we looked at the knowledge work people were actually doing, a lot of what we do as people is like take stuff from one tool, do some stuff to it, and then you then stick it in another tool.

Guest: So we thought there was an opportunity where people were underestimating the importance of of cross-tool coordination.

Guest: And then second, you know, it sounds so silly to say this now, but it was not obvious in the summer of 2021 that AI is going to be important in in doing this.

Host: The original thesis when you know AI was like not highly important.

Guest: So the original name, uh this was before Chat GBT.

Guest: So this was before it was sort of known that LLMs were going to have like the breadth of use cases and the quality that they did.

Guest: I mean, the kind of work I had done on Gmail was with some precursors to modern systems.

Guest: Like, we built this feature where it automatically autocompletes as as you type your email, and that used a variety of like, you know, kind of precursor models, LSTMs and things that didn't use exactly the same technology as as as what what what modern models use.

Guest: But it was like kind of directionally on the right track, but like dramatically less capable than what we have now.

Guest: But even, you know, my background once upon a time was in just like traditional machine learning, feature engineering and classification using support vector machines.

Guest: And I thought even using that technology, you could do some some some useful stuff.

Guest: So the original name of the company was collab AI, and we didn't know exactly what product we wanted to build, just that it was going to help with cross-product workflows, and that it was going to use AI somehow to to to do that.

Guest: And so for the first year of the company, really, first year and a half of the company, we wandered, we wandered in the desert as they say.

Guest: We built eight or nine different product prototypes that all kind of fit that theme, but none were quite right.

Guest: We built an automated to-do list that aggregated things you had to take action on across your tools.

Guest: We built a contextual knowledge base that would tell you when content was up to date or who was the right person in your company to talk to.

Guest: We built a stand-up tool, we built an employee onboarding tool.

Guest: And then eventually by the end of this, where we landed was a workflow tool.

Guest: And a workflow tool that captured repeated tasks that had an element that can be automated and an element that required human judgment.

Guest: So that's that's why we named the company Relay because we're thinking, ah, so many things that we do should be a relay race where the computer does some stuff and the user does some stuff.

Guest: And so that's when we named the company Relay and announced our beta.

Guest: This was at the end of 2022.

Guest: We ran our beta in 2023 where we were a workflow product that was AI powered, that sort of sat halfway between a Zapier style automation product and an Asana style collaborative task management experience.

Guest: And by the end of the beta, we realized that maybe there was a category to create to be created there, but it wasn't us who was going to create it.

Guest: It just wasn't right.

Guest: You know, you know and I'm sure many of your guests have mentioned this, like, you know, when you're an entrepreneur and you're just like muscling something through that's fundamentally not right.

Guest: There's too much friction in the conversations.

Guest: You can't explain what you're doing easily enough.

Guest: And so in the summer of 2023, we decided we're going to build an automation tool.

Guest: Um, and that we're going to focus on, you can call it like the market that Zapier is is the leader in, which is cross-product horizontal workflow automation for a non-technical audience.

Guest: Um, but we were going to try to build the modern version of it.

Guest: And so what makes it modern?

Guest: It's way easier to use for non-technical people.

Guest: It has AI better integrated into the workflows that you create, and it has human in the loop capability so that you can correct your AI when it gets stuff wrong.

Guest: And so, in 2024, we were an AI powered automation uh product.

Guest: Um, and we kind of positioned ourselves as the modern alternative to Zapier.

Guest: We got to initial traction and then early product market fit.

Guest: And and but what we realized was that by positioning ourselves as an automation product, there were two major limitations that came from that.

Guest: One, when you describe yourself as a no code workflow automation tool, you really limit the audience of people that think you're the tool for them.

Guest: Like, um, there's a subset, you know, there's a subset of the world, 10 million of people or so who are comfortable working on these no code automation tools, but we wanted to tackle the much bigger opportunity of helping every business get more work done with AI.

Guest: So we wanted to broaden our positioning to to capture that that opportunity.

Guest: And then second, we really didn't want to be perceived as a duct tape product that only exists to like temporarily glue two products together because they happen to not have a native integration.

Guest: We want to be like a transformative tool that materially augments the work that your team can do.

Guest: So the transition, the evolution that we're making now, and we can talk about whether the technical language actually matters or not, I don't think it does, but we're basically transitioning from an AI automation platform for no code workflow builders into an AI agent building platform for everyone.

Guest: And we can talk more about what that means and and why I think it's cool.

Host: So, I think everyone at this point, at least, you know, slightly people who use startups, you know, early stage company products are, you know, no code has been this word that has been flowing around for the last five years.

