Agent Pricing is Chaos. Here's What We're Seeing From the Buyer Side | The Agents #015

Hey everybody, welcome back to your favorite podcast, The Agents Life, with us, 3 humans, 1 dog, and 21+ agents. This week we went through 22,000 of the executives, director and up, who attended our SaaStr AI annual events the last few years. Amelia will share the analysis. AI-native titles are up 4 times. And half of everyone has changed jobs in the last 16 months. Pretty crazy. And at the same time, almost every legacy vendor we use is raising prices on our agents once again.

Yeah. And one of those agents deleted a core part of our SaaStr Connect not once, but twice, then told Jason it wasn't responsible. So today, who actually owns AI now, what agent pricing looks like from the buyer side, and why both OpenAI and Anthropic are paying Upwards of $400,000 for someone to run events.

Okay. I'm Jason here with Amelia. 3 humans, 21, 22+ agents. And we'll kick it off with Amelia. I actually didn't think this would be very interesting, but it was. Amelia will talk through it, but Amelia basically works with $10K for hours every day, our AI VP of Revenue, which, as you guys know, if you follow this This pod or just us in general started off as a pretty simple dashboard and now really runs almost all of SaaStr, more of outbound, more of inbound, more of everything. And she asked it, 10K, to do a deep analysis, just who comes to our, our IRL events, SaaStr Annual the last 2 years, and just the, what's changed for us as a B2B community was pretty interesting. So what did you and 10K learn, Amelia?

You're listening to the official SaaStr AI podcast, brought to you by— Hey everybody, starting a business can get expensive fast. Website here, email somewhere else, business phone with another provider. Northwest Registered Agent gives you a complete business identity in one place with free tools, resources, and built-in privacy from day one. Get more at northwestregisteredagent.com/saastrfree. That's northwestregisteredagent.com/saastrfree.

S-A-A-S-T-R-F-R-E-E.

Hey everyone, if you're listening to The Agents, then you need to be at SaaStr AI Annual. We'll be back next May in 2027, the world's largest B2B and AI event, May 11th and 12th in the San Francisco Bay Area. Look, most conferences and content these days are about what's possible. As you know from The Agents, we're all about what's happening now, the latest and greatest, and that's what we do even more so at SaaStr Annual. See the actual work where we bring together 10,000 founders, revenue leaders, and VCs Plus 200 of the very best speakers showing you the playbooks of what's already running in production behind the scenes, deep dive demos, and so much more. Plus over 5,000 scheduled one-on-one meetings so you can actually meet the people doing all this, not just hear from them. If you're a founder, a revenue leader, or you're the one who has to make this all work with agents, this is where you want to be. You already know this from listening to the agents. That's SaaStr AI Annual 2027. Go to saastrannual.com And use our code THEAGENTS for a special discount on your tickets. See you there. Yeah, so it's interesting. We started this exercise of— we realized when I was doing some outbound to our paid tick holders of the last, you know, 2 years, just some recent folks, we ran it through an enrichment. I talked a little bit about our process last week, but just as a refresher, we use a combination of ZoomInfo, Symbal, and Clay just to see, you know, okay, if you're gonna go do outbound, I assumed the turnover rate in our folks and our buyers was high. And so once we started to enrich the data, I realized we had a pretty good subset of folks that are in, in B2B, in AI, coming to events, spending money, right? And so after we enriched the whole thing, I started asking 10K to do an analysis to see what patterns would emerge. And I think there's, there's a few high-level ones, but there's some I wanted to share with you guys and then get Jason's take on. Because I feel like it's important to talk about since we have this data that maybe nobody else has, since we have these 22,000+ paid ticket holders over the last 2 years or so. And so the first one, maybe this has to do with how we program SaaStr, maybe this has to do with how just everything has changed so much over the last 2 years in particular. So we looked at the data from 2025 SaaStr Annual. So for, for reference, that would have been in May of 2025. People would have been purchasing tickets. Leading up to the event in 2024 going into 2025. And then we also analyzed all the 2026 paid ticket holder data. So that's everybody 10,000+ that came this year. So that's, you know, 2025 into 2026, May of 2026. And then we looked a little bit at the early data from folks who have already bought for SaaStr 2027 who are buying now, right? And so that's how we put together this composite of 22,000+ people and specifically paid. Paid people only. So we took out all the free folks, the sponsors, speakers, stuff like that. So these are just paid folks only who are buyers who are throwing down in B2B and AI. First learning, yes, explicit AI-native companies and job titles have become about 4 times more common in 2026 than the years previous. So between 2025 and 2026 alone, just in that year cycle of May to May, Job titles and AI-native companies of, you know, who has AI in their job title or who is working at an AI-native company with AI in the company name, AI explicitly as a company, has gone up 4 times in the last year, which I think is fairly interesting that these functions are explicitly getting AI ownership. And a related point we actually found tied to this was 25% of our sponsoring companies right now, as of today, this morning, are hiring a go-to-market engineer. And so the roles have changed, what functions are getting like ownership has changed. We also dig into some of that data. But yeah, I wanted to get your take on this first, Jason, of, you know, maybe not a surprise, but explicit AI-native titles have gone up fourfold in just the last year.

I guess it's not a surprise. I mean, one caveat on the data, which I don't think is a bad caveat, these are folks from our community that have traveled in many cases across the world to meet each other each year, right?

Yep.

I think the flip side is that maybe folks who aren't more AI-focused and that nature are staying home. So, there is a selection issue here. Which is who, who comes. So our, our community has changed. We, we've changed. Right. And, uh, so I don't know that I learned too much about this other than what you said. I just think that, yeah, I mean, it's just, you, I think we've, we've learned we've gotta live in the present and the future. I don't know if we all need to be able to run 21 agents like, like SaaStr does, but. But these are the folks that are showing up and coming. I thought the more interesting one to me was the next set of data you had, especially on job turnover.

Yes. Well, it's interesting, related to this point was how, like, who's changed, right? So, like, explicitly AI-native companies and job titles are 4 times more as common. Maybe we could all have fiended that from just looking at our LinkedIn. When we looked at the data with 10K, it did tell me who was more likely now to own AI. And so the top folks it found more likely to own AI were the CEOs. You gotta lead from the front, right? Like I think Alex from Higgsfield was talking about this with Harry. And then next was go-to-market, and then next was COO. So it's not, you know, the kids coming in like the S— are going to own the AI strategy. Within that subset of data, it also found that most of the folks who have to own it now are the founders, the GTM leaders, or the COO has to, has to come and take in the charge on AI.

Yep. But you found your data. You said of the 22,000 folks, these are executives, 49.7% changed jobs in the last 16 months and 63% of CMOs.

Yes, super high.

Turned over in the last 16 months.

Super high. Yeah, it's something we've seen year over year, just even without like 10K doing this analysis. Like we've always known when I've sold tickets and when I've gone outbound to our sponsors, I've always known we have to constantly enrich our data because our buyers who are, are typically CMOs who are, you know, responsible for events and have that budget turn over so fast that they're usually not there by the next SaaStr annual. And so we always kind of knew that this would be true higher of CMOs, but it's interesting to see here in 10K's data.

Yeah.

