Jason Lemkin: How 1.2 humans and 20 AI agents replaced sales

How outbound, inbound, and reactivation agents matched a 10-person sales team; just 1.2 humans replaced sales, displacing entry-level SDRs and BDRs first.

Lenny RachitskyhostJason Lemkinguest

Jan 1, 20261h 42mWatch on YouTube ↗

EVERY SPOKEN WORD

150 min read · 30,489 words

0:00 – 4:36
Introduction to Jason Lemkin
1. LRLenny Rachitsky
  You used to have about 10 people full time. Now you have 1.2 humans, 20 agents.
2. JLJason Lemkin
  We have 10 desks that used to be go-to-market people. They're all just labeled with our agents. Reple for Replit, Quali for Qualified, Arti for Artisan. AgentFour needs a nickname. Agents work all night and they work weekends and they work on Christmas. We're done with hiring humans in sales. We're done.
3. LRLenny Rachitsky
  The business is doing very similarly to what it was when you had 10 humans.
4. JLJason Lemkin
  If I had two more great humans that wanted to join, don't get me wrong, I would hire them tomorrow. But I'm not going to hire someone that after their third month on the job doesn't know what SaaStr does. I just can't do that. AI is replacing the jobs people don't want to do today, and it is displacing the mid-pack and the mediocre.
5. LRLenny Rachitsky
  How do you see the future of the sales profession?
6. JLJason Lemkin
  We should have $250,000 a year SDRs, but they'd be like at Vercel, they'd be managing 10 agents, not 10 people. The classic SDR junior kid that is hired out of college to send emails, we don't need them. Folks that qualify leads coming in, the contact mes that we see, we have no need for them today. They should be extinct next year.
7. LRLenny Rachitsky
  Someone's listening to this like, "Oh man, my job is in trouble."
8. JLJason Lemkin
  If you can go do this, you're hyper employable.
9. LRLenny Rachitsky
  Today my guest is Jason Lemkin, founder and CEO of SaaStr, the world's largest community for B2B founders, and one of my absolute favorite sales and go-to-market minds on the planet. Jason is not only deeply knowledgeable about everything sales, he's also extremely articulate and direct, and is also now personally going super deep on what AI can do for a sales work. He's transformed his own SaaStr sales team from around 10 SDRs and AEs to one full-time AE, a part-time chief of AI named Amelia, and 20 AI agents. He is seeing the same performance from his AI team as he saw with his former human team, and he's just getting started. These are my favorite kinds of conversations because the guest is living in the future and comes here to show us what the future is like, where we're headed, and how we can get there ourselves, and also just how to avoid all the pitfalls that he had to deal with along the way. We cover all of the things that he has learned about where sales and go-to-market is going in the AI age. He gives a bunch of advice for salespeople and the future of their careers, the future of the go-to-market org, how to win as an AI startup right now, what tools he's finding most useful, what it took to shift his sales team, and so much more. This episode is going to get your mind spinning in the best way possible. If you enjoy this podcast, don't forget to subscribe and follow it in your favorite podcasting app or YouTube. It helps tremendously. And if you become an annual subscriber of my newsletter, you get 19 premium products for free for an entire year, including a year free of Lovable, Replit, Bolt, n8n, Gamma, Linear, Devin, PostHoc, Superhuman, Descript, Whisperflow, Perplexity, Warp, Granola, Magic Patterns, Raycast, ChatBRD, Mobben, and Stripe Atlas. Head on over to lennysnewsletter.com and click Product Pass. With that, I bring you Jason Lemkin, after a short word from our sponsors. Today's episode is brought to you by DX, the developer intelligence platform designed by leading researchers. To thrive in the AI era, organizations need to adapt quickly, but many organization leaders struggle to answer pressing questions like, which tools are working? How are they being used? What's actually driving value? DX provides the data and insights that leaders need to navigate this shift. With DX, companies like Dropbox, Booking.com, Adyen, and Intercom get a deep understanding of how AI is providing value to their developers and what impact AI is having on engineering productivity. To learn more, visit DX's website at getdx.com/lenny. That's getdx.com/lenny. This episode is brought to you by V0 from Vercel. V0 is the web development assistant designed for professionals of all technical backgrounds. Whether you're a product manager, designer, or developer, transform how you bring products to life. With V0, everybody can cook. Don't just show up to reviews with docs and ideas. Arrive with working prototypes that demonstrate real functionality. V0 drafts project plans, generates interactive interfaces, and builds full stack applications without writing a single line of code. And with features like AI and database integration, screenshot import, and sync with GitHub, V0 helps reduce development bottlenecks and enhance collaboration between technical and non-technical team members. The result? Faster iteration and a shorter path from idea to implementation. Vercel built V0 for the builders who want to create at the moment of inspiration. If you can dream it, you can ship it. Visit vercel.com/lenny to get started. That's vercel.com/lenny.
4:36 – 7:13
What SaaStr does
1. LRLenny Rachitsky
  Jason, thank you so much for being here and welcome back to the podcast.
2. JLJason Lemkin
  Lucky to be a super fan and then to get to join, right? It's- it's- it's terrific to- to- to be on the other side.
3. LRLenny Rachitsky
  So this is our second conversation. Uh, we did our last conversation a year and a half ago. I don't know if you know this, but it became a pretty legendary episode. It's something people continue to share and, uh, that conversation was around basically a deep dive into building your sales org. And a lot has changed-
4. JLJason Lemkin
  Yes.
5. LRLenny Rachitsky
  ... for you and the world since then, uh, cough AI.
6. JLJason Lemkin
  (laughs)
7. LRLenny Rachitsky
  And what, uh, what's happened is you've gone extremely deep on what AI enables for sales, for startups, for go-to-market. And what I love about conversations like this is you're basically living in the future and you're here to give us a glimpse of where things are heading and tell us how to avoid the wrong turns and just help us get there ourselves.
8. JLJason Lemkin
  That I think I can do. (laughs)
9. LRLenny Rachitsky
  To start-
10. JLJason Lemkin
  Yes.
11. LRLenny Rachitsky
  ... help us understand just the business you run, this, uh, SaaStr.
12. JLJason Lemkin
  Yes.
13. LRLenny Rachitsky
  What is it you sell? What is the business? What is it, what is it you do, Jason?
14. JLJason Lemkin
  You know, I'm still trying to figure it out, Lenny. Maybe you are too in some ways. Um...
