Latest / The AI Ecosystem Podcast / The MCP Moment: Why Your Tech Stack is Already an AI Ecosystem

Transcript

The MCP Moment: Why Your Tech Stack is Already an AI Ecosystem · The AI Ecosystem Podcast

• Bob Moore: Hello, ladies and gentlemen. We are back on the AI Ecosystem Podcast. I had to think for a second, what's the name of this podcast? if I look different, Lindsey DeFalco: Where are we? Bob Moore: I'm coming to you live from the beach this week. But even when the surfing starts, the podcast doesn't stop. just kidding, don't surf. sitting in my office, taking Zoom calls all day. but hey, happy summer. Lindsey DeFalco: I was I I was gonna say it's obvious that you don't surf 'cause I don't think people surf like three PM. I think that's more of like a six AM activity, but I also I also don't surf, so who knows? Bob Moore: That is really you caught me red handed, yeah. I'm yeah, I did eat some ice cream today. If that is if that's a good that's a good substitute. by the way, I'm Bob Moore, Lindsey DeFalco: Yeah, Bob's Bob's Bob Moore: CEO over here at Crossbeam. Lindsey DeFalco: And I am a Lindsay DeFalco, Chief Product Officer over here at Crossbeam. and if you're watching the video version, you should just be glad Bob doesn't have his sunglasses on, which is usually his his state taking in calls all summer. It's sunglasses on. Bob Moore: Yeah, that is my my my meta Ray Bans with the inductive headphones built in. I doubt I've given this rant on the podcast before, but they are my favorite piece of consumer technology that I think I've bought in five years. And it's because they're not trying to do anything new. They're just combining Lindsey DeFalco: You're Bob Moore: a bunch of awesome stuff that all independently existed. like Yeah, induction headphones, where you don't have to have something in your ear that works super well are amazing. And yeah, front facing cameras are super useful, especially when you can trigger them with voice activation commands like you have in your Alexa at home. And Yeah, sunglasses are great. I hate squinting in the sun. it's like the and and the voice command like linking into LLMs that can then look at the photos you're taking with the glasses. Like these are all just the individual components of photography and voice recognition and audio management through the headphones and the directional mics. Each of these things individually is like a nothing burger. piece of functionality that anybody can put into anything these days. But the way they've combined these into a form factor where you don't even look like, unless there's a tech nerd around who spots you, you don't even look like you're wearing anything but normal sunglasses is so incredibly compelling. and yeah, I go I'll go on a walk in the morning and just throw these suckers on and I can see my neighbor and talk to them and not feel like like an a-hole with my earbuds in every time somebody sees me. And I can be playing in the pool with my kids and take a video all of a sudden from like a first person GoPro style perspective without having to touch anything or like let go or grab my phone. It's like takes you out of that vortex where you always have your phone in your face if you want to be capturing memories or doing any kind of interaction with with people. So anyway, big Big endorsement of that product. I love that thing. Not a paid app. Lindsey DeFalco: Yeah. Bob Moore: Hey, Meta, if you want to send me a free version of the Gen 2s, I'll take them. Lindsey DeFalco: Say that concludes the advertisement portion of our podcast. Thank you. Bob Moore: Yeah. Also Athletic Greens, you can get a free ninety day trial by using code Crossbeam in in your checkout. cool. So what are we talking Lindsey DeFalco: the running Bob Moore: about today? I I I know we're in we're in a wild time and place with C P stuff. Lindsey DeFalco: Yeah. Yeah. I think we wanted to get back to a topic that we touched on a bit more high level in the early days of the podcast, which is and it'll be great to do a check in on like where we are, where the market is. but it's really this whole topic of now that everyone has an MCP, which I don't even think was true even remotely when we started this podcast. So the timeline there is crazy. but now that there's kind of a baseline expectation that, you know, every tech company has has an MCP server, what does that mean for downstream value that you used to be required to get out of custom built integrations or integrations Bob Moore: Hm, yeah. Lindsey DeFalco: built between two different products. So I think that's like the the broader stage set and we're gonna get into some fun stuff around a hackathon project I did related to this that's live now on our Crossbeam website. Bob Moore: Yeah. It is I I think the thing that your hackathon project made me think of, not to bury the lead, is that They're classically speaking, when you think about integrations between products, obviously you can directly integrate two products and like one product in its code very often has to talk to another product through its API, right? Or maybe you Lindsey DeFalco: Mm-hmm. Bob Moore: have some kind of API to API thing, but you always need some kind of middleware like a a Zapier or a Workado or Trey or whatever in like the the iPass universe to make, hey, there's an API here and there's an API there, how are these things gonna actually combined to be something more valuable. So this idea that the number of ways you can combine products grows at a rate of n squared was kind of a falsehood. Because really it only grows at like the order of n, where n is how many companies are willing to build integrations into other stuff. Or at the order of like some s you know, log n or something that is sub n squared, which is like the rate at which they can be supported by these IPass solutions that kind of allow