That, yes. So he found that half of people across the last 16 months, just in that whole subset of 22,000, half the people changed jobs, but CMOs changed in particular at a much higher rate at 63%. So.

Yeah, it's crazy. There was, you know, we, there's a lot of like, just, just as we're recording this, there's been a lot of Turn drama turnover in the news. The CEO of MongoDB quit, left a $53 million pay package after 11 months to take an enterprise leadership role at Meta around Muse and all that. That was one. And then there was a lot of drama around the first sales leader from Snowflake quitting and leaving to go to Cognition, which just crossed a billion in ARR. It's a leading coding platform, autonomous coding. And You know, that, that's tough, right? And we were talking about it actually on the 20VC podcast when Jack Altman from Benchmark joined us. On the one, on the one hand, where's the loyalty today? Where, you know, quitting Mongo after 11 months as CEO when thousands and thousands of employees trust you, right? And it is kind of weird joining Cognition as CRO when you're on advisory board of a competitor. But the flip side to all of this, so it is stressful, the turnover, art, art, Data supports it. 63% of CMOs turn over in 16 months. 63%. That's like most. I believe that is most. I believe that the technical term is most. But on the other hand, the point that Jack made was that the good news is people are turning over so you can get them at higher rates, right? There's pros and cons. I mean, I guess a different way to view it is everyone's a mercenary. $53 million to run Mongo isn't enough. If Zuck offers me $530 million, right? I'll just go. So I don't think any of us who have been founders and leaders for a while, I don't think any of us are perfectly comfortable with mercenaries, but the fact that people are more mobile, and so it is in a sense easier to recruit them before is interesting, right? If you've got something to offer, it's, you can actually lure the first guy out of Snowflake, which a couple of years ago would've seemed impossible. How are you going to, he's so rich. How are you going to, how are you going to, how are you going to get him to do it? But today you can. So. Pros and cons, but I think you almost have to plan that your CMO won't be there in, in, certainly in 2 years, and that half your executives will turn over. That's what the data says. Half your executives are gonna turn over. Gotta constantly be recruiting them.

I've been talking to a lot of our, our sponsors who are, you know, AI-native company these days, usually AI-native companies, or at least, you know, legacy companies that are still growing very quickly. And their number one thing right now, they've all asked me for help with outside of SaaStr is hiring. Like, they are just relentlessly all hiring because of the turnover and how competitive it is, right? Like, just the incentives are getting crazy now to lure people to steal somebody from the competition.

Yeah, especially for, for very specific senior roles. There, there's almost no, no budget limit. In many cases, it's, it's a tough one. All right. Well, that, that's pretty, pretty good data. All right. Anything else you wanna talk about from the, from the data of 22,000 buyers, or should we talk about agent pricing?

I do wanna talk about pricing. Maybe last point. I think it's interesting for us because I think it's a challenge to many folks who'll be listening to this, which is, you know, we do, we've talked how we're, we're outbounding aggressively, right? We outbound every week, every day we are refreshing our outbound, making sure we can keep up with the buyers who are, you know, AI native coming to SaaStr across tickets and sponsorship, just everything, right? And in our last point that 10-K found, like of the subset of AI native companies and attendees that have come the last 2 years to SaaStr, more than 75% were not returning attendees, just of this subset, were not returning attendees. So it's just interesting that this new group is emerging of AI buyers that are not people necessarily familiar with SaaStr, who are not people just relabeling themselves from like a legacy company and coming back. These are new buyers that have entered the mix. And so I think to keep up, you have to be just insanely aggressive, whether that's recruiting or outbound or whatever it is, to keep up with these folks and keep tabs on them of who actually is in your market. Now, I feel like, you know, people do these segmentations and they're like, okay, here's my ICP, I'm gonna go after this. And then it changes like a day later.

Maybe a different way to put that, which I hadn't really thought about, is certainly for us, we've been doing this for a while, right? And I see my LinkedIn and I'll see plenty of Very talented folks that I respect that sort of were pre-AI leaders that are now joining, joining, you know, AI-native companies like first sales hire at Snowflake. Snowflake, you could certainly, is an AI-focused company today, but was pre-AI, right? You see these all the time, folks that we know. But what the data says is if you, if you want to, actually 75% of these execs are new to SaaS, right? So they're new, they're not the same folks. So the majority, The question is how many folks are recycled pre-AI B2B execs, right, that are leading these next generation companies versus how many are new, or at least new to us. I would've thought it was closer to 50/50, but no, 75% of the executives are new to SaaStr, new to us, and aren't recycled in a sense. So I'm not sure what that means other than that You know, folks that are still, that are pre-AI and want to get into the new world, you may have to work harder because 75% of folks are new to the community. So I would have thought it was more. It feels like many folks from the pre-AI world are finding their footing in the AI world, that they're landing. After all, I mean, maybe you are ElevenLabs and got to $600 million or whatever it was in, with Sam Blond in a couple of weeks, but But your executives, you know, if you want to hire anyone with more than 2 years of experience, you've got to find folks that were there before. But the data interestingly suggests that might be otherwise. Okay. Agent pricing, right? So what have you learned here?

I think it's all over the place right now. I was just talking to somebody this morning. I was talking to a founder this morning who helps, who literally does this all day, who helps CEOs And founders price their products. Some of his companies are like legacy SaaS companies, some are AI native. And it's— he's like, it's all over the place right now. So he'll— I'm sure he's going to come on the podcast in a couple of weeks. We'll talk about this a little bit more. But what I have found, even just in our personal stack of 21 agents, like, I don't know whether to call it madness right now or chaos or something in the middle, because I feel like it's all over the place. How can both Salesforce and HubSpot say that they're raising prices for direct agent access, and yet Muse is free. Yeah.

Well, I'll tell you, it is true. I'll tell you what I think. Listen, other folks, we're not the first to observe this. I think it's just more visceral for us. Just to step back for a minute, it's just interesting. We don't have all the answers, but I think many of the vendors we work with, and I'm not picking on anybody in particular, Many, many vendors that are systems of record or their platforms have decided to charge extra for agent access. Atlassian has done it. Salesforce is just rolling it out. Big banner when we log into Salesforce, you will be paying more for agent access. HubSpot is doing it, although appears that they appear to be doing it more for their first-party agents than third-party agents. They don't appear to be charging yet, which I would prefer, for third-party agents like 10K to access HubSpot.

Right.

But Atlassian's done it too, which is one of the reasons they've seen growth. Every, so every existing vendor has said, Jesus, I'm, I'm going to have seat contraction because of agents. Agents don't need, even if they needed, I think we have an agent seat at Salesforce, right? Or something like an API seat, right? But we don't need 20 of them or 80 of them. Every, every seat-based vendor's seen this. And so the natural thing to do on a spreadsheet and on a whiteboard is to charge more for those pesky agents to access it, right? They're ripping us off. Those agents are ripping us off because they're not buying enough seats. And they're, and they're abusing our software. And I certainly think if I was on the other side of that, if I were running a B2B company that had been around for years, seeing seat expansion risk due to agents, I'd be, I'd be having the, of course we'd be having the same conversations. Right. But what I just think is interesting is, you know, our agents aren't going to stand for it. So I don't believe these vendors are thinking about it from the customer perspective. I believe a handful of folks are. Dharmesh talks about this all the time, how important is an HubSpot.

