Welcome to the official SaaStr Podcast, where you can hear some of the best SaaStr speakers. This is where the cloud meets up today on the SaaStr Podcast.
The reason we pushed the envelope and I have 20 agents running now was because, and listen, criticize me, I was just exasperated with turnover. I was exasperated with overpaying, not helping, not working. And I just couldn't do it one more time in my career. Now, if I was you and I had a large team and I had 200 million I wanted to spend, maybe I would come to a different conclusion. So, I'm pushing the envelope on how AI can replace not the best folks, not the best sales teams, not your team, but 99% of the world that can't hire your team.
And it's better. This should scare people that don't want to work harder and smarter than a year ago. Our AI agents are better than a midpack AE or SDR or BDR. Better than a midpack. Not maybe not midpack owner, but I've been doing this for a while. Better than the midpack people I've worked over my career. Not better than the best.
Hey SaaStr, imagine having agents for every support task. One that triages tickets, another that catches duplicates, one that spots churn. That'd be pretty amazing, right? HappyFox just made it real with Autopilot. These pre-built AI agents deploy in about 60 seconds and run for as low as two cents per successful action. All of it sits inside the HappyFox omni-channel AI-first support stack: chatbot, co-pilot, and Autopilot working as one. Check them out at happyfox.com/saastr.
Hey everybody, SaaStr Annual will be back May 2026. The world's largest SaaS and AI gathering for executives. Just this last May, we hosted 10,000 attendees with 68 VP-level and above attendees. 36% CEOs and founders and 25% were AI-first professionals. It's the very best of S-tier attendees and decision makers that come to SaaS Annual and AI Summit each and every year. But here's the reality, folks. The longer you wait, the higher ticket prices get. They're cheap now. They're cheap. So just get them early. Lock in your spot today. Use my code Jason100 for exclusive savings. Get your tickets at podcast.saastrannual.com or just use code Jason100 when you check out. See you there. SaaStr Annual and AI Summit 2026. It will rock.
Today's guest is the one and only Mr. Jason Lemkin, and Jason is one of the OGs of B2B SaaS. He runs one of the most important and prolific media companies in SaaS, which is SaaStr. That's where I learned all about tech when I first broke into this space over 10 years ago. He's got the very best podcast in the market, which is Thursdays with Harry Stebbings on 20VC and Rory O'Driscoll, which everybody should listen to every week. It is a must-listen for me, including all the SaaS content on their own feed. And Jason has been an investor and owner since the very beginning. He led the seed round when Adam was probably like 17 years old or something crazy like that, and Jason has had a pretty prolific hit rate, and Jason's been incredibly helpful for me personally along my journey to owner, and so I'm really excited for this one, and we're going to talk about the future of go-to-market, which you've been talking a lot about. So welcome and thanks for joining us.
So, one of the questions is like recursive learning and like when you have all of these tools, this is one of the questions I'm grappling with today. So, we've got one mind and we've got like an SMS follow-up agent that we're building. We've got X and Y. How are you thinking about centralizing the context engineering, the learning loop? Because one of the questions we're asking is like, do we build one mega model of our own to power all of the customer journey from AI BDR all the way through to AI support, or do you have a mix of all these tools, but how do they share insight from one another? Like, have you thought about that?
It's a, if you're as far as we are or further or where you are, it's one of the great questions you'll start to ask. Okay. And I want to give you my answer before we go. Just for folks that are going to watch or listen, bear in mind if you're not as far as we are, don't ask this question. Get a couple cool tools into production that work. Get it going for three or six months. Then you'll realize once you get three or four agents working, not just one, then you'll have data conflict that they'll be answering questions inconsistently. You'll have trouble sharing data. And after three, four, if you have three or four agents in production for real that are working, not complaining they don't work, if they work and you've done it for three or four months, you'll start thinking about meta agents. How could I have one agent manage all these conflicts and issues? What we're doing today is not that clever. We are using, for example, for outbound, we use Agentforce, Artisan, and Qualified. Okay, we use three and we may add more. We're just basically taking our Salesforce database and giving each of them a portion of it. It's not that clever.