for this in an ideal world, A connects to B and you can just mash any two of them up, which really, you know, because of the nuances of how connectors serve back data and how that data needs to be mapped and wrapped and like look we use an iPass solution for some of our integrations at Crossbeam and we still have to hire a third party company to build a bunch of these crazy maps. Like it's not just a plug-play n squared thing, right? So it's been easier than ever developing like joint ventures and integrations and software products ever was before the era, but it's not without work and it certainly doesn't grow at n squared. Agents actually unlock this n-squared world. Where because of their ability to reason and their ability to sit there as an intermediate interpreter of data, if I have an API on my side and you have an API on your side, and you know, for the sake of easing access to data and creating really healthy, clear contracts and whatnot, we'll just start saying MCP instead of API, right? Like a layer on top of the API that the agents are particularly inclined to work with and understand. Now an agent can just sit there and you literally can n-squared anything. And like, if if I have an MCP and you have an MCP, I don't need to build an integration into your product unless, for some reason, the way in which our customers want to consume that integrated data is not through an AI agent or an AI-driven experience or something with AI in the middle. And behavioral tendencies are pushing us in a way where like. That's true less and less. And and you know, nearly everything is is being consumed with some kind of AI in the middle, if not literally the downstream work that's happening as a result of that combination is being done by AI. so where does this leave us with cross mean, right? So th th th the real th the this this theoretical okay, we're going to a true n squared environment, which is freaking awesome and and a little terrifying. leads us to a place where there's a really cool thing that's true and you exploited that to our great advantage in your hackathon project. And Lindsey DeFalco: Yeah. Bob Moore: I'd love to have you talk about what that project was and and maybe even show it off a little if if you can. Lindsey DeFalco: Yeah. and and just before that, something you said reminded me as a product person and former engineer, there was a whole movement for a while of like no code, low code solutions. And that was like the tagline. I don't know if you remember that era. but a lot of it had to do with like, you know, iPass, okay, no code or low code, whatever. And it's funny because I'm like, that should be the tagline now, but now Bob Moore: Yeah. Lindsey DeFalco: it's kind of like who cares if there's code. Bob Moore: Yeah. Lindsey DeFalco: because I'm not writing the code. But it's funny because like we finally actually achieved a reality where I can build something with no code that is a very deep intelligent integration that can provide value. And it's crazy that we've actually finally gotten there. Now no one's saying no code, low code, because like the paradigm is it's completely shifted and it's it's totally different. So Yeah, exactly. Bob Moore: Yeah, and code is no longer the bad guy because we can generate amounts code. it's like the bad guy is just like the human factor and how many how many tech bros need to internalize it should be you know, no bros, low bros, instead of no code, low code. It's like how many of these Lindsey DeFalco: No I I hate that. Bob Moore: how many of these unnecessary intermediate like steps of development should should be removed from this? We can cut that out of the podcast if we think Lindsey DeFalco: Yeah, no. absolute Absolutely not. I'm gonna make you a shirt that says no Bob Moore: you know what I'm gonna stand by. I'm gonna stand by it. Lindsey DeFalco: bros, Mo Bros, on it. And people can just guess what it means. yeah, no, but I mean, to your point, it's like the enemy used to be human time, like human engineering time. and it was always like, you know, engineers are so expensive, their time is so valuable, and It's it's kinda like the whole, you know, if you have a horse, you're gonna ask for faster horses instead of a car, and we're like, w we have the car now and nobody cares that it's you know, no code or low code. Bob Moore: Yeah, with the self driving car. Lindsey DeFalco: Yeah, yeah. okay, cool. So yeah, this hackathon project. So this has been something that, you know, I mentioned we've chatted about before on the podcast. I talk about a lot, and I get like a variety of reactions of like head nods and and eyes glazing over when I'm like If I have an MCP and you have an MCP, then we have an integration. Bob Moore: Yeah. Lindsey DeFalco: and so and you know, then like you said, you extrapolate that out. Like I have a tech stack of tools that I use to make product decisions. I can have an integration with all of them if they all have MCPs without actually writing any code. So what Bob Moore: Yeah. Yeah. Lindsey DeFalco: I what I wanted to do is turn that into something that's actually tangible. and that actually showcases what you know what I mean when I'm saying two MCPs side by side is is an integration. So the hackathon project was and I'll I'll walk you through the different steps and the but I'll I'll spoil the ending, which is that the output is this massive directory that is a logo wall. That essentially outlines all of the use cases that Crossbeam's MCP can achieve when paired with another technologies MCP server. And that's really the point. And that's probably the most basic version of the point is that in a one-to-one in a one-to-one relationship of Crossbeam MCP plus you know, other technology company MCP, there's a lot more you can achieve than either one of them on their own. So yeah. Bob Moore: It's so