15. LRLenny Rachitsky
  Yeah, that's true.
16. JLJason Lemkin
  So, you know, I- I- I am a two-time founder who started writing a blog, my God, in 2012, about all the mistakes I made after I sold my startup to Adobe.We started doing a couple meet-ups. You've done meet-ups. Um, before anyone did this stuff, we did a big meet-up in 2015, thousands of people came. Then we do 10,000 people a year at SaaStr Annual. I've actually also invested almost $200 million, almost 10X lifetime, um, only into founders from the community. Um, but the reason I don't know is, wh- what I... Eh, eh, folks that do... are connected to our content, I'm just passionate about helping other founders see mistakes and make less of them. So anyhow, we're- we're a large community. I do invest but- but it turned into a business. There was no revenue in the beginning. I'm sure there was no revenue for- for Lenny in the begin-
17. LRLenny Rachitsky
  Yeah.
18. JLJason Lemkin
  And ours was less intentional. But we do do eight figures of revenue a year and it's work. It's- it's... That sounds great, okay, that you do eight figures, but there's a lot of costs, especially on the events side. The media side has, uh, almost no cost. And it's work. We have 100 sponsors, and- and as you know, like, you have sponsors, but there's a certain level where it becomes a lot of work. Like getting two folks to sponsor your podcast with a couple emails, no work, right? Uh, if you wanted to have like four Lenny podca-... Like it just, it just scales. So we have built our own go-to-market team over time, and I've lived the frustrations. Folks have followed it and then, maybe
7:13 – 10:11
AI’s impact on sales teams
1. JLJason Lemkin
  I'm rambling a little bit, the interesting thing, the a-ha moment that happened to us is going into May of this year we had one AI agent in production called Delphi that we both used for Digital Lenny and Digital Jason, really interesting learnings. And we went into our 10,000 person event this year with- with an ei- you know, a seven-figure budget and eight-figure top line, and two folks on the sales team who are, were paid high end of market. I have many flaws, but paying well and being loyal are not one of them, and two of them just quit at the event. They just quit on site. Okay? And I... This is like the third time I've done this, like the eighth team I've built, and- and I turned to Amelia, our Chief A Officer, and I said, "We're done with, you know, hiring humans in sales. We're done." We are gonna push the limits with agents. We're gonna push... Even if it doesn't quite work, okay? And I knew from this Delphi, this general agent, that it would sort of work because going into Annual, this general agent, this Digital Jason, closed a 70K sponsorship on its own. So when I saw that a horizontal agent, not trained for sales, not trained for GTM, could close one deal, we were like, "Let's deploy a couple of these apps. We have time." And I just can't pay a junior SDR $150,000 a year to quit. I just can't. Like, criticize me, but I just couldn't do it one more time. I just couldn't do it one more... And I actually think this is an en-... When I talk to CEOs at leading AI companies, they kind of don't want to do it either. They want to have the smallest sales teams they can, as much for cultural reasons, right? Even if, even if Replit only goes f- from 0 to 200, it could have been 220 with a smaller sales team, I think Amjad's okay with it. Right? So it's an enduring thing. But anyhow, so we push the limits and now if you walk into SaaStr's office, it's kind of funny, we have ten desks that used to be go-to-market people. They're all just labeled with our agents. Repli for Replit, Quali for Qualified, Arti for Artisan. AgentForce needs a nickname. Maybe we can make one up with Salesforce. There's Amelia's corner office at one end. I'm in the, I'm in the back of the office and it's just agents. It's the quietest office. And net-net, Lenny, it's about... Here's the meta learning for when we, when we're recording this. The pro- the net productivity is about the same. It's not better. It's not worse. Um, but it's so much more efficient and it's scales, because software scales. So... And, uh, we can talk about what we've learned. I think it's important that it takes time to train these agents. They don't work out of the box. Um, but when you dial them in, when you take your best person or your best script and you train an agent with your best person and best script, that agent can start to become a version of your best salesperson, your best person. And that's what we've learned and how to perfect it. And I just think because... And criticize me, anybody, you or anyone watching or listening, maybe it's not cool I didn't want to hire my, uh, 28th rep that was gonna quit, that... But I just couldn't do it one more time in the age of AI. I'm like, "It's time to go to the bleeding edge (laughs) and just see what we can push
10:11 – 14:18
How SaaStr's AI agents work and their performance
1. JLJason Lemkin
  the limits here." (laughs)
2. LRLenny Rachitsky
  Okay. I- I- I love all the directions we're already heading. Okay. So just, uh, help people totally understand what-
3. JLJason Lemkin
  Yes.
4. LRLenny Rachitsky
  ... your business is, basically these people are selling sponsors for your conferences. Uh, is that-
5. JLJason Lemkin
  They're selling two things to-
6. LRLenny Rachitsky
  Okay.
7. JLJason Lemkin
  ... because it's just relevant to the deep dive. They're selling-
8. LRLenny Rachitsky
  Yeah.
9. JLJason Lemkin
  ... sponsorships which average about 70 to 80K.
10. LRLenny Rachitsky
  Mm-hmm.
11. JLJason Lemkin
  And then they're also selling tickets which is the high volume for... This is like the- the self-serve version. They're selling tickets that are anywhere from a couple hundred bucks to, if you're a VC that comes the night before, could be two grand. Okay? No- no gifts for VCs that decide the night before. Founders that decide early get, get in at about ten- twenty percent of cost. And- and- and it's work to sell these tickets, right? And so you can just post an email like you do and you probably fill up Lenny's summit, but if you want to max it out you gotta do work. You gotta do drip campaigns. You've got to reach out to people. You have to reactivate folks that came to Lenny's summit, you know, three years ago but you want them back because they're good people and that just requires work, and as your base scales... You know, you have... How many people subscribe to Lenny's newsletter now? 1.2 million or something?
12. LRLenny Rachitsky
  1.2 million, roughly.
13. JLJason Lemkin
  Okay. How many as, how many of those is a human willing to reach out to? Approaching none.
14. LRLenny Rachitsky
  A few thousand, yeah, right?
15. JLJason Lemkin
  Im- imagine you hired a 21-year-old SDR fresh out of junior college and said, "Here's my list, 1.2 million people. Start calling them." (laughs) "I want 'em to come to Lenny's summit." But anyhow, so we have this low-end version which is tickets, right, which is four or five million a year, and then we have this higher end sales cycle and they're very different, and actually they have different agents. And then we have a different agent to get people to come back, lapsed people. So we have lapsed, high-end and low-end agents, and they have different workflows and we actually use different vendors for now. For now-
16. LRLenny Rachitsky
  Got it.