And that requires Amelia, who went from our head of CMO and everything to chief officer. Now she's got to manage three different segments of our base and use three different tools. It's fine, right? Because it's better than humans. But they don't, the only way they communicate with each other is by pushing data back into Salesforce.
Right. And that's okay for today, right? And in fact, there's been a little bit of renaissance, as you know, with Salesforce at it. This is both a negative for Salesforce in that the agents in some ways are more powerful than the CRM, but it's also a renaissance because Salesforce becomes the hub for all these agents.
Right. So Salesforce is more important to us. Right.
But it's a big question that I have a lot of answers going into 2026. This meta agent versus the human having to be the boss of all the agents. I think, I don't think we're there yet. At least we haven't figured it out yet. Maybe you'll figure it out ahead of us. And if you're asking the question like an owner, maybe we should build our own orchestration layer that is automated with AI. I mean, listen, more power to you and love, and you know, you guys have raised nine figures of capital and have an S-tier team. I don't think that is the right answer for 99.8% of the world. It might work for owner and I'd love to learn from you. It might work for us, but most folks should not think about, most folks should realize you need a nerdy human managing your agents. Someone that's a GTM nerd in a good way that loves this stuff, that's quantitative, that likes to get in the weeds, that likes to interact with agents, that likes training, that loves data, someone that just loves data, you need that at the top of your stack and your agents will have conflict.
Right. And we haven't figured it out. But like you said, we all quickly see this issue, but you've got to be in production with several conflicting agents to really feel this issue. So, you know, don't put the cart before the horse.
Okay, that's really good advice. And so just going back a step for those who haven't been following us along with the podcast stuff, LinkedIn, Twitter, like just give them a perspective of how drastically your business has performed over the last 12 months and what you've built.
Yeah. As strong as the SaaStr brand is, okay, this is an enduring brand, okay, helped hundreds of thousands of folks, half the public B2B companies have founders love SaaS, okay, it's still hard to recruit. It's still hard to recruit. Over the years I've desperately tried to recruit GTM folks. And first I tried folks that were refugees, right? They got fired from owner, but I still knew Kyle and so I hired a bunch of like folks that were almost like Kyle, but got topped or fired. And I'm like, 'Well, come hang out at SaaStr for a while. Help me sell.' And that was fun, but they either thought it was beneath them or had other issues and they quit. Okay. Then I tried to hire some kids, right? But they didn't want to learn what we do or sell sponsorships. They just wanted to basically sell Clay, something that was easy, right?
So we never cracked the code on our sales team, at least the way I want. I never had the stable, high-performing sales team. I had pockets of excellence, but it was never stable. So we go into May SaaStr Annual 2025 and we also pay people really well, dude, really, really well. And we had two folks on our sales team making very high six-figure salaries that just quit going into disaster. They just quit. They just quit for no reasons. No notice, no reasons. Just ghosted. And listen, maybe it was right of them. It sounds critical. I'm no longer critical. And Emil and I turned to each other. We're like, we're going all in on agents. We're done with this. I am done paying an SDR $150,000 a year, an AE $500,000 a year for basically inbound spoon-fed leads and renewals and quitting on me. Maybe you can be critical of me. Okay, fair enough, right? But I'm not like the worst boss. Like I'm pretty loyal. You've seen me. I'm pretty grateful. I pay people well. I don't tolerate laziness or other things, but do your job with me. I'll stick with you for 20 years. And at this point in my career, I couldn't take it anymore. So I'm like, okay, Emil and I said, listen, anything we can do in AI in May, and May is a long time ago now in AI, like we're going to do it. We are going to do it, right?
And the first learning was, so we just went around all the sponsors at SaaStr this year and we just said, 'Who wants to help us?' And Artisan was there and we didn't know what the right vendor was, okay? And you and I talked at SaaStr about what we were doing and we were still on the journey, but Artisan said to their credit, they're like, 'We'll help, we'll do it all for you. We'll give you the right forward-deployed engineer and we'll do it for you.' So, and this is a meta-learning, we picked Artisan as our first agent. Not because they were the best. They're very good and we can talk about it, but because there's a two-by-two of quality of vendor, but quality of FDE, quality of folks that will help you deploy it.