cool. I I do think there's to tie this back to like the theme of our podcast. You think about the limiting factor in technology partner ecosystems historically, it's been how many of these integrations can you build that are actually credible, where you actually have customers that use them in order to get to that one plus one equals three story. In order to get to that we're better together value story. And the thing that Lindsay is about to show you is kind of an articulation as to like you can make an argument that we now have a you know marketplace of several hundred technology partners and very viable real world use cases of how. you can use our technology and theirs at the same time in a way that makes both of the technologies more valuable. And yeah, you know, no no additional layers of code or integration or IPass or anything else needed to exist in order for us to achieve that. and it's You know, as you think about account mapping and co selling and cross selling and building out ecosystems and kind of being a part of kind of what we consider to be a new world of tech stacks that come together, it is a game changer. and it has it even has me looking at our business in a in a pretty different way. Lindsey DeFalco: Yeah, and also a really important note about this hackathon project is I didn't win. So that tells you the quality of the rest of the hackathon projects. but yeah. Yeah. So we're gonna Bob Moore: Yeah, why we talking about this? We should talk about the ones Well, you know what's funny is that the one that got first place is is also in the family of the eradication of dependencies on some of these solutions we're talking about. Lindsey DeFalco: Yep, yep, that is very true. Bob Moore: Yeah, yeah. Lindsey DeFalco: there was a theme. Okay, cool. Yeah, so I'll talk through a little bit about like what I did and how I got there. So, and I'll I'll get to a visual for those of you that are on video and then I'll also say the URL out loud so you can just go look at it yourself if you want. so what I started with is I wanted to identify all of the different technologies that our customers used. so essentially get an idea of what are the tech stacks that our customers are actually using alongside Crossbeam So for that, I started my project with hashtag ad. Trace, which Bob has talked about on previous podcast episodes, which is the third-party technographic data provider, the entirely separate separate product that we've built in-house here. And so I started with this technographic map of all of our customers to see what technologies are used the most by by which of our customers. And so that was really the point of that is to narrow down what technologies to focus on based on who would actually use them and the you know the the hope that these companies are already plugging in the MCPs for the technologies that they use the most. And I'll kinda pause after every step, Bob, if you want to like chime in with anything but Bob Moore: Yeah. This already is like a really neat it's like you when you're when you're looking at the very, very early stages of defining your the total addressable ecosystem, basically, this is a really cool complement to Crossbeam, right? This is exactly yeah, I'm I'm I'm full of t shirts today. but like the back in the day, right, the only way to to do that kind of analysis was To get onto Crossbeam, hope that they're all on Crossbeam, partner with them all on Crossbeam, get them to share data or at least share overlap counts and do some big comparative analysis. Now, you know, we have that level of deep data with our our pre-existing partners, but what Lindsay was able to do here is also compare at least a rough cut of the data against 20,000 other companies using the data from Trace, even before we're connected on Crossbeam to validate the fact that they ought to be a partner in the first place. Which then leads to what what is the next step that Lindsay talked about next. Yeah. Lindsey DeFalco: Yep. Yeah, exactly. So the next step is do these companies have MCP servers? I also did an additional layer there, which is do they have MCP clients? and maybe we'll get into that after we talk about the MCP server to server. Bob Moore: I'd love to talk about yeah, what's the difference between those two? Maybe we should actually let me take you on a tangent. What's the difference between a Lindsey DeFalco: Yeah. Yeah, let's do it. Yeah, yeah. Bob Moore: server and a client? Lindsey DeFalco: Yeah. So everything we're talking about right now with two MCPs, you know, working with each other side by side, it all requires that there is an MCP client or some sort of AI platform that is able to access both both of those MCP servers. So if you look at Claude, Claude is, you know, or Chat GPT, like it would be very reduction reductionary redu reductionist. It would Bob Moore: Productive. Lindsey DeFalco: redu reductive. That's the that's the that's the version of it. It'd be very reductive to say that they're MCP clients. but they serve a a bigger purpose that covers the use case that MCP clients serve, which is that they can Yeah. Yeah, Bob Moore: They're like super clients. They're like yeah. Yeah. Lindsey DeFalco: exactly. and so what all the connectors do is allow you to connect, you know, multiple servers into this super client. But that's not unique to those AI platforms and so that's where MCP clients come in. So Bob Moore: Would a good example be so sure, you can load up Claude and you can pull in the Crossream MCP and also the Gong MCP and you can Lindsey DeFalco: Mm-hmm. Bob Moore: compare call recording data to your ecosystem data and boom, there's an example. But also Gong has their own agents, and those agents can be clients that can also talk to the cross-ream MCP directly as like primary. gong owned clients and there you might also wire in like the Chili Piper