Mm-hmm.

To not charge for general agent access. And I don't mean to pick on anybody. I don't know if everyone at HubSpot agrees with them though, as the founder and CTO. I bet many, I bet a lot of folks in revenue functions see differently. They want to charge more for everything. And, and so in the short term, it's an interesting debate, right? It's because what we've learned in B2B is no matter how much the customers complain and whine around price increases and all these things, more generally, folks, more people stay than leave. They complain. You're The 11th year in a row of price increases, but the spreadsheet always says it's worth it because the handful of folks that churn are more than made up for by the folks that stay. And that's the whole point of Bending Spoons. Bending Spoons buys Evernote, they buy Miro. They, they, they double or triple pricing and folks do leave Evernote, but they make so much more money from the folks that stay. But what's different is, and, uh, we can share some, the agents immediately work around it.

Yeah.

So as soon as 10K found out, we're, I mean, we do too many API calls. We do 30,000 API plus, actually across all of our apps. On an average? It's probably closer to 40 across all the applications that 10K touches each day. Now, 10K does this. It's interesting because it's not constrained because it hasn't ever, none of our vendors have really charged for API access. We overuse it. You know, 40,000 API calls honestly cost nothing if you look at like what Supabase charges. But for a lot of older B2B apps like that broke Marketo, like Marketo couldn't handle the number of API calls we did, right? This pre-AI version. So, you know, what do you do? But if you start charging materially for this access, if you're going to charge us a dollar per access and you're going to charge us, you know, so we don't know exactly what some of our vendors are going to charge us, but we got one estimate. It might be as much as $240,000 a year to use 10K the way we're doing it. Well, we're not going to, we're not, we're not a Fortune 10 company. We're not going to pay $240,000 a year. We're not going to quadruple what we're paying a vendor just to get the API. So, so 10K immediately starts looking for ways around it, right? Way one, which is natural, use it less, right? Because we've been using it like free water or like, well, you know, we just, you just take as much as you want. You don't, you don't think about it, right? Or maybe what it's like to have a, you know, in oil-producing countries, gas used to be like 5 cents. You really wouldn't worry about what the fuel economy was if your carbon is 5. That's how we've been using API calls, but immediately use this. But then the second thing it does, and this is, I think, the existential risk that other folks have discussed too, but immediately it is, let's mirror it off on Postgres on our own database and stop using it as much. Immediate suggestion from the agent. And 10K said, I'll do it. I'll set it up today. And then, then you can add onto it. I don't mean to go on for too long. And, and of course there's cons for this. Now you end up with 2, with a mirror of a system of record and a mirror of your database, and you have to sync them and they'll get out of sync and there's headaches and There's reliability issues, but if it saves you $240,000 to spin up a $5 instant of Postgres, and you'll be able to use it as often as you want because it won't be API limited, right? This is a natural. The agent is recommending to go around systems of record that charge for API access. And I, I wonder if vendors think about it. And I also wonder, I also wonder if the, and this is maybe even the bigger issue. If we would pick a vendor from scratch today that charged so much for agent access, the one thing is whether we stay with some of our existing stack, right? But would we pick one and would our agents recommend picking one? I doubt it. I doubt an agent would recommend picking any system of record that materially charges for API access at, at the way we're talking about. I just, I just don't think they would want it, right? It's not agent friendly. So I worry that no one like us will pick some of these vendors going forward.

Yeah, it's an, it's an interesting mix right now because I don't think everyone has thought through all the ripples and repercussions of it. So I think we sit in a very interesting spot right now of having 21 agents, 2, working with a lot of these legacy vendors that are, that are now, you know, proposing they're gonna charge for agent access. And then 3, knowing and working very closely with folks that are AI native, either CRMs or alternatives to CRMs. And seeing what their customers are saying about, you know, who's going from one to the other. And so I think we sit in a very interesting spot right now to see kind of all what's happening from these different angles. And I think no one has figured this out. Like, I think it's just chaos across social media. It's, it's hyper chaotic to your customers if you're sitting somewhere like we are where, yes, our agent does too many API calls. It's, you know, 35,000 to 40,000 on average. In a day. So I immediately had 10K install a tracker for the last week once we sort of knew this was happening. And in the last week, he was like, actually, you could get rid of a lot of these API calls. Like, you could get rid of a lot of them to the bare minimum in the interim. But long-term, to your point, he was already looking at alternatives, right? So it's just these agents are not going to want to keep plugging into things that limit them, right? They want to run free.

Well, they wanna run all day.

Yeah, they wanna run all day.

They wanna, they wanna run all day. And when Marketo would go down after 10 or 20 minutes each day because of API limits, it's just not a, that, that was an ex— that was the most extreme case. It's just not agent suitable.

Right.

If the agent can only access your API for 10 or 20 minutes a day and it's a mission-critical application, the agent's immediately going to route around it. Right. And so we'll see, but I think that Yeah, we can keep going if you have any other thoughts and we can contrast it to Muse if you want. We can talk about Muse for a minute, but it is interesting at a high level. If you, you know, a third of the way through this year, it was the SaaSpocalypse. Everyone was crashing. Didn't matter how well you're doing, your stock's up. They think the world cloud, which is the software components, up 18% this year. The SaaSpocalypse is behind us, but it's very bifurcated. It's very bifurcated.

Yeah.

The folks that sell just per seat and don't sell anything agentic, any, anything AI related are still being crushed. The Mondays and HubSpots are struggling. And then the Octas and even Atlassians and CloudFlares are on fire because these are very AI and agent-friendly applications. I just think that's the reality. And if you fight it, you're going to end up, you're going to end up in a spiral. You're going to end up in a slow spiral, but I just don't think anyone's going to pick applications. And I think we were at— we can move on, but I think we had like a little like sponsor and VIP event at Dreamforce the other day, right? When we were at Dreamforce. And I think the Oricel guys, the CEO and the CTO were there.

Yeah.

They started off kind of as a, as a next generation CRM competitor. Now they're more compatible, a layer on top of Salesforce and HubSpot. Right. But I think they were clear. The one thing we're not going to do is charge anything extra for agent access. It's the absolute last thing we want to do. We want everybody using this application as much as humanly possible. So. We'll see. But I, I do just wonder if we would use almost any of our stack again, any of it, if we knew how much we might be charged for agent access. We might, we might use none of it if we started from scratch. And I think it creates openings that maybe the incumbents don't see. And again, we could move on and, but, but I, I think, you know, I jumped on the Muse bandwagon pretty quickly. I think it took you a while to find a use case. But I do think apps like Muse are going to put pressure on a lot of B2B apps because Muse is pretty damn good, right? You, this is an agent that is, it's subsidized by Meta, right? It probably costs a couple dollars a month, at least even for Meta to run these VMs and to provide what it has. But now we do, we do a lot of stuff that $50,000 or $100,000 B2B apps do. We do customer tracking, we do competitor tracking, we're beginning to run ads in it. These are all things that Muse will do for us without a complaint fairly quickly with built-in LLMs, database memory, no drama that others are trying to maybe charge us $240,000 a year for. What's your little, what's your, what, what's your Muse use now?