Okay. Another vendor, you know, is nameless. We talked to their head of sales, argued with us. He was like, 'Well, you've got to pay us $100,000 upfront. We'll see if you're worth it and then I'm not going to do anything until you agree to pay me 100k upfront.' And then I talked to the CEO later and the CEO asked me why did we lose the deal? I'm like, you didn't lose the deal, but Artisan came in and did all the work for us, as did Salesforce with Agentforce, as did Qualified and others. So when you go to pick a vendor, pick someone that is the right partner with you to train and deploy your agent. We are not smart enough to do this. This is not SaaS of 2023 where someone buys a product and hopefully later in the year they get it going. Agents are too much work to train. Every agent that you're going to use in GTM, including Clay, including the rest, requires weeks of training before you can go live. There's a, we can talk about why it requires. So, if you're not getting the help that you need, don't do it. And ask to speak to your forward-deployed engineer or your solution architect or whatever it is. And if you can't talk to that person, just find another vendor. I would rather have a worse vendor and know how they're going to deploy my agent than the world's fanciest brand where I don't know who's going to do it. Don't sign the contract until you talk to your forward-deployed engineer or sales agent. Right? So, I mean, we're going off course, but then we just went all in and I'm like, we're not going to have SDRs we pay, you know, six figures, four quid on us anymore. We're just done. And then we said, let's go read their emails. And like some of their emails were just terrible. And then the worst one was like we had a senior SDR that we paid a lot of money for, made up to $200,000. And basically what this SDR would do is I would go on LinkedIn and I would see, oh, Kyle left from owner to go to Qualified, right? And Qualified's in our ICP. So I would say go follow up with Kyle. And the SDR would be like, I'll add it to my list and get to it when I can. And half the time the SDR wouldn't even follow up. Now the agent just does, the agent doesn't argue. The agent just follows up with Kyle. So anyhow, the reason we pushed the envelope and I have 20 agents running now was because, and listen, criticize me, I was just exasperated with turnover. I was exasperated with overpaying, not helping, not working, and I just couldn't do it one more time in my career. Now, if I was you and I had a large team and I had 200 million I wanted to spend, maybe I would come to a different conclusion. So, I'm pushing the envelope on how AI can replace not the best folks, not the best sales teams, not your team, but 99% of the world that can't hire your team.
And it's better. And this should scare people. This should scare people that don't want to work harder and smarter than a year ago. Our AI agents are better than a midpack AE or SDR or BDR. Better than a midpack. Not maybe not midpack owner, but I've been doing this for a while. Better than the midpack people I've worked over in my career. Not better than the best. And so we just don't need them. We're good. And so these jobs, these midpack jobs, they're just in terminal decline in GTM. And you should be aware of it. They're in terminal decline. We won't need the mid. Now there will be pockets, okay? And the hottest AI companies in the world can't find enough people to hire today. We could talk about that. And so they may absorb the mediocre, ironically, right? Because they have to hire so many humans, but step up or maybe find a new role.
That's sort of the Jevons paradox of sales, which I think people are stumped on. Like if reps are so much more productive than ever, then are the unit economics of an incremental rep so good that you're just going to actually hire more of them? Like where we are, we have like nine high-impact production use cases of AI. AR booked per dollar out on a per-AE basis is 3x any team I've ever managed before, and I've managed some pretty good teams, but the argument is not, hey, let's hire less reps. I'm hiring like 25 people in sales over like a 75-day period. But how do you think that applies more generally to the market? Like does a rep being worth so much more mean we hire more reps, or do you think this like mid-to-lower pack, these jobs just go away?
Well, let's step back for a minute for how I know it's different because it's a different company, but if you had to try to normalize it, how much more productive on a dollar basis, on a quota basis, is your AI-infused team today versus prior teams? Like how much more productive literally are they?
Okay. I might challenge you on it, but it's whatever it is, it's a lot. Okay.
So, it's not like-for-like because we do month-to-month contracts and it's SMB.