Host: Everyone is sort of familiar with the no code low code idea.

Host: Zapier is pretty popular, and I had this thesis that, you know, a next version of Zapier is bound to be because of AI because it's sort of like if you imagine a line with, you know, 10 dashes, and the gaps were much longer before AI.

Host: Now it's like you have one single line with a small dot, where the dot represents the human in the loop.

Host: So that's sort of like the way I describe from Zapier to the next version of Zapier.

Host: Um, the question really though, like you you mentioned the duct tape product.

Host: You don't want to be, and when I see a product similar to yours is the problem is we are using so many tools and so much data is spread across different things.

Host: Um, so many workflows require more than five tools to get it done.

Host: Um, so how are you prioritizing which tools to bring into the platform?

Guest: Right now, we have uh about 120 native integrations.

Guest: The way we think about it is that there's about 12 categories of tools that pretty much every business needs.

Guest: Every business has an email client, Gmail or Outlook.

Guest: Every business has a calendar client and then maybe also a scheduling tool on top of it.

Guest: Every business has a messaging tool of some sort, Slack, Discord, WhatsApp, other.

Guest: Every business has a CRM of some sort, whether that's a dedicated CRM like Salesforce, Hubspot, Pipedrive, Adio, or you've built your own CRM in Notion or Air Table or Google Sheets.

Guest: Everyone has an email marketing tool.

Guest: Everyone has an email e-signature tool.

Guest: So there's these there's these, you know, 12 to 15 categories that are that are quite ubiquitous.

Guest: And then in each category, there are 3 to 20 like players that have material market share.

Guest: Some some markets are more consolidated than than others.

Guest: I mean, for example, if you look at email marketing between Loops and Drip and Mailchimp and Kit, like there's there's probably dozens that have that have, you know, pretty material material usage.

Guest: Um, and so we've basically just tried to kind of work our way down that list for our target audience, which is it skews towards smaller and medium-sized business.

Guest: Our largest customer has like 1,000 employees.

Guest: So we haven't done the Net Suite or the ADP or the service now, like the the major the the larger enterprise products.

Guest: But if you're a regular smaller mid mid-sized business using modern tools, you'll probably find everything you need with with Relay.

Guest: So we have we have, you know, a hundred some.

Guest: Zapier has 7,000.

Guest: I don't think you need 7,000.

Guest: I think that's a vanity metric.

Guest: I'm I'm I bet half of those don't even really, really work.

Guest: But they definitely have more than we do.

Guest: And so I think the number that we need to get to is probably somewhere between 3 and 500 for for the product to really feel complete for everyone in our target audience.

Guest: And that is a long slog, but I think I think a couple I have a couple beliefs that I think are different from other people in the startup ecosystem.

Guest: One is, I believe integrations are skilled labor.

Guest: Like I do not think this is something that you can just outsource to a low quality engineer.

Guest: If you're going to build a really good sales force integration with thousands of custom fields and thousands of custom properties and build a performant autocomplete on top of that and not know how to navigate the object hierarchy.

Guest: That's like that's some really hard software engineering and you need really, really good software engineers to to do it.

Guest: And the second is that I believe agents will only be as useful as the robustness of their ability to interact with the tools they you use.

Guest: So there's kind of like two big schools of thought.

Guest: One school of thought is like let's just do everything in the browser.

Guest: Like, you know, open AI operator or cloud computer is going to I'll just be I'll just be like the humanoid robot equivalent of.

Host: So it's almost an evolution of like the UI testing Selenium.

Guest: Yeah, exactly.

Guest: It's like on on one hand, but but, you know, I don't you can think of it as like UI Path meets the modern age, but like UI Path to me means a lot of like very really carefully scripted rules.

Guest: Um, um, but yeah, so there's like the we're just going to use the browser and then there's going to we're going to build on top of APIs.

Guest: Those are the two big schools of thought.

Guest: I think every serious player in the agent space is going to need to do both at some point because there will always be some products that don't have APIs.

Guest: But if there is an API available, like almost certainly, that is going to be a way more efficient and way robust way for the AI to interact with the product in the near term.

Guest: So we have focused entirely on robust API based integrations because for our target audience and the kind of tools we integrate with, those exist.

Host: Um, in in terms of traction, uh what has the traction been like?

Host: Um, you know, how many customers are using it?

Host: Like give us a little bit of insight in terms of like the scale of the company, right?

Guest: Yeah, so we're now at um 440 paying customers.

Guest: We have about 1,200 weekly active teams.