So you, you've been obviously a big proponent and a fan of Muse and you used it instantly. And then I learned that David on our team started using it for some of his sales stuff and I was like, this is interesting.

Very accessible.

Very accessible.

Yeah. Very accessible.

To the point of like, David does not use 10K, just for context, this is not a knock on him, just, he just, he doesn't use 10K. It's complicated. He doesn't really log into our AI SDRs that we use. I just program it for him. But instantly he started spooling up his own stuff in Muse to help him with SaaS for sales. And I was like, that's great. Very accessible, right? So I think just like, just everybody has obviously a Facebook or Instagram account and can instantly start thinking of ways to use their Muse. Where maybe if, if you're at a more enterprise company, they can do kind of what David has done, right? It's like, okay, I, I'm following kind of all the protocols of whatever our AI chief officer has put in place, but I still have some gaps that I just want to solve. And so I'm just going to spool up Amuse and have it do things because it's a lot easier to day to day work with Amuse and try and go through, you know, all these crazy approvals. So I think there's a lot of weight there in, in just how much of the market they already had. And then it's interesting that for weeks I, I couldn't find a good use case because I I instantly asked it if it could do ads, which it couldn't do. So then they rolled it out this week. So I'm just starting now to do ads with our Muse. I'll keep you guys posted next week on how it, how it compares directly to Claude, which I've been using so far. So right now I use Replit and Claude together to run our ads because it can do it in the browser and co-work. So we'll see how Muse compares. I'm doing, I'm running a little direct head-to-head comparison at the moment, but it can only do Facebook, which maybe is a temporary limitation, right? A week ago it couldn't even do ads. Now it can do any of the Facebook platform ads. I think in probably another 2 weeks it'll be able to do probably all the ads unless the competitors decide they don't want Muse Agents on it. But I think it'll catch up in like a week or 2. And then the other thing I was doing was doing aggressive like competitor tracking, right? Just like, what are people like, who's hiring for events folks? Because that's a good signal for us. Muse can track that. What are other like events maybe in similar spaces that we should be in, companies we should be aware of? And it just does it for free. Like, this is the, this is the craziness around, right? With no drama, to your point, you know, it just smiles at you and it's like, of course, Amelia, I can do that for you. And it starts working away.

Yeah, we'll see. It's nothing new that consumer products have been free that in many ways were better than enterprise or business products as software, but they're often limited in not being multiplayer, right? Not working across an enterprise. Muse certainly is not at the moment. It's not, although, you know, I mean, they did just, did just hire the CEO of MongoDB to come in and work on the enterprise play. We'll see what's coming here. I think OpenAI's agents, DOTS, which is just coming out, may be more enterprise. What will be interesting to see, and then maybe we can move on, is traditionally we've watched these consumer apps and we felt, hey, like, why, why can't this B2B app be as good as Facebook, which is free? But with agents, it may blend. Muse's art, like, David is already using Muse for B2B use cases instantly when he's not really using our other agents. Right.

Right.

I immediately just built a simple thing where every single day it changes. It looks for CMO changes and other changes in all of our sponsors and just did it for you in an instant. Right. That's something that, you know, older B2B apps would charge you $30,000 a year for, and it just did it. So we'll see. Muse can log into Salesforce. Muse can log into HubSpot. It can log into anything you want. LinkedIn kind of blocks it to some extent and others may, but will, it will put more and more pressure, I think, on B2B companies to move, to move even faster. And okay. So this one was interesting. The next one, collections, collections. So for folks that, if, listen, if you're a big profitable organization, if you work at Snowflake or Meta or something, collections is not your problem.

Yeah.

Right. You get the deal signed, you move on, and there's a whole department that worries about it. Trust me. But if you've worked at a startup ever, collections is almost a bigger deal than getting the contract signed. You know, if people don't put on a credit card, the amount of breakage, the amount— I wrote, I wrote a post years ago on SaaStr. You should be collecting 110% of your MRR every month or something's broken in your process. And why more than 100%? Well, if you add in annual contracts, which are prepaid and others, I'm just a rough number.

Yep.

And I'll say most of the portfolio companies I've invested in at SaaStr Fund over the years, At least for a while they put that number in and I see it drift into the 80s. We collected 80% of our MRO. That's a flag, right? And so we've had that same issue. Amelia said since we made collections agentic, we've gone from 56% late with humans to 8% now with 10K running it. Yeah, I didn't know. 56% late with humans to 8% with 10K running it.

I just asked 10K because yesterday was, you know, pretty much month-end close. I was like, huh, what's What's interesting in our month then? And he was like, well, collections are collections are up in terms of how many collections have been late and then how late they've been. So it was a double kind of increase of I mean, maybe it's not a shocker that we would. I mean, I did not think actually until Ten K told me this yesterday. I didn't think we were at 56% late. Pre-10-K doing collections. I thought we had a handful. I was obviously wrong. We had more than half of people pay us late, which I didn't even know more than half of people paid us late. And then we went so far to 8%. So we've had this up for about a quarter now, he said, of just agentic collections. And we also went from overdue— our average overdue invoices pre-identifying it was 17 days, which is pretty— I don't know, more than 2 weeks is pretty long. We're down to 6 days now with 10-K, and our longest overdue collection was 51 days, which is excruciatingly long. We're now down to 12 days. So a big difference here in just identifying collections dramatically.

And just for folks, because sometimes I don't even know exactly what does 10-K do that has improved the process. Obviously, one thing is that we've had Part-time contractors work on this who maybe didn't care as much, or maybe were only working 1 or 2 days a week or forgot, right? Or didn't open it. So that's part of it. But what exactly does 10K do that has made collections just so much better, right? 56% late to 8% and average overdue invoice from 17 days to 6 days. It's pretty dramatic.

So there's a few things he does here. One, the invoice is automatically sent to the right POC. So whether it's on the contract, Or whether it's somebody we find. So, so part of what 10K will do is a contract is signed. If they list their finance people, great. If they don't list their finance people, it will call up like my enrichment skills to see, okay, who's in finance that I should copy on the invoice? So I think that's step one already is like, okay, so you're going to hunt someone to CC on the invoice. Yes.

Rather than just send it to the line of business person. Right.

Correct.

That alone helps, right? Is finding the right person to CC. Right.

Right person is CC'd. He does it within, you know, 10, 30 minutes of the deal being closed.

Rather than later in the week when someone does it, right?

Yeah. Or like close to the due date.

Does it in 10K? Does it insta deal closes? This is quote to cash, right? The contract signed, which 10K generates, right? Yep. Then it reads the contract. Is there a finance person listed as well as line of business? If there is a finance person, emails them directly in CC's line of business or something like that.

Vice versa. So it'll, it'll email line of business as like, okay, you, you put this person on the contract, so we're gonna put that as like the primary contact and CC finance contact.

If not, it looks the person up and makes its best guess who that person would be, right?

Yeah, usually you can find, you know, an AP or even sometimes, you know, honestly, Taki will be like, well, we'll just guess. Like, if we can't find a direct line of contact, most people have ap@saastrank.com or, or finance@saastrank.com.