17. JLJason Lemkin
  ... we use different vendors.
18. LRLenny Rachitsky
  Okay. And so, uh, previously...
19. JLJason Lemkin
  Yeah.
20. LRLenny Rachitsky
  Before this future world, how many SDRs did you have? How many salespeople in this-
21. JLJason Lemkin
  We would have-
22. LRLenny Rachitsky
  ... amount of teams?
23. JLJason Lemkin
  ... two to three SDRs and up to five AE s.
24. LRLenny Rachitsky
  Okay. So like eight, nine people-
25. JLJason Lemkin
  Yes.
26. LRLenny Rachitsky
  ... full-time working on SaaStr-
27. JLJason Lemkin
  Yes.
28. LRLenny Rachitsky
  ... to bring in sponsors and to bring in tickets?
29. JLJason Lemkin
  Yes. Although, yes, a lot of it is inbound and renewals, but no-no- but yes, to br- to manage that business, to manage the sales side of that business.
30. LRLenny Rachitsky
  Got it. Bring 'em in, manage them, help them be successful.
14:18 – 19:19
How go-to-market is changing in the AI era
1. LRLenny Rachitsky
  Okay. I wanna spend time on the different agents you've built. But first of all, just kind of zooming out, having gone through this, this experience, how do you see the world of go-to-market changing next year, in the coming years?
2. JLJason Lemkin
  All the plays work. It's the playbooks that are kind of broken in the age of AI. All the plays work. Outbound still works. Webinars still works. Podcasts still work, okay? Events still work. All this stuff works. All this stuff works. Why is ElevenLabs out doing a road show, right? It works. Why, why do they go on Lenny's podcast? It works. So the plays all work. It's just the playbooks are broken because at, at... For folks that aren't in the age of AI, growth has decelerated so much that nothing seems to work, okay? It's working, it just works so much worse than 2021. But the plays still work. They just, they don't have enough ROI. There's not enough budget for old school SaaS from 2021. The ones that are blowing up, right? The Vercells, the Replets, the ElevenLabs, they have so much demand, so much demand that, you know, that they're still running the plays, but they're, they're doing them differently. They're doing them from a hyper PLG focus because there's so much demand, and they're often picking and choosing which prospects to ta- to, to contact. So like, for example, Bolt is probably a distant number three behind ReplIn and Lovable, right? But one of my old sales guys runs sales there, and I talked to him when they went from zero to 50 million in like six months. He's like, "We honestly just have so many leads, we just... Our... Half our job is picking which ones to respond to." Right? And he's like, and he also is like, "We closed a seven-figure deal we stole from Lovable because no one called them back at Lovable." So your traditional B2B SaaS company, even ones at billions of revenue, even the HubSpots and the, and, and, and all of them, they don't have so many great leads, they don't call them back. (laughs) So that is a different world. Um, not easy, different world. And then this, this world where nothing seems to be working is just because the demand has evaporated, right? So both ends have an incentive in 2026 to push the limits for AI, for go-to-market. The ones that are hyper growing can't touch everybody. They can't do everything. Not everyone like Vercel will build their own internally. We can talk about why most folks should not build, they should buy. For the same reason that's always been true in software. We can talk about it. At the low end, you still need humans, but ruthless efficiency is going to be the name of the game for 2026. So anything where AI works, the demand is inexhaustible. So everyone's either looking for more efficiency or they just can't service the massive amount of inbound they have. Um, and so maybe that doesn't totally answer your question, or I got a little bit off track, but that's how the world's changing. Like when we first did this, which wasn't that long ago, in a way, pre the AI explosion, all B2B companies were kind of the same. Like they grew at somewhat different rates. Some blew up faster like a Samsara, some took longer like a UiPath, but come on, they were all... They all kind of grew the same way for the same ACV, for the same deal size. Now, just like in venture and everything else, it's wildly bifurcated, right? You've got the low end, which is all about price increases and forcing things onto the base, and at the high end, we have something we've never seen since 2020, which is everyone in the market at once. Everyone in the market at once. This is something that people don't understand. Why, why are, why are these companies doing so well? Why are they blowing up? Because they're... It's not just that the software... We, we love this stuff, Lenny, right? All these new tools, we love them. But it's not one law firm looking at Harvey, it's everyone. It's everyone, right? It's not a few folks looking at video on the internet, it's everyone trying to make video on the internet. (laughs)
3. LRLenny Rachitsky
  Because there's a lot of push from the top of like, "We need to adopt AI, we need to be more productive."
4. JLJason Lemkin
  Now, everyone.
5. LRLenny Rachitsky
  We've got the... Yeah, now.
6. JLJason Lemkin
  Not the traditional. Like the traditional metric was, in most categories, 3 to 5% of prospects would be in market a year.
7. LRLenny Rachitsky
  Hmm.
8. JLJason Lemkin
  So you'd send a trillion emails and you do cold calls and you'd hope maybe 2026 was the time they were willing to dump Salesforce for your new product. So add all that up, 5%. In many categories, we're up north of 50% in market.So that just totally changes. The plays still work, showing up in person, actually knowing what the hell you're selling, knowing how to get through procurement, all of those at work. But other than these weird windows, artificial windows in 2020, we've never had so many people in market at once.
9. LRLenny Rachitsky
  And this is for AI products specifically or-
10. JLJason Lemkin
  Yeah, that-
11. LRLenny Rachitsky
  ... even beyond that?
12. JLJason Lemkin
  ... that have massive ROI.
13. LRLenny Rachitsky
  Hmm. Yeah, productivity.
14. JLJason Lemkin
  Right? I want, I, I need to bring a VibeCode tool into my company, Lenny. Okay, go out and do the work. Compare Replit, Levable, and whoever else, and buy one. Okay? Harv- like why are Harvey and the others and Lagora are doing so... I mean, they're great tools, but everyone's like, "We need to automate how we review contracts and documents with AI now." They want a leader and they're gonna do it. And that, that will slow down. Like not everybody can be in market every year. It, it's exhausting enterprise. So this, this is a version of the AI bubble that will end, and we will revert in some ways to old school. But when everybody's in market, it just, it just changes how you run the whole thing. So the b- so the fastest growing ones and the slowest growing ones both have incentives to use AI here, just dif-
19:19 – 22:03
The future of SDRs, BDRs, and AEs in sales
1. JLJason Lemkin
  for different reasons.