MCP and then get a bunch of meeting data or something in there all kind of coordinated. Is that like yeah. Lindsey DeFalco: Exactly. Yeah, that's exactly it. And really the question is, do you want to pull in data from other companies into your own product and use that within your agents and your AI tooling? In that case, you need to be an MCP client so you can connect to other people's MCP servers, which is the Gong example that that Bob just gave. So and Gong does have this. You can and actually Crossbeam's MCP server is connected in there. So you can pull in Crossbeam data. into specific agentic features within Gong. So that's an example of an an MCP client. But then there's companies where they will never build an MCP client. So Crossbeam is an example of that, at least now. Never say never. But you know, our our strategy is more focused on getting our data into other MCP clients outside of Crossbeam. And so then we have our our MCP server. So you know, a lot of companies that have MCP clients will also have MCP servers, but not everyone who has a server will have. a client. So I I say that because looking through all the technologies when I did the exercise of seeing what all the different products people use, some of them are are not gonna have servers because they want people to do the work inside their product. So their goal is going to be to have an MCP client that can ingest all of this external data to to do whatever they want. Bob Moore: Got it. That makes perfect sense. So people might be and this is true of for example some of the larger CRM providers that haven't moved as quickly on their MC stuff, right? Like they want you to Lindsey DeFalco: Exactly. Yep. Bob Moore: just build agents in their platform. They're stingy about their own data and letting it go and get pulled into other places in hopes that they end up being the home where a lot of this stuff gets built. So there's a little bit of a turf war around like do as I say, not as I do with being partnerly and Lindsey DeFalco: Yep. Bob Moore: data open. Yeah. Lindsey DeFalco: Yeah, yeah. And it's I mean, CRMs are a perfect example, right? You look at like HubSpot and Salesforce, they're gonna focus their energy on being able to ingest other MCP servers as opposed to having their data be available in an MCP server. And that is Bob Moore: Mm-hmm. Lindsey DeFalco: where they've spent most of their time is they're like, Hey, we have a directory, you can you know, register your MCP connection as like an app or connection or whatever we're gonna wanna call it. And we'll pull it into Agent Force, we'll pull it into, you know, Breeze, we'll pull it into, you know, whatever slice of the the CRM functionality they're looking at. So yeah, that's a that's a great example. And they're not gonna be as excited about an MCP server, because why would they if they want people to work inside the CRM? Bob Moore: Mm-hmm. Yeah, yeah. Lindsey DeFalco: cool. So I did that slice to see who has servers and who has clients, because the whole narrative about, you know, how how we integrate is different if we're talking about a client than if we're talking about about Two servers, two MCP servers. but for the purposes of the entire output of the project, I just focused on MCP servers side by side. Bob Moore: Mm-hmm. Lindsey DeFalco: okay, so once I had this, I now have this whole table that's like all of our the top technologies that our customers use. I have do they have an MCP server, yes or no? Do they have an MCP client? Yes or no, and then links to the documentation for all of the the MCP servers. So the next step is now what can they do? And what use cases can we jointly achieve with our MCP servers? And Bob Moore: Mm-hmm. Lindsey DeFalco: so this is where I set Claude off overnight, over the course of a day. I spent I I kind of maybe accidentally ran it on Fable. I can't decide if it was intentional or an accident, but did a great job. spent a lot of money overnight, which which is the point of the hackathon, right? Bob Moore: Of course, yeah. Bankrupting the core business is always a Lindsey DeFalco: Yeah. Bob Moore: the number one principle, as we mentioned on our our earlier hackathon episode. Lindsey DeFalco: Yeah. but in reality actually I think I think what it ended up being was almost like a dollar a page. and once we get to the output of this, like it would have taken somebody months to put together all this content. Like it is it would have taken me I probably could have done three a day, maybe, if I was like only doing that. Maybe like five a day. So And we generated 171 of these of these pages. So okay, so we have the we've we've now sent Claude off on this mission and it's what does their MCP server do? So what MCP tools are available? what functionality can you actually get from that MCP server? It has all the information about Crossbeams tool set and our MCP server, and so now the exercise is okay. What use cases can we jointly achieve if you use Crossbeam's MCP server and let's say Clay, Clay's MCP server. That's the one I'll I'll pull up as an example. and this is where the content that we actually put on the marketing site started to get generated is these one pagers that can basically answer the question, if I use clay and clay has an MCP server and I use Crossbeam and Crossbeam uses an MCP server, what use cases can I actually achieve by asking one question that hits both MCP servers? So I'll pause again, but I I think I'll move into showing next, but if you wanna Bob Moore: I think the thing we're we're also saying without saying it here is, you know, I mentioned earlier, okay, n squared, right? Like the number of MCP servers that exist squared gets you the number of theoretical integrations. I do want to say the obvious thing out loud, which is a bunch of those integrations will make absolutely no logical sense and have