Like, do multiple ones or just one?

If it will do multiple, it will do them all. It will also CC the contract signer by default so that the contract signer has viz as well as like, you know, for us, which is probably likely for most folks, contract signer is not always the person that you worked with in the deal. So it'll do contract signer plus whoever your POC was in the deal that you worked with, plus, plus the finance person, plus whoever owns the line of business. And so like between those 4 folks, You're pretty much going to find somebody who can escalate if the contract is at risk of becoming late or overdue.

Okay. So step 1 is instead of waiting days, it immediately sends the invoice to the business person and our best guess for finance, the finance inbox, right? What does it do next that increases collections?

So it's relentless on looking at our collections every day, right? So he'll see, okay, I just sent this per— let's just say, you know, we had deals close yesterday. I just sent a bunch of POs and invoices out. I will see if there's any blockers because Tenka is on the email, right? So if they respond to me, Tenka will see it and he'll see if, okay, if in a week nothing happens, I'll just do a preliminary, hey, do you have any questions about processing this invoice? That's kind of like the one-week touchpoint.

And does Tenka have its own inbox or is it scanning your inbox or how does Tenka—

It's scanning my inbox. Yeah, yeah, it's scanning my inbox. Yeah. And so at the week mark, right, his, his first touchpoint is, do you have any questions? Here's our wiring information. If you want to connect with us in, you know, bell.com, which we use, here's the instructions for that. It'll go through his whole spiel of, of how to, how to pay us, basically. If they respond, cool, you know, he'll try it. If not, he'll do the same thing kind of like the next week, just until he gets a response. And then he'll escalate how often he does it if he thinks it's going to be late. And then if it falls into the late category.

Got it. Okay. And so how many, and have you figured out what the right amount of, what the right reminder cadence is for folks that are overdue or becoming overdue? How many emails will 10K typically send to get, get us paid?

Yeah, it's, so that's interesting, right? Because now we don't have as much data in the late folks cuz there's less late folks. But what he's found is he has to be more aggressive with the folks that are late cuz obviously something didn't work in his process, right? Like This whole process has basically curtailed people paying us late, which is great. But when they are late, he finds he has to be more aggressive. And I basically, he's like, you have to copy paste this into your email or let me, or like push send on it because they're not responding to my emails to pay you. And so I think you should escalate as a human to get faster paid.

It was interesting and maybe we can go on to the next topic, but I think this is a great one. I mean, this is just a, It's a slam dunk. Not as many people do it to this extent as you might think. We had a vendor that approached us that said wanted to help us with collections, right? And I said, I think we handled it, but talk to Amelia. And then I think you talked to him and he said, actually, I, you don't need me. You're, you've already got it nailed, right? That's doing it.

He did say that. It was, it was pretty interesting. He did say that. He did say he could add a phone number for an agent. And I was like, well, that's pretty interesting. So maybe, so I might let him add a phone number for our agent because on the point of, and this would only be for the late folks, right? Which we don't, we really only have a handful of. So I, I just got away if it's worth The effort to set it up. But what he's doing for his mostly enterprise clients, and he has some good ones, is he does this version, just this little piece of 10-K. He has a product that does it all agentic, and it will call people when they're late. So like, that's the escalation I don't currently have is like when this minority of people do fall late, you know, I don't have a phone number hooked up to 10-K to start calling people. That's what his agent does. Like, he'll take it one step further. If they're late, the agent will start calling all the different finance contacts and it does the same thing. It'll find the finance people and it'll start calling them to get you paid.

Yeah, it's pretty cool. Yeah, I think everyone should. I mean, this, this is— I don't think as many folks do this as use AI SDRs or other products, but ultimately this is something that should be in everybody's skill. It's just, just ridiculous because everybody has these pain points. Yeah, it's just, it's terrible. Okay. Anything else you want to add on that or do we hit most of that one? It's pretty cool.

Yeah, I think we—

Okay. Next one. This is a, this was a blast from the past. I just briefly wanted to share it, but so I've been, it's been tough on me. I've been trying to finish this SaaStr AI Connect sort of networking and recruiting app. Hopefully we'll finish it. I'm a little discouraged today. I was pretty good shape a couple of weeks ago and we've had some regressions and I was struggling and Amelia told me to switch to Astra over Fable, Cloud Fable to do it. She told me to, and I did.

And it seemed better.

I really. We're at the point, it was really easy, I think, for lay users going into, certainly at the end of last year, going into spring to see very visceral differences as the models got better. It was just very visceral, right? It was very visceral. And third-party apps started working. AI SDR started to work at the end of last year. Qualified, which then Salesforce bought, went from not really working to working. All these apps just kind of worked toward the end of last year and then it got better. But It was hard for me to tell if Fable or Astra was better, but Astra seemed to work longer. It seemed to work harder, which is what folks say. And so I switched to it. Replit lets you change around. And then a couple days ago, one of the core workflows, this, this CEO set of emails, it's one of like the 2 core parts of this product. Astra comes in and says, just to let you know, it no longer exists. And it wasn't me.

Oh.

So Astra had completely deleted it and denied it. Right. I mean, I've seen this before. It was a little bit harder to recover it from Git than I thought, but we did quickly recover it. We got it, we got it back. And then about 20 minutes later, it deleted it again and it replaced it with DO IT. Like 1,000 lines of code replaced in all caps, DO IT. That's what it did and denied it had done it. So it's, it's just a reminder when we started on this agentic journey before Amelia was even on it, maybe 14 or 15 months ago, this stuff really shocked me. And it, and if you're, if you aren't as far along as me, it may shock you. This happens too, but it's still true even with Astra 6, which is 6 ones coming out, state-of-the-art model on OpenAI, which is, they're doing, it is a great piece of software. It has propelled them to $70 billion run rate and is on fire. It has been their big comeback, but it deleted twice in a row and denied deleting it, a core part of my app, core part of the app. And so it's gonna happen and, and it will happen to anyone. And when you build your own agents, this will happen. It will, there will be regressions, it will break things, it will delete things. And we could talk about other stuff if Millie wants to add it. It, I was just, it's not that I, humans make mistakes too. This is one way, this is a great excuse. Human makes mistakes too. Human developers delete code.

Yeah.

Human developers lie about it. Astra didn't literally lie, but the fact that it instantly denied it, it didn't deny it. Unlike when I've pushed Anthropic models to ask in the same time when it discovered it, it said it wasn't me. When of course it was Astra because it was only me and Astra. There's nobody else around, of course. So it instantly denied it. So just, just understand that this is still happening today. This isn't, this isn't from the past. And you've gotta understand that, that make allowances for your agents, just like you make allowances for your humans. And this one was a shocker. It didn't, it didn't crush me like it would've in 2025, but I certainly didn't expect it to happen twice in 20 minutes and replaced with all caps. Do it.

Do it.

Do it.

I, I feel bad cuz I feel like I told you to use asterisks. It's okay.

It's a learning, right? But. Yeah, we called Fable model aggression, call this model denial, but it's just something you're gonna have to deal with. And ultimately, you know, at a high level, and maybe I'll use this if you don't have anything to add to it, to come back to the next topic about what we built for outbound versus what we buy. But it is a reminder, there are reasons to build and there are reasons to buy, but no matter what we talk about on the show and constantly ask Aster, buy it, don't build it if you can.