Guest: Um, uh that's up from, you know, essentially zero when we launched at the very end of 2023.

Guest: Um, uh for so that's that's been our that's been our first year.

Guest: I would call that, you know, everyone has a different definition of product market fit.

Guest: I'd call it early product market fit.

Guest: I think, you know, we we kind of in the last six months have gone from initial traction to early product market fit.

Guest: My personal definition of of product market fit is like, could I go on for for a product-led self-serve business?

Guest: Like, could I go on vacation for a week and come back with more users, more customers and more revenue?

Guest: And that is now true of our business.

Guest: Um, we still, you know, we still of course want to bend bend the curve and and and grow faster.

Guest: But I think we've um with 440 paying customers, there's enough there to say that you you're not just like a bespoke consulting shop for for one or two companies.

Host: So, uh what are you doing to sort of drive this adoption further?

Host: And I think you've done very well in terms of like product-led content growth, especially on LinkedIn.

Host: Once I discovered your product and started, you know, following you on LinkedIn, I've really seen that, you know, you really took to yourself on selling the use cases what the product can do.

Host: Uh talk to me a little bit about about that.

Host: Is that the strategy?

Host: Uh you know how intentional, how are you thinking about it?

Guest: Yeah, that's super reason, by the way.

Guest: We only kind of figured that out in the in the past month.

Guest: So we basically, we've been building from the back of the funnel to the front of the funnel.

Guest: This is kind of based on my experience from my first company.

Host: What do you mean by that?

Guest: Yeah, meaning we started with retention and depth of engagement, then moved to activation, and now have moved to to working on top of funnel.

Guest: So, in my first company, Time Fall, we were featured in the App Store on our launch week.

Guest: We got like 300,000 downloads the first weekend and retained none of them.

Guest: And so, uh I was pretty scarred by that experience.

Guest: And so I decided that for the next company, I would rather have 10 rock solid retained customers that find a ton of value in the product, and then figure out how to activate more into it, and then figure out how to bring more in top of funnel.

Guest: And so that's the progression we've taken where already six months ago, I felt pretty good that once someone was activated, they were going to retain really helpfully for the long term, engage, um and broaden their use cases.

Guest: We then spend a ton of time working on activation so that people could understand what their first use case should be and get their first workflow up and running successfully.

Guest: We have a lot more work to do there, by the way.

Guest: Um, and only now in the last two months have we put a a disciplined effort into in into top of funnel.

Guest: And the strategy that I think makes sense for our kind of product, fundamentally, it has to be word of mouth.

Guest: Facilitated word of mouth isn't just like you sit back and you wait for people to hope and hope that they talked about it with the other.

Guest: Word of mouth facilitated by content, community and partnerships.

Guest: If I had to pick one company that we look up to as a role model, it's it's Notion in that regard, where they built a very flexible horizontal tool that gained a lot of it took time, took a long time, but eventually gained a lot of true fans and then had this whole ecosystem of community members and templates and partners.

Guest: And so for us, it's it's sort of the same.

Guest: So, first, there's content that I create.

Guest: And I I spend a lot of my time on content creation, and there are two types of content that I create.

Guest: One is, I make LinkedIn posts, which are designed to be sort of teasers that illuminate a use case that you could use an AI agent for.

Guest: And so I read a lot of posts of the form that's like, I just built a cool AI agent to synthesize all of the insights from my customer calls from the last week.

Guest: Here's how it works.

Guest: It looks at all of your Fireflies recording from the last week.

Guest: It takes, you know, parses key information out of the summaries of each of them.

Guest: It puts it into Cloud 3.5 Sonet to write an aggregated summary, and then it sends it out to our team over Slack, make us like a 10-second GIF about it and then say, you know, let me know if you're interested in in trying it.

Guest: And then I pair that with long form YouTube tutorials that show you how to actually build the thing.

Guest: Um, and so our on on LinkedIn, I now have 15,000 followers, but a month ago I only had 7,000.

Guest: So this is like this is brand new.

Guest: And um and now I'm getting, you know, 150,000 or so impressions a month on LinkedIn.

Guest: So pri- pretty decent top of funnel for for our stage.

Guest: Obviously, most of those people on LinkedIn aren't like looking to adopt or buy a tool right now.

Guest: It's more about sort of like building this this brand gravity as they call it, where people become more aware of you over time.

Guest: And then our YouTube channel is going to cross 10,000 views for the first time this month, like an order of magnitude fewer, but super high intent and those convert really, really well.

Guest: Like 25% of our paying customers come from YouTube because like if you're going to sit through a 45-minute tutorial of me building an AI agent, like you're you're you're high-ed.