no business value for anybody, right? Like there's only so much, like if I You know, if I'm connecting Lindsey DeFalco: Yes. Bob Moore: a construction software company's MCP to my barbershop's MCP, there's not much they're gonna be able to do to deliver business value to their mutual customers by by combining that data. So part of the the The beauty of what you were able to do is in theory, if there are thousands and thousands of those MCP servers out there in the world, you could have built thousands and thousands of pages, and it's like Crossbeam times everybody. But what you were able to do is using a little bit of Crossbeam magic, a little bit of trace magic, and a little bit of LLM magic, reduce that universe of thousands down to kind of the the folks where all the conditions are met. There's an overlapping target market. There is an actual real use case that we can articulate and kind of show evidence of. And the MCPs exist and the underlying technical underpinnings are are actually there. And that that's what makes this thing so powerful because what came out the other end is actually the stuff that people want and are going to care about. And if they actually go try to use it and use these tutorials and guides and information, it's going to work. And it's going to be really cool. Lindsey DeFalco: Yeah. yeah. the entire exercise of all the different steps it takes to get to this output and actually generating this output. this was probably the most fun I've had doing a hackathon because it truly it truly would have been like a a behemoth of a project pre pre AI from like the, you know, designing of the actual pages which we had our wonderful brand designer help with, the the rate at which she was able to churn these out. using AI tooling as well. It's just crazy. So this is the output and I will I'll kind of read through it and try to narrate for folks that aren't on video. So the header of this page, it's at crossbeam.com slash partner dash MCPs. So MCPS, which now that I'm saying that out loud, kind of sucks. anyway, here we are. So the header of this page is Crossbeam MCP works with your AI go to market stack. And that's really the entire pitch of having an MCP server in this day and age is just that. It's all of a sudden you are automatically a part of this. AI go to market stack without needing to build these direct integrations that you previously would have needed to build. So, what we get if we go further down, in the middle here, we have a section that talks about the different core use cases that you can achieve with multiple of these MCPs that we're gonna get to. and the header here is from a plain language ask to a go-to-market workflow. So modern day no code, low code. but the point here is that you can ask one question. so we have you know an example prompt here and it says, pull the greenfield accounts from Crossbeam where a partner already sells with the contact who can introduce us. Use 6sense to rank them by buying stage and enrich the top 20 in clay. So insert any tool names that are in your stack. And it's giving you an example of how you can jointly use a variety of different MCPs. The thing I didn't get to in the hackathon and that we won't get to yet is like We're talking about one to one here. but in the middle of this page, there's examples where we have like, you know, three or four different MCPs. And, you know, like Bob said, this really does expand out to as many as you want. And there are MCP marketplaces out there and you can imagine like on the marketing site being able to select your whole tech stack and then we can tell you what use cases you can satisfy. And that's kinda like Bob Moore: You're totally right. I it's not order of n squared, it's order of n factorial, which is even more gnarly. 'cause you you Lindsey DeFalco: Yeah. Bob Moore: could in theory have any permutation of any universe of all the MCPs that exist, which is like you get soon you get into like there's more ways you can do this than there are like grains of sand on the planet Earth kinda kinda number. It's Lindsey DeFalco: Mm-hmm. Bob Moore: crazy. Yeah. Lindsey DeFalco: Yeah. and I want to talk about that after this actually because I think the whole challenge with all of this is like knowing what to do with it. And there's too there's so much possibility right now that it's hard to like rein in what you actually want to do, which is what the market struggles with. Okay, so let's get to the good stuff. This is it. if you're looking, dun dun. If you're not, it's a logo wall. says Bob Moore: Yeah. Lindsey DeFalco: your go-to-market stack is already an AI ready ecosystem. And we have all of these logos. So there's 164 of these. And again, they're ranked by specific to crossbeam what we think our customers will care about the most based on the data we have from crossbeam and from trace. Bob Moore: Yeah. Lindsey DeFalco: So let's see, let's pick one. da da da da da da is there a specific one you want to test here, Bob? Bob Moore: Why don't you click on let's do Gainsight? Lindsey DeFalco: Okay, let's do and now we're testing this project because I definitely haven't clicked Bob Moore: I know this is like the live demo. Lindsey DeFalco: Okay, let's do there's GainSight and then GainSight CS. I don't know what the difference is here. all right, so now we're on the page that is specifically for GainSight plus Crossbeam MCPs. And this is this is the one where all of the specific use cases and information about what you can do with those two MCP servers exists. So let's see, it generated Five different use cases here. So da da let's see. Prep a co-cell call with the questions the account already asked. So Crossbeam returns the contacts a partner shares at the account plus a recent partner deals opened or closed one. Gainsight searches community posts and topics, so you walk in knowing