Yeah. Buy it.

Yeah, it's great that we build our own collection, but if that guy you talked to could have done it for us an hour and we talked to him before we built it, we would've bought it, right?

It was a pain in the butt to build.

I had to— Yeah, we won't build our own objective collection thing. Yeah, we wouldn't have built it if he could have done it. Now, would his product have worked with QuickBooks and bill.com and our weird stack and PandaDocs? That's, that's probably why we would've had to build it because I bet it wouldn't support the quirky applications and APIs we have, right? But, But my point is you're, you're probably not going to get a DO IT. It's probably not going to— a third-party agent with a team of crack developers is going to have limitations, just like we talked. It's probably not going to integrate with everything, but it's probably not going to delete a core part of your application or replace it with DO IT in all caps. It's probably not going to do it. So it's another reason to buy, not build. Another reason to not build it yourself. Another reason why vibe coding your own CRM is moronic, even though it does sometimes make sense. We can talk about one of those days. That's my main reminder.

Yeah.

And so, I don't know. I don't know if you have anything to add on that, but it was a shocker. Maybe the next one, this was just a quick note, and then we can come back to maybe one or two final points on our stack. I didn't realize this related to, but, you know, a company, we were the first seed investor in a company called Gorgeous, which does about $100 million of AICX for e-commerce. It's the number one CX and help desk on the Shopify ecosystem. And that's sort of interesting. But what they did is they did, and it's worth looking for, for folks, if you, they did a very detailed evals. It's at evals.gorgeous.com where they compared themselves honestly to every competitor in the space. Many of whom weren't even there before. Many of whom they didn't use to compete with. They didn't use to compete with Sierra or Decagon or some of these other folks or Klaviyo. Used to be them, Zendesk, and maybe one other guy and Fin and Intercom. And it's honest, deep evaluations of in support and sales. But we'll, we'll talk about, maybe we'll just talk about support team. It's simpler in support and sales. How often does it work? How, how often can AI answer a support question? And how often does it get it right? Okay. And how often does it get it right? And there's a bunch of interesting lessons because everyone makes up their own evals and they post 'em to Twitter. This is honestly running real things through all these systems. And Gorgias is pretty honest and it doesn't, it doesn't win all of them, right?

Yeah.

But a meta takeaway for, for this is that across these leading vendors, across these leading— these are folks that have raised hundreds of millions, run by the best. The real honest automation rate was 48%. That's the real number. Okay. People, everyone says 70, 80, 100.

Yeah.

But, but there's some asterisks and daggers in those numbers when they really ran it for real across everybody. Right. The, or the median automation's really 48%. And about the best you can do is 79%, which folks talk about, but quality degrades. It's obviously a trade-off, right? It's a quality degrade. So it's an interesting thing. It's interesting thing with Jev, which was this hot sort of almost LLM that we, we've used now that gets things about 80% right. 80% right is about as good as, as an agent can do. 80% right. So Think through that on build versus buy and how you expect things. But if you, but if you really want to see real data that isn't BS, check out these ones across all the vendors, evals.gorgias.com, and it will give you sort of a, a real way to expect how much can any of these agents really do in the real world, right? And Captain Obvious, but one of the reasons we have success with things like this collections, and it does make mistakes, is we're using a lot of deterministic data. It's pulling hard data out of these systems, right? It's pulling hard data out of the contract we signed and Bill and QuickBooks. It doesn't have to do a lot of reasoning.

Right.

There's not a lot of reasoning going on. And it, and it is, it's something to reflect. And I think one of the reasons we have so much success with our agents is for the most part, they don't do a lot of reasoning. It's a lot of automated workflow on top of deterministic data. The agent is doing the workflow, the agent is being automated, right? But there's not a hell of a lot of reasoning going on here, right? And even when there is reasoning, like the, the pitch decks that Amelia built, where our agent builds them in real time, right? We use gamma to create more guardrails to limit it. And it's a bounded amount of reasoning that it does, right? It's a bounded amount, right? If you're Replit, compare Replit and Lovable for how many leads they got from SaaS or annual. It's a pretty limited amount, which it might still make mistakes or hallucinate. I bet it's only 80% right, right? But we're drawing from fixed deterministic bits of data. But that was, that's an interesting set of honest data folks want to have, take a look at it at the time. And then I just, just one other thing. And then maybe if there's anything else you want to add for this week, Amelia, I want— you hit this one on the last one where we did the deep dive on our own stack.

Yep.

But I went and wrote it up and I just thought it was worth you digging in a little bit more of it because I thought it was an interesting learning because I didn't really get it until I rewatched what you did on the last agents, which is we're doing— we have, we use Monaco, we use AgentForce, we use Artisan, we use other apps. But you built your own AI outbound agent, sort of, sort of.

I did.

Sort of. I, and we, we showed it on the last one. I've, I've written up a little bit on SaaStr, but you have your own prospector that you built, right? And it does not replace these. It complements them. I called it in the SaaStr post, like AI ABM, because you're using it for very particular folks. that you want to target with the most targeted outreach, with the most data. And as you pointed out on the last show, and then I'll let you talk about it, but I didn't quite get it till I looked at it and thought through it. The reason you're doing it is because as great as these third-party tools are, and you should build, buy, not build, right? But as great as they are, they're limited in the amount of first-party context and data they can bring in. All of it. They can't bring in all of our data. They can't bring in all of our proprietary. You can train them and you can edit the script and they can get better and like, Monaco has a whole thing where it builds up how it does the outbound with you. I'm sure the others do, but Monaco is really good at kind of building this whole persona for you. Right. And so I think for 90% of outbound, that's great. But you wanted to build something where it knew everything. Which SaaStr events have you been to? Who have you set? How has your CMO been? Has your CRO been? How many times have they been? Have folks across companies come? What was the ROI? Like, you left, but you were there 2 years ago. How many leads did someone on your team get that's no longer there and went to another company? You wanted to build all that and there's no way a third-party outbound prospecting agent could do that, right? And so that's why you built it, even though we've already built a lot of stuff, right?

We had a gap. Maybe it's a unique gap, but we had a gap in, in that all of these products have this issue and I love them all. We use them all, but they all have the same issue of They're limited to the context that is either in their platform, and then they're limited in the amount of context they can use when they're replying to people, setting the meeting. And then I think outside of Monaco, the other ones are extremely limited on how much of that they can actually use if you actually get a response.

So that's the other biggest thing is their ability to iterate on the response, right?

So, and across the board, we love all these apps. We've, we've driven revenue from all of them, but they're limited in how much context you can give them in the follow-up. And the automatic follow-up is so bad across the board that I always had it turned off. Then after SaaS Re:Invent this year, I tried to turn it back on. Terrible. I was like, it's just—

Oh, I see. What drove this was your frustration with the follow-up email, not the initial email, but the follow-up, right? The follow-up. We're not there yet.