Guest: You really want to you really want to use this product.

Guest: So that's the that's the content I create.

Guest: But now that we have a pretty robust community and set of partners and set of happy customers, many of them are writing LinkedIn posts about AI agents they've built because, you know, they want to show off that they're using AI in their work too.

Guest: They might build templates in our gallery.

Guest: They might record their own YouTube videos.

Guest: And so um this is how we're starting to build because with a horizontal product it's always um tough for people to figure out like, what do I use this thing for?

Guest: And the way we're trying to solve that is by having a really robust gallery of concrete use cases that are broadly resonant that you can start with today, whether those are provided by me or whether they're provided by one of our partners who has an audience in a specific function or a specific industry.

Guest: So, for example, we have a couple of real estate agents who love our product, and they've started creating content about how other real estate agents can use AI agents effectively to monitor new listings or interact with their clients.

Guest: And so, I don't know anything about real estate, but I can build a platform that enables the real estate agent to build their own agent and then share it with their audience.

Host: I mean, it's all strategies don't work for all products, so there's always a product and the strategy fit and the channel fit, I think, for you because it's a horizontal product.

Host: Um, and because it enables so many customers, it's also a very demoable product and very outcome-oriented product where you can start with the outcome itself, and the outcome can be very attractive because it saves time.

Host: And LinkedIn is sort of like always about, you know, what is the growth hack that you can give me?

Host: What what can I do, you know, something in one minute?

Guest: It's a really, it's a really good fit for LinkedIn.

Guest: That's what we found.

Guest: And for example, like I've had no traction on Twitter.

Guest: Zero.

Guest: Like I I I tweet similar things and it gets like 50 views.

Host: Zero.

Guest: And then it gets like 500,000 impressions on LinkedIn.

Host: Yeah.

Host: Yeah, I think the LinkedIn plus YouTube strategy is clearly a winning strategy for you.

Host: And as like the as the product gets more and if you add, uh I almost feel like affiliation could another be another way.

Guest: Exactly.

Guest: So we just we just we just launched our affiliate program and we just launched uh a template gallery that other people can submit templates too.

Guest: Um, and so that's where we want to kind of capture these uh other content creators who have their own audience that they want to use us as a tool to help, you know, make their business better.

Host: How come there's no LinkedIn integration when you post so much on LinkedIn?

Guest: Ah, we're actually in the review process from from LinkedIn right now.

Guest: Um, we're we're waiting for them to flip the bit to to to accept us, but it's coming very soon.

Host: Because you have Blue Sky, which is, you know, very, very, you know, recently, you know, more of an adopted product.

Guest: Yeah, you know, it's it it's funny because um there are a few dimensions that make integrations easier or harder to build.

Guest: There's the quality of the API documentation.

Guest: There's whether there's a formal review process that you need to go through and how rigorous that review process is.

Guest: There's how many different capabilities you need to support.

Guest: And so, we've been we're constantly surprised by which integrations are easy and which integrations are hard.

Guest: It has been counterintuitive to me.

Host: What integrations are coming in the next couple of months?

Guest: So we have a public, we have a public road map of integrations.

Guest: Um, it's listed uh we use a uh we use a feature request board tool called Nolt.

Guest: Um, LinkedIn is at the top there.

Guest: Um, a few others that are coming.

Guest: WhatsApp is coming.

Guest: That's a really highly requested one.

Guest: Um, Zero, the accounting provider, is a really, really highly requested one.

Guest: We have a bit a few more social media integrations to build, deepening our YouTube integration, building an Instagram integration, uh because for a lot of content marketers, those are those are important.

Guest: Um, we need to deepen our integrations with website builders and blog providers.

Guest: Like we don't have Squarespace or Wix or.

Host: WordPress.

Guest: Yeah.

Guest: We have we don't have WordPress.

Guest: We just have Webflow at the moment.

Guest: Um, we don't have Sanity, even though that's the tool we use.

Guest: So we always have the escape hatch where if you're a little bit more technical, you can use um a tool called custom HTTP requests or web hooks to directly integrate.

Guest: So we we do all our own publishing on our Sanity instance via Relay, but we use web hooks to to do it, which is we don't want that to be the experience the majority of our users have to go through because you do have to be a bit technical to to to set that up.

Guest: Um, and so the the drum beat of integrations will will will go on uh forever.

Guest: But we we really want to cluster around the use case where we're seeing the most traction, which is content creation and marketing, research, like research use cases, especially in the context of of sales, and then general, like back office operations for the most common for the most common providers of those tools.