what those users have already raised. Bob Moore: She Lindsey DeFalco: I mean, that's pretty interesting. Bob Moore: Yeah, that that's actually and we do have a you know, a a classic style partnership with Gainsight, to a certain extent and a lot of customers in common and that kind of stuff actually does happen for real. Lindsey DeFalco: Yeah. And I think this actually pulled from Gainsight Community, which is Bob Moore: Yeah, I think this might be the gain set like the content end, like the Yeah, Lindsey DeFalco: Yeah, yeah. Bob Moore: yeah. Lindsey DeFalco: Let's see if the other one did. here Bob Moore: Yeah, customer seats. Lindsey DeFalco: we go. Yeah, yeah. Okay. I switched over to the other gain site one. and here we have like build an expansion list from healthy customers with partner coverage, check an at-risk renewal and log to partner play, match partner performance to customer health. And some of the example prompts here are like Let's see. Pull my accounts with healthy scorecards from GainSight, then ask Crossbean which partners share them ranked by win rate. So yeah, the again, all of this was was claw generated, but I was pretty impressed with how it was able to actually understand what GainSight does, for example, what Crossbeam does. And then all of this is again like framed in what the MCPs can actually achieve. So if you Bob Moore: Yeah. Lindsey DeFalco: have both these MCP MCPs connected, you should be able to just put these example prompts in and actually get results, which is pretty freaking cool. Bob Moore: It is very cool. and I think I'd be remiss not to call out the SEO and AEO benefits of this project Lindsey DeFalco: Yeah. Bob Moore: as well. because this the there's a little bit of a throwback here back at Stitch, my previous company. Hey, Lindsey DeFalco: Hey bingo. Bob Moore: on the bingo board, we built a bunch of these content microsites that were not nearly as good or as cool as this and required much more human labor. But at their core I had a very similar concept, which was we owned the domains two, like the the letters T O and then the name of all the big data warehouses. So we owned two redshift.com, two bigquery.com to Azure SQL.com. Lindsey DeFalco: that's cool. Bob Moore: And we stood up these websites that had the names of companies with APIs as the subdomain. So it would be like Trello.to Redshift.com and the website would render for you this really cool landing page about like, hey look, Trello has these APIs, you can get this kind of data out. Hey look, Redshift is a data warehouse. You can store a bunch of data there for analytics. How on earth do you put these things together by Stitch? Right? So it was Lindsey DeFalco: Mm-hmm. Bob Moore: a great sales pitch for Stitch, but you what Stitch was? It was one of those freaking eye pass providers that sat in between that we don't have a name Lindsey DeFalco: Mm-hmm. Bob Moore: anymore. So what's cool about this is It's not I I think what it brings to light is if you're asking questions about how A could potentially combine with B, both from a value story perspective and a how do I do it perspective, it gives both agents and humans a really good landing zone to find that. And by the way, the only examples we give here is where either A or B is crossbeam. And that's how it benefits us, right? Is it it it tells all those stories from the perspective of this is why Crossbeam should be in the mix, this is why Crossbeam should be in the stack, and that's kind of our whole MO from a marketing perspective in you know, the second half of this year. Lindsey DeFalco: Yeah. Yeah. And there's there's two specific examples too that I think are interesting from like an ecosystem standpoint. So let's see Euler here. Euler's a PRM, right? and they have an MCP server. And this first this first example here, which for those of you on audio is register a partner deal from Crossbeam Overlap. If I had a nickel for every time this has been requested as an integration, as a feature, I would have, I don't know, maybe $5. But this is such a primary use case for our customer base. And being able to integrate with PRMs, especially in an AI native way, is just it's something that we've gotten requested over and over and over again. And it's like one of the more high value things that that people want to be able to do and don't know how to do. And all these examples are like check attribution against real overlaps, recruit and onboard suggested partners. Like this is y people that have Euler have an MCP set up, have Crossbeam have the MCP set up, they have an integration between their PRM and between Crossbeam. so this is having like, you know, deal registration done. We have a deal registration integration now because we have an MCP. Bob Moore: Wild. Lindsey DeFalco: And there's one more in here, which is Bob Moore: There's more Lindsey DeFalco: but wait, there's more. So another one I wanted to call out, which I think is just really cool, again, because it's it's one of those things where I think people have kind of mentally given up on there being an integration between them. but the hyperscaler marketplaces. so AWS Partner Central is what I have pulled up right now. Bob Moore: Yeah. Lindsey DeFalco: And AWS Partner Central has an MCP server. I don't know if you know that. If you didn't, now you do. and I have run an example, a live example actually of the AWS Partner Central MCP with the Crossbeam MCP. and it's really freaking cool. So there's an example here, get an opportunity summary next to your sharing status. so actually that's not the one I'd show. Here we go. Turn shared accounts into ace opportunities. So again, this like registration workflow where you need to go register a lead or a deal in a Bob Moore: Yeah. Lindsey DeFalco: partner central, you can do that automatically with a workflow based on some trigger