2. LRLenny Rachitsky
  How about the sales profession specifically? Are SDRs gonna be replaced fully, AEs? How do you see the future of the sales profession?
3. JLJason Lemkin
  The classic SDR junior kid that is hired out of college to send emails and respond to, respond to inbound emails and maybe get back to them later that day or the next day, we don't need them. We're not gonna need most of them. SDRs that knock on doors in a lot of industries aren't gonna be displaced, right? The email-based cadence SDR will be 90% displaced by AI next year. The people have different nomenclature. I call BDRs, folks that qualify leads coming in, the contact knees that we see, we have no need for them today. They should be extinct next year. There is no reason in the age of AI I have to hit contact me, wait two, a day or two for a 21-year-old that doesn't know what Linear does to say, "Hey, what do you do? How much are you willing to pay me? Maybe I'll set up a call with Lenny later this week." There is no need to do that with AI. The AI, our AI alone, one of our agent fully qualifies everybody on the website. So they don't even know they're being qualified and just sets up the meeting with the salesperson. So this SDR, this email-based SDR and these human qualifying leads, which is not good for the customer, it doesn't feel good to be qualified, does it? They will be mostly extinct next year. I'm guessing with your... Now the AE, the classic human doing the sales, most of the tools aren't there yet for the most part. I think 70% of their jobs will be safe by the end of next year, but I think it will decline to 40 or 50. I don't think there's any reason what we're seeing in other categories a great agent can't close the deal too. If there's not a lot to negotiate on price and the agent knows the product better than a human, at least for folks like you and me, I mean, do you like to talk to a human in sales?
4. LRLenny Rachitsky
  Sometimes. Uh, I'd rather, I'd rather just cha- chat with a, yeah, a really smart AI.
5. JLJason Lemkin
  Yeah. So that's all in progress now, but the classic... And the tough part, and a lot of folks ask this question, Lenny. They say, um, "Okay, Jason. I, I see that in, in your data. How are we gonna build the sales profession if there's no entry-level jobs in SDRs and AEs?" And that's a meta question across all of AI. We're already seeing AI concentrate strength in sort of mid-tier folks, isn't it? And we're already seeing lots of folks cut back on entry-level hires, you know, Shopifys and others aside, where they'd rather have the six or 7-year-old engineer that's a cursor machine rather than train some kid. It's just more efficient today, right? I'm sure you see that across a lot of folks you talk to. It's gonna happen in sales too. So the folks that know how to manage an agent work with an agent, the folks that know their product for real, they're gonna become more valuable and the rest
22:03 – 23:43
Why leadership roles are safe
1. JLJason Lemkin
  are gonna become less valuable.
2. LRLenny Rachitsky
  It's interesting because, uh, I'm an investor in a bunch of startups, as are you, and I'm actually seeing a lot of asks for go-to-market people, salespeople. Do you think this is kind of a temporary because there's so much demand, they're like, "Oh, we need people to help," and then this will start to become more AI over time? Or are they just looking for these really senior people that you're talking about?
3. JLJason Lemkin
  Well, listen, whether you're managing humans or orchestrating agents, you need leadership. We've yet to produce an autonomous CEO. I know folks talk about how on, on Twitter that will, AI will replace the CEO, but I don't know that that's literal as much as figurative, right? So we're still gonna need the C-suite. We're still gonna need VPs. I mean, it's become so much work to manage a million leads, right? A half million leads. So we, we need these leaders whether... The, the question is, and I've seen your call out there and I saw your tweet on it. The question is, how many of the folks that had the current playbooks are the right folks for the future? I'm thinking maybe 20% of the folks I talk to. 24 are still panicking about AI. And, and I'll tell you how to not, how to be in the 20% if you want to know, but I think very few like the Janine from Vercel are gonna make the jump. So we'll see. There will be huge organizations, right? Like Denise just went from Slack and Salesforce after 14 years to be CRO of OpenAI. She's working, gonna be pretty high level, right? So she-
4. LRLenny Rachitsky
  CRO.
5. JLJason Lemkin
  ... may not know how to, need to know how to implement the agents, but most of the folks your companies wanna hire, I would just make sure they could, th- they really wanna roll up their sleeves and do the job of 2026, 2027, right? Um, uh, uh, just because they worked at Slack does not necessarily mean they have the skills
23:43 – 28:40
How to be in the 20% who thrive in the AI sales future
1. JLJason Lemkin
  at your startup.
2. LRLenny Rachitsky
  You said that you had some tips for folks to actually be this 20%. What are some... If someone's listening to this like, "Oh, man. My job is in trouble. What should I focus on?"