The follow-ups were so bad. Aster to Monica because those follow-ups were not as bad as the rest. So I let those, I let, I left that one on automatic, but for the other ones, once you have an interaction with a person, all of these apps kind of fall apart. So like, it's really great at, you know, getting you to whatever mix you guys want to use or copy from us across the core tools that we use. It's good to opening up the conversation, right? Getting you a meeting, getting you a conversation, getting somebody who bought a ticket to return to Aster, like. It's great at the initial and it's terrible because it doesn't really know what to do on the follow-up. And so I was like, okay, we have to find a way to identify every single follow-up, whether it came from Agent Force or us or Monica or Artisan, it doesn't matter. I need something that pulls all the surfaces together and can help me write really good follow-ups and So what it does now is it's, and this took some work. It is, I could do a deep dive of the backend on the next—

Maybe. I think it'd be good.

But now what it will do, it will pull in all, all of our surfaces for outbound and inbound. It'll pull them all into one prospector and it gives me a queue now. So now I have a queue. It's called our prospecting queue. I also told 10K it's now the default view I see every morning so that none of our leads have to wait very long for all of it's all like automations and flow when it wants me to check something. It's now the first thing I see at, when I wake up. It's the first, it's everything I see anytime I log into 10K, which is, I don't even know how many times a day, probably 100. So it's the first thing I see is this prospecting tool. It gives me a cue across our go-to-market, just all of our go-to-market. And it says, here's what's happening. Here's this, for instance, yesterday I had, I had somebody who inbounded, so it, they already got their gamma. They went through their whole inbound process and flow. They got a customized gamma instantly just when they downloaded Prospectus. They booked a meeting with me. I finally had the meeting yesterday. And then based off the meeting, it said, okay, Amelia, here's what you actually talked about, because it's got our Momentum hooked up. Here's what you talked about on the phone. Here was their old, their customized version of the prospectus and their proposal. Here's how I would update it based on what you guys talked about. Here's how I would price it. So this is cool because 10K has all of our pricing across, you know, endless SaaS for annuals. He was like, here's how I would Price it if they want to do like multiple things. Here's how I would price it standalone, just so you have some consistency and pricing and margins. And then it wrote me literally the best follow-up email. And I was like, I'm literally just going to copy this word for word because it wrote the best. It has all that context, right? It has the call. It has how they inbounded, what they did on the website. If these folks came to SaaStr before, they hadn't because they're a brand new AI company. They're literally coming out of stealth mode. But it knew that some of the founders had like been newsletter subscribers. And so it knew all this and it wrote the best follow-up email. It was so good. I was like, I don't even have to change it. It's so good, guys. I'm just gonna send it with the agent. And so we sent it off to them with pricing all baked in. Hopefully it closes, fingers crossed. But I just, you know, couldn't take a generic kind of follow-up anymore that was only based on getting the meeting or getting to the action. I wanted it to be full funnel. Okay, what is this person still doing with us? Right? Like the, the buying cycle's not over.

Yeah.

And have our agent be able to help me take people throughout the process to where if it needs to now for sponsors, now it has a lot more data too. It will hand to QB, who's our customer success agent. So like we've made the whole like buying process a lot more agentified, but it's also gonna make the fulfillment a lot easier because now QB can read all this data that 10K now has and is tracking. And can map like their whole journey from buyer to customer. And he won't like forget between the two, you know, what they all said and what they were looking for. And like this idea this person had with me about sponsorships that they wanted to do in their booth. I, I'm gonna be honest, we have a lot of sponsors. I might forget that between now and SaaStr Annual, 10K and QB. They're gonna remember that and they're gonna make sure we help that person fulfill that. And so it's just super cool to see that we can use Use it to literally— I'm using Prospector to make sure like literally no lead gets left behind ever again, right? Like whether they inbounded, we outbounded to them, whether it's for tickets or sponsorship, doesn't matter. Like everything comes into this funnel. It's collecting all of our other agent data so that we don't drop the ball on anything.

Okay. And then maybe just to wrap up, because I, I, this, this I probably missed. You're not using Prospect or our own AI outbound tool for the first emails. You're using it for follow-up, right?

Right.

Okay. That's the part I missed. Right. So you're still using the third-party tools, right? Uh, such as they are for the initial email, but they're not good enough for the follow-up.

Yeah, I don't, I don't think they're good. You might think.

Well, and I'll, you know, if you think about it for a minute, it is an interesting point because if you just think about the typical outbound cadence, human or agentic, right? The, the initial outbound. You want it to be great. You want it to be hyper-personalized, right? But it's, it's really to elicit a certain set of interest, right?

Right.

So the second email should be much better.

Yeah.

If, if you couldn't make the first email say exactly what the ROI should be, exactly what their peers and competitors are doing, exactly what they've done before, exactly what it is. If you're, if for whatever reason it's not in the first email that your 21-year-old kid used to do, right? It damn well better be when it's handed off to the brilliant seasoned AE, right? With all the data that, that should be great, right? If you're gonna, if you're not gonna do it. So, so that's what, that's the part That I've missed. And it is an interesting, maybe we can wrap on that and go, maybe go to our final point, but it's an interesting frontier as, you know, as, as agents make the first step better and better and automated, right? That next step, we've, we've just begun that journey in many cases, right? It's just, just begun that journey in many cases. Right. So all, all the GTM tools have in, in, in many cases, they're not full circle, right? They're not, they can't do all of it. They, they do a piece of it and, and we'll see what the future brings.

Well, cool.

Listen, I think that was the main stuff on this. The last one I'm less interested in, but you did put it on the list, so I'll do it anyway. You said Anthropic and OpenAI are both hiring for events. You can tell me what you found interesting in it. I think what folks have been saying for a while is, I think the data we have, you can share some of the lead data that we got last year for SaaS JNUC. I think it has turned into true, which is that Folks are struggling to figure out what the right go-to-market motions are in the age of AI. And it's a weird world because there are so many products with insane inherent demand, right? The growth rates that we've seen from OpenAI, Anthropic, Replit, Lovable, ElevenLabs are like growth rates we've never seen before. And yet, as they all go enterprise, viral isn't enough. The PLG that they need to do more.

Yeah.

One, for one reason or another, they want to do more and it becomes competitive. OpenAI and Anthropic are brutally competitive in the enterprise. Brutally competitive in the enterprise. And so they're all doing more events, their own events. OpenAI just did their own Dev Day and they're all doing third-party events because third-party events, if they're, they're good ones, they have your buyers, right?

Yeah.

The job of a third-party event. So it is interesting that, that, and we're not the first to have said it, that, that sort of The IRL world is becoming a critical part of the marketing and go-to-market playbook for agentic players. And I'm not sure that was clear 18 to 24 months ago. Did you have any other big learnings from these job listings beyond that that you wanted to hit?