that happens in in Crossbeaver, some some change that happens with your partner. Bob Moore: That is incredible. ca I like getting into a whole other universe of technology platforms that exist historically just to do that, right? And it's like kind of there's a lot of data manipulation and kind of providing a human in the loop contextually you know, adaptive management of how to go from, hey, this is something that I'm working on in my siloed CRM environment to this is something I want to transact through the marketplace. And how do I send the right information over and structure it in the right ways and scale that and create pipelines around it? There's a whole industry around that. and there's a bunch of companies that have gotten a pretty material scale with that as their core value proposition. That and and like helping companies get listed on the marketplaces. And again, right, with an agent in the middle and two MCPs, it kind of eradicates I I think the necessity of that highly structured workflow-based approach to doing that work. and if you Lindsey DeFalco: Mm-hmm. Bob Moore: build the right harness around whatever it is that has access to those two MCPs, it can be immensely powerful. and I think you know a lot of our own collaborations with the hyperscalers are are are heading exactly in that direction. Lindsey DeFalco: Yeah. Yeah. So those are mm those are a few examples of you know, different technologies that are in and around our stack. And that was that was the hackathon, is creating these these landing pages that really showcase the the joint use cases and you know, these like I don't want to call them free integrations. So they're not free, but these integrations that you get built in when you have the the MCP servers connected. Bob Moore: Yeah, totally. this is so cool. and it's great to see live. And I will, you know, you pulled some data recently, and I don't know if it's cool for us to show it here, but I'll talk about it in broad strokes, which is just the consumption of MCP credits and the number of requests that are happening of our specific MCP server is growing. Way over a hundred percent a quarter and some months way over a hundred percent a month, like month over month right now. And it is in the this is not like it went from two to four, it's like in the tens of thousands to hundreds of thousands of requests kind of kind of zone. it I didn't even realize the rate at which that's happening and it is not slowing down. particularly with our MCP just now going into going into GA and we didn't even have our sales reps like armed to talk about it, right? This was all like inbound demand up to this point and now it's kind of released to the masses. for folks that are on our supernode tier and up, it's and connector tier with the with the platform fee, I believe. it's crazy time. and I I just Lindsey DeFalco: Hmm. Bob Moore: can't wait to see those numbers continue to climb. And it almost feels it feels inevitable because that modality shift is when you look at the data We have our our sales teams kind of like armed to the teeth to bring up the topic and they're doing a great job of bringing up the topic of MCP. We see it all over the place. It's in so many call transcripts. And yet despite that, as many customers are bringing it up before the sales rep can even mention it. It's like almost a fifty-fifty split. and that is also so fascinating to see. it's like it's it's a standoff as to who's gonna get the opportunity to say the word MCP first. think this kind of stuff is why. It's why all the companies care. Even if the people that are buying or the people that are selling any of these products, not just ours, are fully selling or articulating what we're talking about here. I think the underlying motivation from their companies themselves and the people in their companies standing up these AI stacks is all driven specifically by this realization, right? That this like this is the ultimate meta integration. movement and it will eliminate so much work and overhead and confusion and cost in the universe of ops. and man, are we there? We're like there. It it's it's it's about the human adoption and it's about going as like low bro, no bro as you can and bringing this stuff to life. Lindsey DeFalco: I'm definitely getting that on a shirt for you. Bob Moore: Yeah. Lindsey DeFalco: yeah. I think the other thing I love about it too, from an adoption standpoint, is that once once anyone on the team has access to the MCP server, then anyone on the team has access to create a proof of concept to prove why it needs to be part of XYZ workflow. And that's actually what happened interestingly with with us with this MCP server. for a company called Interpret that does they essentially analyze like feature feature requests and discussions about certain product functionality from call transcripts. And we got the MCP connected and I was like, well no one no one's using it directly on on the team. But then somebody put together a proof of concept that was on a different team about how it could help. I think it was something in like the the customer success world. and now it's part of multiple workflows. Because individual people make these proof of concepts and then basically can, you know, rally to get it set up by ops to be part of you know, larger workflows or orchestrations. And so for people out there that are, you know, building MCPs or thinking about them from like a strategy perspective, that's something really compelling that you know never existed with with APIs in a no code, low code Bob Moore: Yeah. Lindsey DeFalco: way before. So that's been really fun to watch as well. Bob Moore: Yeah, and it it it does it reinforces what we've been talking about on this podcast all along, which is the collapse of companies that matter into are you a system of record for something or not? like