3. JLJason Lemkin
  It's gonna sound simple. It will work, and most, almost nobody's doing this. Pick a tool, an agent, an agentic tool to solve one of your problems. It doesn't almost ma- Just one that's the most painful or the one that's most ac- it could be support, it could be SDR, it could be inbound qualification. Pick one. Pick a leading vendor. I don't care which one it is, okay? We can talk about how to pick a vendor, but pick a leading vendor that treats you well, that you like, and do it yourself.Train the agent, ingest the data, do the iterations, understand how this damn thing works, okay? The folks that are lost today have never done it. We literally, we've turned into a consulting shop, Lenny. It's kinda crazy. I don't, I don't know what I think about it, but we literally just did a job. Amelia, our chief AI officer, and, and Mia, she, she, she drove it, we did a call with a public B2B company worth well over 10 billion that you would think is an AI leader, okay? And we did a call with their team, and they're like, "We're str- we wanna figure out this AI SDR stuff." We're like, "Great." They're like, "We think we're just gonna buy this tool and just give it to our SDRs to figure it out on their own." (laughs) Okay, one, no chance. Two, we asked them, "How much of this have you done yourself? Like, have you, have you done it yourself?" And it was just crickets on this call of 20 people. No one had done it theirself. So they thought they could take an untrained agent with no training and just magically give it to a bunch of young 20-year-old SDRs and, and this magically would sell on its own. It doesn't work that way, right? So the way all these agents work is you... There's a lot of jargon which is intimidating, ingestion, orchestration, training. It's not that hard, guys. It's just different. It's the same B2B stuff we've been doing for over a decade. You go to a website, okay? You give it a URL of your website. You give it a URL of what your Wiki is. You give it a URL of your training docs. Maybe you upload your prospectus. You upload a few documents, it ingests the data, and ingesting means it uploads. (laughs) It means it processes the data and it does some other stuff you don't really need to know, some ragging, some vectoring. It really doesn't matter. You, you upload some stuff and it kind of knows it and isn't great at it. And then it will turn it into... Ideally, it will turn it into questions. And you answer these questions and it, they will be get... The more you answer and train it, training this just answering questions and getting better and better. So first you upload a bunch of your stuff, then you spend hours training it, often with the help of a vendor, someone called a forward deploy engineer, which is a scary term. It means someone's gonna help you do this. You upload your stuff, you try to get it right, and then you basically have to make sure it's right. QA, testing. And every day when that AI SDR sends out emails and do practice emails, they will say some dumb things. Maybe it's hallucinations. It really doesn't matter what the technical term is. And you correct it. And each... And if you do this for 30 days and every day you spend an hour or two correcting those mistakes, by the 30th day, it's gonna be pretty good. And this is, anyone can do this that has been in B2B or SaaS. Anyone can do what I just described. It is not that different than other things we've done. It's just sequenced differently. But nobody does this. Everyone's panicked. And if you, if you can go do this, pick any tool, pick, pick, uh, Agentforce, pick Qualified, pick Artisan, pick whatever you want. If you can go do this and get it live into production, you're hyper employable. All the companies you talked about that need a GTM person, they will hire you. You could be a, magically be their chief agentic GTM officer because... But almost anyone can do this if they want to. It's just gonna take a month of your time, and it might take you 50 or 60 hours plus qualifying the vendor, right? And in the old days, like, when we first did our podcast, you'd hire an agency and disappear. That's how you would do this stuff. Don't, it don't work that... The agencies don't know how to do this. You've gotta do it yourself. But if you do, man, you will rock. You will just, you will just rock. And you will learn, right? You will learn, um, and you will learn the limits, and you will learn the agent can, what it can do and where it can't do, and then you will learn how to do the next one, right? So, like, we're pretty far on Agentforce, which is Salesforce's one, which Mark talks a lot about, but we're, we're probably one of the only organizations of our size on it. I will tell you a cheat code, which is pretty interesting. So we've got, we had three of these agents working for sales. After training it and learning it and spending months down to learn it, we got it down to one prompt. And prompt is another almost intimidating word, a string of text that describes what you want this thing to do, okay? We took that prompt and gave it to Agentforce. Within a day, it was pretty good. So you will, if you can do one of these, it'll be really hard. It'll be brutal. Then the second one will be easier, and then you're gonna be like the master of the universe in AI if you can do it yourself. But if you're waiting for people on your team to do it, if you're waiting for an agency to do it, I think you're gonna be out of a job, right? So this is the... Like, everyone comes to us as an experts. We're not, we're only experts
28:40 – 30:10
Why you shouldn't build your own AI tools
1. JLJason Lemkin
  'cause we did it 20 times. (laughs)
2. LRLenny Rachitsky
  I think what might be helpful here actually is to do a, a tour, a kind of a quick tour of the agents that you've built and what they do, which ones have been most impactful. And then as you do that, what products you use. What, what powers these agents that, uh, that you like and maybe don't like?
3. JLJason Lemkin
  If any, if anyone goes to sastr.ai/agents, we'll, you'll see everything we built. It's all bulleted out. You can copy us. And I'll walk you through it, but two, two caveats or things that, uh... I've built a lot of stuff in Replit. We can talk about it for fun. I'm like a top 1% user. I love it. None of the GTM stuff we built ourselves. Don't build it yourself. You're not Vercel. You don't have a full time wicked awesome engineer that wants to build this. Could all this stuff be built yourself? It's the same idea of building your own Notion. You could do it, but don't do it. It's, these products are expensive. They're not so expensive it's worth... And then maintaining. The pace of innovation is so fast, even if you can hire someone to build it internally, it will become obsolete if you're not careful in a couple months. So we've built a lot of stuff. We could talk about, um, uh, I, we, I built a, a, a calculator to do startup calculations and used 800,000 times in 90 days.
4. LRLenny Rachitsky
  For, uh, valuation.
5. JLJason Lemkin
  Yeah.
6. LRLenny Rachitsky
  Right? These are the-
7. JLJason Lemkin
  For valuations. I built a pitch deck reviewer that's reviewed almost 3,000 pitch decks. Lots of fun stuff, but none of the GTM stuff we built ourselves, none of it. So just a caveat, don't build it yourself unless you're Vercel, unless you have a reason. It... That was a great pod. It was wonderful. Don't do it. Don't do that yourself though.
8. LRLenny Rachitsky
  Basically if you, unless you have a, uh, awesome go-to-market engineers.
9. JLJason Lemkin
  For... And they really wanna do it, they're really are chomping at the bit to do it, um, don't do it. So
30:10 – 36:40
Specific AI agents and their applications
1. JLJason Lemkin
  I started... And this is not where other folks would start, but there's some learnings. I started... There's an app called Delphi, which makes digital clones. And, um, you used it for the Lenny bot a long time ago.
2. LRLenny Rachitsky
  Lennybot.com.
3. JLJason Lemkin
  I saw you do it. What's that? Yeah, lennybot.com.
4. LRLenny Rachitsky
  Lennybot.com, yeah, check it out.
5. JLJason Lemkin
  I saw it a long time ago.
6. LRLenny Rachitsky
  Yeah, it's great.
7. JLJason Lemkin
  It was interesting, but I, I, I didn't... It didn't click, and then Brian Halligan who's the founder and chairman of HubSpot, did one too.... uh, they're Sequoia backed, and he was working at Sequoia, so he helped them early on. And then I kinda had a magic moment, and this is the way it works in AI, when I saw the combination of the two. So yours was real, like, people should love LennyBot because it's got, it's... If folks who haven't tried it, try it. It has ingested, I know, a, a scary term for some. It has ingested every single interview you've ever done, right? Every word of content you've written. So i- it can combine them all together. It can combine the Vercel story and, uh, and, and, and, and, uh, what, what you did with, uh, with Dillon at Figma and can synthesize the knowledge, and it's pretty good. It's pretty good. What I liked about Brian's better than yours, though, was that it was Brian. And LennyBot is kinda Lenny, but it's also kind of all your guests, right? I think that's the superpower of it, right?