But even that, if it were truly 3x, if it were truly 3x, I'm spitballing. You could probably get away with a 30 or 40% smaller team than before and still hit your growth numbers. Not a third, right? But 30 or 40%. So, there's a little bit off in your math, right? Because if you look at a lot of the AI leaders that are hyperscaling, they all have sales teams. But in fact, their leadership is the same people we know, right? I mean, the CEO of Slack that was at Salesforce for 14 years just took over as CRO of OpenAI. Okay. The CBO of Vercel was the CO at Stripe. These are not like brand new people. These are the classic best B2B people we know are running AI companies, but for a variety of reasons, they are running them leaner. Some of it is because so much of the revenue is PLG, right? You know, if the first 300 million Cursor was all inbound, same thing is true with Replit. You just don't need as big a team, right? Because you don't, so there's some reasons. Some of it's cultural, like if you have that much self-serve, do you want the headaches of a human-driven sales team, right? And then some of it is just assigning massive quotas, two, three, four, $5 million quotas. The quotas, I mean, the quotas at OpenAI for the enterprise team are pretty damn high.
Yeah, they're really high. And they don't have enough people. Like they don't have enough people, but they're really high. So, I think there's a lot of confusing factors. If you're growing at OpenAI's rates, or if you're a Salesforce and you're repositioning your whole company for AI, you might hire a massive amount of salespeople. But it doesn't mean that the salespeople don't have to be much more productive. So, I don't think it leads to, it will only lead to more net jobs because AI is such a huge amount of our growth in our economy today.
Right? If OpenAI, I don't know how many sales folks are at OpenAI or all of them, right? All and Anthropic, but let's say they have 10,000 reps on their sales teams or could at the end of next year, let's make up a number, that might be too high, right? That's plus 10,000. But how many are we losing at the public companies growing 8% or 2%? Probably more than 10,000. So I don't think it's positive for, if the new folks have bigger quotas and are more efficient and the old ones are being relentlessly brutal on efficiency, I don't know that's going to lead to massive job improvements. I do think that what will happen, and I wrote about this, the $250,000 SDR. I do think that the elite folks, the truly elite folks, not the ones that think they're elite on LinkedIn, the ones that really are five times, 10 times more productive for real, they should be paid compensation that is two to three times higher than it used to be. And this is already what's happened in engineers in AI, right? The average, you know, the average employee at Anthropic makes $400,000 and the average engineer makes 800K and plenty of engineers at OpenAI are making 20 million, but it's not charity. Like the productivity that they're expected to create is like nothing we've ever seen.
Yeah. The leverage is wild.
Yeah. Super high leverage. So, traditionally in sales, there's been no leverage. Every year, you've got to add more. In fact, there's been anti-leverage because it gets harder and harder to get that incremental dollar, right? For folks that crack the code here for real, which is not going to be most folks, but folks that crack the code, you may well find your way into a job that pays two to three times more than before. If you can have someone managing instead of 20 SDRs, right, if you can have one human managing 20 agents, no problem paying that person two or three times more than a traditional SDR director of outbound. Like no issue, right? But it's not charity. We're expecting 10x the productivity.
But that's probably not a BDR turned agent manager. That's probably like a growth PM or a data science turned GTM.
Probably. I think in the short term, like there are no magic hires out there. We're going to have to look at our own teams and it might be someone, it might be that nerdy SDR that has a math degree that still joined owner because it was cool and is like pretty good, right? But loves all the data, right? And so Kyle says, I'm going to promote her. And then he promotes her again and she's running the whole thing, right? I think we're going to find just as many of these people inside. And if you put a job description on LinkedIn, I want a senior GTM engineer for our sales team, I think you're going to be unhappy who you get.
Yeah, we were trying to hire this GTM AI lead and it was really difficult to fill that funnel with people who had enough business acumen and sales savvy. We ended up hiring a guy who was at Atomic in the venture studio and like had started a couple companies and so had the business savvy but just loves AI and loves go-to-market.
You need someone, they got to love it.