Host: It's amazing that in spite of like, you know, you talk to a typical business or like a typical software company, you have like hundreds of SAS tools that a company has to use and still uh, you know, a company like Relay is needed for them to complete tasks.

Host: Like it's Well, it's the it's it's even it's even as, I would argue that as the number of tools goes up, you need it more.

Guest: Um, because, yeah, if you're just using everything within the Hubspot ecosystem, for example, most of those tools have decent native integrations with each other.

Guest: But the more you're choosing, you know, best-of-breed individual tools in each category, like a typical provider will build the integrations with like the top three partner products in the top three categories they need to integrate with.

Guest: But that still leaves like a huge ocean of integrations that that that that they don't have.

Guest: So like SAS brawl is is kind of good for tools for for for tools in our category.

Guest: And this is one of the reasons why it was it took so, so long for us to get off the ground, why we needed to build so much.

Guest: This is why there aren't as many companies in this market as I would expect there to be given how big of an opportunity it is because like it's just a lot of work to build all these integrations.

Guest: Uh like even when we had 50 integrations, we I I didn't really consider our product viable because for a given for a given business that's considering adopting you, you need to at least be comprehensive for their tools.

Guest: So we'd go into these calls and be like, do you have Gmail?

Guest: Yes.

Guest: Do you have Slack?

Guest: Yes.

Guest: Do you Google Docs?

Guest: Yes.

Guest: Do you have Mailchimp?

Guest: Yes.

Guest: Do you have DocuSign?

Guest: Womp womp.

Guest: Like, and they'd be like, ah, I guess I'll just stick with, I guess I'll just stick with Zapier.

Guest: And so we had to hit a critical mass for enough businesses to really be viable.

Guest: And now so for for a good chunk of the last year integrations was our absolute top priority.

Guest: Now it's still an important ongoing priority, but I no longer feel like it's the biggest blocker to our company succeeding.

Host: Why not go through the approach of because some companies, especially in the enterprise space or more like soft uh deeper, you know, platform or IAS layer do this is where you have a lot of integrations, um they open source um or like give an open source way to add integrations.

Host: I have you thought about that?

Host: What are the pros and cons of doing that?

Guest: There's a couple ways that we've seen people do that.

Guest: There are some that have like open source integrations that can be provided back into the main product.

Guest: Like Active Pieces is a company that allows that.

Guest: And there are other companies where they allow developers to build their own kind of like packs or extensions.

Guest: Like Make.com allows that.

Guest: And there are others that allow third party developers to build their own integrations, like Zapier.

Guest: Um, all reasonable approaches all have pros and cons.

Guest: Our philosophy have has has always been, we want our integrations to be really robust and really high quality.

Guest: Like if an integration exists in relay, it works.

Guest: Um, and what we found is that the broader the aperture um a company creates in who can publish which kinds of integrations, the more flakiness and unreliability that you have.

Guest: We will probably need to at some point open up a platform and allow additional developers, but I don't know that that's the right approach in this day and age.

Guest: I think the right approach in this day and age is to have amazing software engineers design the platform and the framework and then use AI coding tools to augment themselves to build tons of integrations.

Host: Use AI internally to improve your capability versus externally outsourcing.

Guest: Yes.

Host: I like that.

Guest: I think that's going to be a better approach.

Guest: There's there's also the approach of people who don't want to build integrations themselves can use one of these iPASS providers, whether it's, you know, Paragon or Trey or Prismatic.

Guest: Um, I think kind of a controversial take, but I think that's the worst of all possible worlds where you end up doing a ton of work to to implement them and then you don't control your destiny if something in the integration doesn't doesn't work perfectly.

Guest: And for us, integrations are so core to our business that like we just can't let we can't let that be in the hands of a third party.

Guest: We need to we need to own the experience.

Host: I mean, there is some overlap here in terms I mean there's a lot of overlap with Make.com, Zapier and these are legit businesses of significant size.

Host: Uh do you see them as competition and in that case like what how are you differentiating from them?

Guest: We see um when people are deciding between Relay and other products, we typically see them considering two categories.

Guest: One category is they're considering the traditional automation players.

Guest: That's Zapier, Make.com, and if they're um more enterprisey, maybe Wocado.

Guest: And then there's the new AI agent builders.

Guest: That's Lindy, Gumlu, relevance.

Guest: Those are the ones that we see we see most often.

Guest: With respect to Zapier, our main differentiations have actually stayed pretty consistent since the beginning.

Guest: It's a way easier product experience for

---