8. LRLenny Rachitsky
  That's the way I think about it. It's not just my intelligence, the, the lessons of every single person I've had on this podcast.
9. JLJason Lemkin
  So it's great. But I thought, "Hey, maybe I could finally do one that's, that's in between the two." Like, I, I've been a founder, and I, and I've written 10,000 pieces of content, so that's a little bit like Brian, but I have more than Brian. And I'm not Lenny in terms of productivity, but I've got a lot of voices. So I'm like, "I'll try it." I used Delphi. I instantly broke it 'cause I had too much data to ingest.
10. LRLenny Rachitsky
  (laughs)
11. JLJason Lemkin
  It took about a week to get going. Um, and it worked, and people... And it's just like you. People, some people spend hours a day on Digital Jason in another browser, and they don't do what they do with Lenny, but they'll, they'll ask about their sales woes and what to do with their sales team, and they'll, and they'll upload LinkedIns and ask if they should hire people. And then a curious thing happened, which is that because we do these events, people started to use it for questions for the events. "Hey, how do I get a refund?" "Hey, can I get a discount?" "Hey, where is the San Mateo County Fairgrounds?" "So Jason Lies is really in the San Francisco Bay Area." Like, "Who's speaking?" Or, and, and like, there's endless questions, right? And we used to use pre-fin intercom, and we're so busy we would answer like two weeks later. Like it was ter- the worst support ever. And the, the agent just started doing support on its own. And then it did this thing (laughs) where it sold sponsorship on its own, right? So, so start, you can start with... So one place to start, if you haven't started, is in support, okay? And you don't have to buy Sierra and you don't have to buy Decacon necessarily, and you don't have to buy Fin, but one potential place to start is, is your support. Like, can you do great 24/7 support? Can you? Most, most apps can't. In fact, some of the worst offenders are AI leaders. They, they have no support at all on their website. So that's one place to start. Um, and then the next place we started... O- so for us, the low-hanging fruit, the next place we started is, "Hey, we wanna try outbound." Okay? Because we don't have 1.2 million names like you have, but we have like 400,000, okay? And we have data on them, so we wanted to say, "Hey, come, come back to our SaaStr event." So, we didn't know what to use, and I'll tell you some learn-... So we picked a, uh, a YC company called Artisan. They've gone like, from like nothing to 10 million this year. Um, we picked them, but this is important, why? They were at, they were a sponsor at SaaStr. We didn't know, and they offered to help us the most. This is the critical insight. We didn't know if Artisan was the... I have opinions now. We hadn't deployed, but w- another vendor argued, the CRO argued with us. He said, "I need 100K upfront before I help you." Okay? Another one said they were scared of SaaStr, they didn't want bad PR if it failed. Fair.
12. LRLenny Rachitsky
  Oh, but yeah. (laughs) Like... (laughs) Certain companies, yeah.
13. JLJason Lemkin
  We don't wanna be your first.
14. LRLenny Rachitsky
  Got, got a taste of that. (laughs)
15. JLJason Lemkin
  Okay? And Artisan said, "We'll do it." And, um, we had nothing. And here's a c- interesting thing about Agentic stuff. It's like support. If you have nothing, it doesn't have to, like, change the world. If you're literally doing nothing and you start to do something that's high ROI, like, you're gonna get return, right? So, we did that one. We trained it. It's great. We did about 60,000 emails, um, saw pretty high rates. Um, then we said, "Well, we'll try inbound." We, like, we don't wanna have this depressing experience where a salesperson quits and it's two weeks later until they talk to... So we used this vendor called Qualified, which was founded by the XCMO of Salesforce that does a lot now, but mostly focused on Qualified stuff. That immediately worked. Like, we had someone at 11:00 PM on Saturday night that wanted to sponsor, and they sponsored and it worked. And it, and it worked great. Um, but again, they helped us.
16. LRLenny Rachitsky
  And this is an agent that is emailing with prospects, selling them on a sponsorship.
17. JLJason Lemkin
  Well, if literally, if you go to saastrannual.com, and anyone should buy a product like this. It doesn't have to be Qualified, but, but... And there'll be... The bubble, the intercom-like bubble-
18. LRLenny Rachitsky
  Oh, okay, got it. The chat.
19. JLJason Lemkin
  ... is tuned to, to qualifying inbound prospects.
20. LRLenny Rachitsky
  Got it.
21. JLJason Lemkin
  Folks that say, "Hey, I want to sponsor Lenny's podcast." "Sorry, we're sold out through 2028, but if you wanna be on the wait list, (laughs) sign up here." Okay? Or, or even better for us, it would qualify folks out that weren't a good fit, right? It would save so much time, and it would do it 24 hours, then it would just set up the meeting. So, the reason that was a great second one was because no one was willing to do that. No human was willing to pick up the phone and talk to these people. So it was such low-hanging fruit. Um, but the key to the first two, and, and if you're gonna pick an agent, is they, they offered to help the most. The, y- your... I, at the end of the day, Lenny, these are all running on Cloud4. They're all basically using a bunch of APIs mashed together. That's not new to software, right? Mashing a bunch of APIs under the hood. But deep down, I don't wanna get anybo- trigger anybody, many of these leader, the leaders in AI GTM, they're more similar than different. They're more similar than different under the hood. It doesn't mean the features are, are parody. So, because you have to train them, because it takes a month, the world's best software with no help training you is not one 99% of people should buy. So today, in the old days, we would qualify the best software. We would make a matrix and we would do their thing and we would compare features and do this. You gotta do another column which is your forward deployed engineer or a solution architect or your SE, and talk to them and say, "Who is gonna help me?" And before you write a check, get on the phone with Lenny and see if Lenny's really gonna do the deployment. And if Lenny rocks and the other vendor is better and won't help you, don't do it. And that's why we've had so much success, is the first two we did, yeah, they were startups, right? So they worked harder. But Artisan and Qualified just did the work with us. And we're not stupid, but it was work. We needed help, right? And so, that's what I learned, is you, you have this partner, the FD and the vendor, and, um, and a lot of them actually might not take your business if they don't think they can help you. The best ones turn away a lot of business today, which is interesting,
36:40 – 42:11
Challenges and learnings in AI deployment
1. JLJason Lemkin