Yeah. And he's ripping through projects and shipping stuff. And we built our cold outbound email program. He built the infrastructure in like three weeks and like everybody else told us that was like a multi-month project and it was just like prompt layer and a bunch of models and he was writing some custom code and then we were sending 1,800 cold outbound emails every single day from all of these domains and we had help. We had like Eric Nowalski doing all the domain infrastructure and helping us think through.
Okay, that's a lot, dude. Like it is a good story, but most people can't do that.
I just don't know if an exec that worked at Atomico that founded four companies wants to go to most startups.
Okay. So, again, going back to the beginning, I do love your stories. I want to hear them. But I think I worry that people, like for most folks, I think they're doing the opposite. They're putting up a job notice for an AI executive to manage this stuff. I actually think in two years it might work because we'll have veterans. We don't have veterans today. So when spaces change rapidly, you've got to find the person on your team that's the AI nerd. Some folks can hire the person, but I just don't think most folks, and we talk to like a lot of CMOs. Amelia talks to a lot of folks, comes to advice. Okay. They do not have the capability to do what you just described. In fact, even worse, Amelia just had a discussion with the CMO at a leading $10 billion company that is AI-first, right? And they said, 'Listen, we think we want to do an AI SDR tool. I'm going to buy it and just hand it to the SDRs to figure it out.' No training, no anything. Just had budget. So, I'm going to spend $400,000 for this tool and I'm just going to give it to each SDR like we used to do with Outreach or Salesloft or Mixmax in the day. Here, go run your thing. And you and I will laugh, but this is a very sophisticated team at a $10 billion-plus public company, just because that old paradigm doesn't work today.
Buy Salesloft, let the reps use it on their own, right?
Yeah. Maybe you can centralize. So most like it's just, you got to find that nerd, that GTM nerd that loves it, that wants to run it for you internally and in fact is chomping at the bit to do so.
Yeah, chomping at the bit. Right. That's good advice.
And so one of the questions I get which is related here is like, okay, my CEO says we want to do AI and I want to pursue being an AI-native company, but like my CFO isn't really giving me the headcount I need. Like we've got, you know, he's giving me like one extra RevOps headcount. They're asking that team to figure it out. I don't really have a budget for like a bunch of speculative tools and I'm really surprised by the number of friends of mine working at good companies who are in this position who are like, I just don't have the budget to pursue it. I'm like, what are you talking about?
It's the wrong way to think about it.
So yes, this is a real story. Here's what you have to do. This will solve all these problems.
Pick one tool that you believe in, that looks like it is the easiest to deploy, that solves a medium or higher ranking problem. Okay? And for many folks it is this AI SDR, but it doesn't matter. Pick one problem. Could be RevOps, right? It could be, pick one tool you believe in based on a demo and some research. Okay? And here's the second, just one that is easiest to deploy, that you believe in, and that it solves a real problem in your org, and then deploy it yourself. The first one is not about headcount. It is not about, but it is only about the budget for this one tool. You only need 50k, 60k or 100k, which is not nothing, but find the budget. If you don't do it yourself, it's like hiring a CRO that's never sold, but it's not quite the same, but it's the same fallacy. 99% of folks we talk to, and we almost are like a free consulting shop. So many folks come to us at SaaS now for help doing this because of all of our data. Do it yourself, dude. Mr. CRO, Mrs. CMO. Then they're flabbergasted. They haven't deployed a tool themselves in a decade or ever, right? And you have to oversee it. Pick the vendor. Do the training. Do you even know what training is? Do you even know what a forward-deployed engineer is? Do you know what ingestion means? Do you know how these tools are trained? Do you know what it takes? And it's only 30 days of work, guys. So it's going to take you 20 or 30 hours of those 30 days. You're going to have to make the time and then my God, you'll be ahead of 98% of the world. You've actually qualified, chosen, demoed, trained, ingested, and deployed an AI agent at scale. That is better than 99% of the folks talking about it on Twitter. They haven't even done it.
You do one and then what you do is you walk back to your CFO and you do what we do. Hey, we just sent 70,000 automated emails. They're better than humans and it generated 15% of the revenue for SaaStr London. 15% of the revenue.
Can I have some more budget? Okay.