are you an authoritative source for some very distinct piece of data or knowledge or insight that cannot easily be replicated or retrieved elsewhere. And differentiation through workflows and differentiation through the application or UI layer is like reducing to zero. So and which which is funny is you look at things like the CRMs not being super generous with what comes out of their MCPs, because there will be a finite number of folks that do get to hold on to being the application layer, but it's going to be the absolute biggest companies in the world with the most gravitational pull and the most market share and the largest pre-existing ecosystems. And I think all the big players are trying to make a make a stand and and make a play for that and ask everybody else to integrate into their client environments. and everybody else will, by the way. But with every single one of these that you allow yourself to integrate into, you are sub you are you are accepting your fate that your application layer will not be your differentiating factor anymore. And like you better you better have a damn good differentiate a proprietary source of data or intelligence behind what your company's doing or a reason why you're gonna be able to generate that by virtue of someone using your offering. and that's the world we're about to go into. And you know what, I don't mind that world. it's an extreme meritocracy in my opinion and it gets us closer to like it's it's a datatocracy really is is what it comes out to. Three for three on the on the Lindsey DeFalco: But yeah, you're you're on a roll today. Yeah. Yeah, it is funny because I do Bob Moore: Register that domain name before this podcast comes out. Data tocracy is pretty good. Lindsey DeFalco: yeah, it does like it, it feels like a repeat of the same there's always been a discussion of what app are you gonna log into, especially for really congested markets like tools for sellers, right? Like that's always the topic of, you know, do we integrate into this other tool where they already log in? or you know, are we trying to build a surface that they're actually gonna use? And it I think this is just accelerating that. And it's making the the big players that already had that that stamp that like my god, we're making that stand. They're just digging in even more. but it makes it easier for everyone else to get their data into their products, which I think will actually accelerate their ability to be successful in being Bob Moore: Nice. Lindsey DeFalco: the only ones. So it's just like a big a a big accelerant. Bob Moore: I love it. Hey, I this has nothing to do with our our chat here, but I was talking earlier about the metaglasses and I wanted to close on one other example of like one of my favorite new pieces of technology that knows how to use LLMs and AI technology, but just combine it with very simple things that we've known about and had for a long time. And you're actually looking at it in my frame right here. This little red box is called a sticker box. And It this I believe that this was like a GoFundMe project that turned into a real consumer company, but it does a really simple thing that is gonna be slightly out of frame, but I'll see if it's like slightly visible here. chord's pretty long. So I can say, show me a picture. here, I'll just hit the button. Two people recording a podcast across Zoom. Plays a little music, it says thinking thinking. It has done my speech to text. It is a transcript of what I said. Including the it's like real polite. There it is. So it generated a photo Lindsey DeFalco: Wow. Bob Moore: of the two of us recording a podcast, and what's in here is a thermal printer. I don't have any stickers in it right now. But what it does is you load the stickers up and thermal paper just prints a pattern on it because of heat, a heat signature. So you don't need any ink cartridges, you don't need toner. It's not Lindsey DeFalco: Well Bob Moore: receipts print, like if you get a receipt printed at the at the store. So This thing immediately then prints on the thermal paper that's in a roll on the inside, the exact image that you just generated, and you can rip it off like you would rip off a receipt and boom, you have a sticker. I tell you Lindsey DeFalco: That's really cool. Bob Moore: my kids love this thing, but I love it just as much. And it is again like thermal printers have been around like, I don't know, since since the sixties, in in like popular use. And it's just like a little Wi-Fi connector and the dumbest, like lowest cost, lowest token model that can generate like a 64 by 64 pixel black and white image and some speech and tech stuff, which has like reduced down to three. And all of a sudden you have a thing that literally like I show this to people aged two to 102 and their jaws drop on the floor and they're like, We're living, we're living in a dream right now. Like how is this even a thing that works? But it's just piecing this stuff that we have together with AI It in the same way that the metaglasses do. and it just really w creates something really transformative. Cost a hundred bucks. crazy probably cost Lindsey DeFalco: That's really cool. Bob Moore: them five bucks to make it. the most expensive thing is probably Lindsey DeFalco: Yeah. Bob Moore: the tokens it costs to generate all the images. yeah. Lindsey DeFalco: That's wild. I love that we are I don't know if you guys know, we got so popular that now we have to do advertisements at the beginning and end of our podcast. So that Bob Moore: That's right. Hey, sticker box, use code CrossMe. We twenty percent off your first order. They now have holographic sticker paper. All right. Hey, on that note, Lindsey DeFalco: Incredible. All right. Bob Moore: always a pleasure, Liz. We'll we'll talk again soon. Lindsey DeFalco: Yeah, a good one. See you soon. Right.