  right?Uh, interesting learning from this for folks is, um, a lot of folks say... And they would say it to you if you use these, Lenny. They say, "You have too much data. SaaStr is not like us. We're a startup. We're tiny. You have 400,000 people in your database. Lenny has 1.2 million." It's not like that's... I, I got... I, I, I only have 300 customers, okay, or 200 customers. What I've learned is, um, y- that's wrong. If you have 300 customers, how many folks have come to your website ever? 30,000? How many leads do you have? How many folks in your database? How many folks have you tried to reach out before? More than a human's doing. And th- then all of a sudden they have an a-ha moment. I'm like, "How many folks do you have in your HubSpot," right? "How many folks do you have in your CRM?" They look it up. 31,000. Okay. How many folks do you have talking to them? Zero. You don't need the scale of numbers that you and I have to make these agents work. You need a... You need a little bit of scale and you need a little bit of traffic, but not as much as you think. So, all the learnings... We have a lot of folks that honestly they don't want to do the work. They're like, "Well, SaaStr has a lot of scale. They have a lot of years." It's not true. And it turns out to also be, like, true with the training. I'm sure you've seen it with LennyBot. Like, I thought having 12 years of content made the difference. Nah. It's having, like, a couple of months of really good content and a long tail beyond that, but you don't need as deep training and as much as you think. You just need a bit to be really good. So, anyone that has any skill whatsoever, even a couple million in revenue and up, can benefit from these products, right? So, we did horror general-... general bot, got us to a certain place. Then we did SDR, the-... for outbound, then we did inbound, and then we did Agentforce really early with Salesforce. And we didn't know what to do with Agentforce at first, right? Um, but we decided we would reactivate the folks that sales decide was not worth their time, folks that reached out to sales... And this is true at every startup. We even just talked about some of the AI leaders, where a human says, "You know what? I don't think this is enough commission. I'm kind of busy. I got a $4 million deal with Meta going." We just took Agentforce just on those, okay? And, and we trained it on very similar prompt. It had 70% response rate. Those are people that were dying to interact with us. (laughs) 70% is so good at... And this is something humans were not willing to do. It wasn't worth their time. And I know this sounds critical, and maybe I'm going to trigger some sales folks, but the reality is if, you know, if, if you're in a lead-rich environment, okay? And there... And I, I, I think that there's lead rich and lead poor environments for, for even big companies, but startups, like, there's not enough, but y- eventually you become lead rich, okay? Reps just don't follow up with a lot of them. It's just human nature. It's... Even you, I bet more folks want to sponsor t- the newsletter than you can let in, right? Do you hu-... Do you pick up the phone with all of them?
2. LRLenny Rachitsky
  Uh, I reply to all of them and then we just tell them we're, we're full, but-
3. JLJason Lemkin
  Yeah. But you see the point, right?
4. LRLenny Rachitsky
  ... yeah. Um, we, we do follow up.
5. JLJason Lemkin
  Even your scale, you see the point, right?
6. LRLenny Rachitsky
  Yeah. Yeah, it gets challenging.
7. JLJason Lemkin
  Um, and let's imagine all of a sudden you had six months of inventory available. I bet if you spooled up an agent and emailed all those folks back automatically, you'd fill up, you'd fill up the docket, right?
8. LRLenny Rachitsky
  Mm-hmm.
9. JLJason Lemkin
  So, anyone can-
10. LRLenny Rachitsky
  Yeah.
11. JLJason Lemkin
  ... do these sorts of... You think you can't, um... Unless you're so small that you have sufficient humans to talk with every potential lead, every person that touches your website, every person that clicks with anything, you can benefit from AI. So, that was kind of our journey, and then we've done a lot of other niche stuff. I'll tell you at the end, the, the... where we are today, this is, uh... Maybe this is almost too much learning is, we're at the point where maybe we can't do one more because right now when we, when we did Delphi in the beginning, when I copied you with Delphi, even me, I spent almost an hour a day training it in the beginning because w- when we started to use it for support, it had an initial... It started telling people the wrong dates, okay? And we can talk about why. So, I had to fix it and it made some mistakes and so when people started using it, I had to spend an hour each morning firing up Delphi, reviewing the issues and answering them. I don't have to do it anymore. It's well-trained. Um, we have so many agents going and so many emails that Amelia has to spend, you know, 10 to 15 hours a week reviewing the outputs. And it's exhausting because agents work all night and they work weekends and they work on Christmas. It's a big issue, right? This is not... Being the orchestrator or the chief AI person is not a good job for lazy people because the agents never sleep, right? So, it is, uh, so much time now to manage these 20. This is interesting. We can't... I don't know when we're going to do the 21st. We may be full. And for folks that are startups, this is a reason to go harder because everyone was in market this year, okay? Everyone. And it's going to keep happening, but business process change remains an issue for business software. Business process change, at the end of the day, and so many founders get this wrong, and 99% of sales folks, they don't care about business process change, sales folks, they just want to get their commission. Doesn't really matter what you pay for an app for a, for a customer as long as it's fair. It's all the work to do to change the way you do your business, right? So, we're even... We're at the point where we're overloaded, right? And so just be aware if you're, if you're at a startup or even Salesforce or HubSpot, maybe, maybe close those deals in the next 12 months 'cause the window may close where people say, "Listen, that's the coolest agent I've ever seen. I'm exhausted from the last five. I had to do five last year. I just can't... literally cannot bring one more app into my enterprise." And so that's going to be a headwind that today everything seems like it has tailwinds, right? Everything is on fire, but people are going to get exhausted for having so many agents. Exhausted.
12. LRLenny Rachitsky
  Man.
42:11 – 47:31
Making AI-generated emails good (not just acceptable)
1. LRLenny Rachitsky
  Okay. There's so much to, uh, to learn from in what you just shared. Something I definitely want to ask about, as people hear this, uh, agents sending off emails, agents talking to your clients. Uh, we get... I get a ton of emails that are terrible.
2. JLJason Lemkin
  Yes.
3. LRLenny Rachitsky
  H- what have you learned about making these outbound emails good and not just, you know, noise? How do you make these conversations high quality? How do you... Are they... Can they get great?