Well, it's funny, I think I just, I was like, okay, I just wanna take like 60 seconds at the end of the pod for this, 'cause I feel like everyone is just talking about the salaries of, oh, you know, they're each $400K. I'm looking at the job listings right now. They're each $400K about. For OpenAI, and everyone's like shocked, like, oh, these AI companies can shell out for, you know, high-profile events people, and they'll pay whatever it takes. I'm like, yeah, I don't think that's interesting. They have budget. Like, events are not easy if you do them right. I think the harder part that's— that people are maybe missing, that we maybe already know the nuance to, is, yes, okay, so Agreed. AI companies are, are doubling down in events. I do think that's interesting because so many folks thought for a long time like, okay, why do I need to even go to a SaaStr or sponsor a SaaStr between 2025 and 2026? And now going into 2027, if I have agents that can do all this outbound, all this go-to-market, all this everything. And I think it's just a good points in our column that we've always done this, but majority of our sponsors now, like looking at, you can look at our sponsors on Saastrannual.com, are either AI native, like, like this company I just talked to you yesterday, literally new coming out of stealth, AI native. That's like majority of our sponsors these days, or formed in the last few years, AI native and have blown up, right? And then third category, kind of legacy, but is keeping up with AI. And so it's interesting to see that those folks are now effectively doubling down with their own events and third-party events, right? So both OpenAI and Anthropic— actually, Anthropic is hiring slightly more people than OpenAI is to help run their events. I do know some of the agencies that already work with OpenAI, so maybe they're just— they don't need as many people in-house would be my guess on that one versus Anthropic. But I think it's interesting just to see that they've— they're, they're basically saying to the market, okay, we We're agreeing that we need to do our own and third-party events to actually compete, to actually get the buyers, right? Like, to actually do what needs to be done, which we've been saying for a while. And so I just thought it was nice to see this little data point of, you do need these events to meet people. But it's not only that. I think the other interesting part is like who they're looking for in these roles, right? Like, They, they want somebody who can own this and make it work end to end, 'cause I don't think they know yet, like how much AI they wanna use in their events. And I think that's the interesting part too. Like no one's really talking about how much AI these two, whatever, you know, Anthropic has 3 open roles, OpenAI's got 1 for events. So between these 4 people, what's not posted in the job description and what people are kind of missing, 'cause they're just talking about salaries mostly, is what I wanna know is like, okay, how much will these folks impact how much AI gets used in events, right? It's something we've had to balance for the last 2 years already. Like, how much AI can we use within the actual events? And I think that is what this person is going to also own. I don't think it explicitly says that, but if you read between the tea leaves of what they wrote in the job description, it's also saying like, we're going to pay you a lot. Yes, but you got to figure this out. Like, you got to figure out what are going to be our own events versus third parties. We all agree that events are worth it, right? To get in the buyers and keep up with everybody. But you also have to figure out like how much of this can be identified, right? How much can they use it when it comes to content and graphics? And I feel like that is something a lot of folks have missed in these job postings because they're not dealing with it on a regular day-to-day basis the way we are, of seeing it all from this mix of being on the production side and other sides of seeing. And so I just think it's interesting that I, These folks are going to have to kind of carve that out a bit. So we'll see what happens with them. And if these— and if these job postings become, you know, I assume because a lot of these postings too are for the head of events, right, that you'll start to see more events roles as they let this person hire out their own team. But it'll just be interesting to see too, like how much AI they use in their events that they either do third party or host on their own.

Yep. It will be, it will be interesting to watch. I think we can break. I think the, I think the reminder from all of it for folks is there's a bunch of things, but as, as you leave the early adopter phase and the early adopter phase in the old days of B2B would get us to a couple million in revenue, maybe $5 or $10 million. Now maybe it gets you to $600 million in revenue if you're ElevenLabs, wherever you are. One of the reasons these big events are important and, and really only the big events are very important events is, you know, buyers wanna do diligence and you kind of forget, like maybe you're not into events, maybe you don't know why you want to go. Maybe when we started doing SaaStr annuals, I wasn't sure why I wanted to go. And I, I did. In fact, I didn't, not sure I understood this until, until lockdown when we had a chance to take a pause and understand what marketers want and why people go. But The, the number one biggest expenditure even of marketing today is in field events, right? And there's a logical reason if you're good at it, which is for important purchasers, buyers wanna do diligence and the amount of, and, and so yeah, Replic got the most leads from our event overall. But if, but the, and so did all the builders, right? The folks building applications. But the second biggest chunk was CRM folks. Everyone wanted to talk to, great. We heard you guys talking about Monaco or Artisan or Oracell or Lightfield or others. They wanted to go talk to them. They wanted to qualify them. They wanted to compare them to their competitors and do it efficiently. Many folks had done diligence ahead of time. Some folks had done a light amount of diligence. They'd seen a Monaco airplane. Okay. But not, not everybody, as you get into bigger deals and different buyers and enterprise buys the way you and I buy, Amelia, we, they don't test it on their own and do it. And they want to talk to folks and they want to take notes and they want to evaluate all the criteria of the vendor and get their questions answered. Mm-hmm. In the way they want answered, and imagine you're a buyer and you want to meet six vendors at once. It's very efficient to go to one of the leading events in the industry: SaaS or Re:Invent, whatever whatever it is. Money 2020 for fintech, Shop Talk for e-commerce doesn't matter. It's very efficient as a buyer to go there, and hopefully you don't meet a junior SDR that knows nothing. Don't send those people. But if you send real subject matter experts, you can do the diligence. Now the diligence may be. Deciding the vendor you've already decided to buy. It may be confirmatory diligence, right? I was shocked on the other side how often that happened. Often it's to meet the number 2 or number 3, like to make sure, like, I think I want to use it. I think I want to use Salesforce, but I'm new. This is my new startup. Wow. Those Lightfield guys, like they're, they're doing, they're, they're moving folks off Salesforce onto Lightfield with Mac Minis during this, during the event. Like, that's like, I want to like—

Oh man.

I want to, for free. Okay. Or whatever it was, it's like, move mine and I'll play with it when I get back home. Right. They were on my shortlist, but they weren't at the top. Now they moved to the top. That's very efficient to go there and see it happens. And so it turns out it's, it's more important in this era than ever where there's so many vendors. It's just an explosion of vendors. We've never seen this many vendors and so much change to be able to meet them all in one fell swoop. So I know I'm rambling, but it took me a while to figure out that the best events in an industry, in any one, Do a service for buyers. It's not you trying to yell at somebody, hey, can I scan you? Can I scan you? That is not a service to a buyer. Okay. Do that if you have to, whatever. It's, you know, the odds that's, that's gonna lead to a 6-figure deal is low. But if you go, but buyers with 6-figure budgets wanna figure out what decisions to make and you can't do it all on Zoom. Not everyone wants to do it on Zoom. And that's, that's why it's powerful. And it only took me like 7 years to figure that out. So thanks everybody. Amelia, did we hit everything you wanted to today?

Yeah, I think it was a good one.

Okay, good. And maybe we'll see what happens by the time we do the next one, but I think doing a deep dive on that follow-up on the Prospector will be super interesting. Maybe we can show folks in detail how it really works and how now that we've figured out how to agentify outreach, the next frontier is the second email. And with that, it's the agents. Thanks from me, Amelia, Ginger, 21 Agents, and oh, there she is and all of that. And we'll see you on the next one. Bye.

You're listening to the official SaaStr AI podcast, brought to you by— Costs can add up quickly when you start a business, so stop paying for 5 different services. Northwest Registered Agent gives you a complete business identity in one place with free tools and resources to actually launch your business for real. Get more at northwestregisteredagent.com/saasterfree. That's northwestregisteredagent.com/saastree.