4. JLJason Lemkin
  It's a really, really, really good question. So, the two, maybe the two biggest learnings, um, take...Your best person on your sales team, the best marketer you have. Take their email copy and use that as the template for your AI. If you ... The, the, the terrible mistake people made, people all, everyone in 2024 said these products didn't work. There were two reasons they didn't work. It was before Claude 4.0, right? Net- Repl.it didn't work. Lovable didn't exist. Gamma didn't really work before 2025, right? Before Claude. Like, the LMs reached this point where they would work for these use cases. So that was one threshold. The other thing that happened in 2024 is the vendors kind of lied and said, "Uh, just turn the product on. It'll get you revenue. No, no need to train it. No need to do anything. We'll just do everything as this magic s- AI savant." It's not, it's not the way it works this way. What you do is, an a- an agent will be successful in go-to-market and sales today if you take what works for your best person, train. I know this seems like a scary term, but it's not. Tr- Upload that text, okay? Train the agent on it and let it iterate an A/B test from that. Agents are really good at A/B testing. They're really good at creating variants. AIs, like, you know, ask Claude or, or Ch- for a variant of your best email. Say, "Give me three versions of my best email." They'll be pretty good. That's all the agent has to do, is take your best email you ever sent and stick it (laughs) through an API and go ... It's, there, it's ... I, I'm m- I'm making it sound simpler than it is, but, but not by too much. So train it, and then what, and then what it'll do is then give it some data sources. And the data source could be as simple as Salesforce, okay? And it, and, and then if it has any data on Lenny, it can pull data, and it can lightly personalize that email, okay? And even better, it, a lot of these products track all the visitors to your website so they can see what's happened and they use other APIs, and so they can personalize your emails more. And so what ends up happening is the emails that the AIs write are pretty good, okay? If you're getting terrible emails, it's a poorly trained product from a bad vendor. You should be getting emails when you get them and you're like, "This isn't s- as good as Jason said on Lenny's podcast, but it's pretty good." Okay? That's what AI can do today. And the magic is if a human isn't even doing that or if your mediocre humans are worse. And I'll tell you, you know, one of the first lessons I learned when my last startup was acquired by Adobe, Sam Blonde was one of our sales leaders. Then he became CRO of Brex and others, and he ... We inherited a bunch of reps from Adobe. We didn't ask for them. And he's like, "My God, I never read everyone's emails before. These are the worst emails (laughs) I've ever read." So the AI can do better than that. The AI can do better than, than that. And so you just train it on your best, and it'll be pretty good. And so, so you just haven't seen a well-trained agent. And then what I learned ... And then another question folks ask is, "Okay, Jason, that email was pretty good. It wasn't as great as you said on stage, but it was pretty good. But do you, do you tell people it's an AI, or do you, or do you hide it?" And what we learned from sending hundreds of thousands is it doesn't matter. People, we're, we're, we're, we're l- we're, we're, we're in an age where people don't really care as long as the email adds value and they know they're gonna get an instant response. We've tried both. We've tried to say, "Hey, it's Digital Amelia or Digital Jason." We've tried to fake it. And what we've learned is now we just send it. We, we just send it and no one cares. And sometimes we'll get, especially founders will get an email back. They'll be like, "Ha ha, I can tell this is an AI, but it's pretty good. Can I do a meeting?" That kind of says it all, doesn't it?
5. LRLenny Rachitsky
  Mm-hmm.
6. JLJason Lemkin
  So we're worrying or we're creating issues as excuses to not do the work.
7. LRLenny Rachitsky
  Your point about how sale-, human salespeople's emails are not great already is really powerful, 'cause all we're looking at is these okay emails from AI and you're saying, "Okay, but humans, they're not actually that much better if you actually look at them."
8. JLJason Lemkin
  My God, they're not. The be- Listen, the best outbound emails you've ever gotten. Um, like for example, I know you've done a bunch of investments. Do ... A lot of them are inbound to you. They want Lenny involved, right?
9. LRLenny Rachitsky
  That's right. That's right.
10. JLJason Lemkin
  Some of them are just so good you can't believe it, right?
11. LRLenny Rachitsky
  Hmm.
12. JLJason Lemkin
  A few.
13. LRLenny Rachitsky
  Yeah, yeah.
14. JLJason Lemkin
  How many are that? But, but a lot of them aren't, right?
15. LRLenny Rachitsky
  Right. Mm-hmm.
16. JLJason Lemkin
  So like the best founders and the best sales execs and the best SDRs will spend two hours researching an email. Okay. Who exactly at IBM should I reach out to? What did IBM ... Who else exactly is a competitor that's using them? Exactly what was the ROI? They'll give you a perfect story. Like, if you get the world's best story. Here's your competitor. Here's how they use it. Here's exactly when they bought. Here's the ROI. Here's the case study. That's gonna ... That's a great email, right? How many 21-year-old SDRs do that? No, they, they, they use a tool, an automation tool, whether it's Outreach or Gong or SalesLoft or Mixmax or an AI-based, but they do no work. It's not gonna be that great. It's not gonna be that great. So that's why for, for ... People get a little confused. The bar for good enough for AI GTM, it's not as high as we think. (laughs) It's just like a, you know, a, a facsimile of your best person reproduced as best we can. It's gonna beat your mid-pack person. It's gonna beat the person that literally knows nothing
47:31 – 52:39
When humans still beat AI in sales
1. JLJason Lemkin
  about your product.
2. LRLenny Rachitsky
  Is this, is this an opportunity for humans to continue to thrive, this layer of much better emails? Uh, this came up when I had Jenn Abel on the podcast. She ... I asked her, like, "What tools do you use? What do you use?" She's like, "Nothing. I just write it out (laughs) artisanally. Uh, and that works really well, 'cause everyone's sending AI emails." Is this just like where go-to-market salespeople still can exist, this much better email? Or is that also getting away?
3. JLJason Lemkin
  Look, for, for, for if you have a high performing human team hunting high dollar value logos, and when we're ... And this is classic stuff. Lenny and I are, and, and Jenn are in a, in a conference room. We put a whiteboard of the 50 best folks that we want to sponsor Lenny's podcast. There's only 50, okay? And we write Notion and we write Linear and we write Repl.it. There's only 50. It's a th- And they, we're all trying to sell them these new sponsorships. They're, they're, they're half a million bucks for two years. Take it or leave it, okay? And I give ... We divide them up and we say, "Lenny's best at this, Jason's best at jen, and we each take 15 or 17." Dude, no need for AI there, is there? Agreed.
4. LRLenny Rachitsky
  Right.
5. JLJason Lemkin
  No need for AI there today.
6. LRLenny Rachitsky
  'Cause the, 'cause the ROI is really high, a lot of the time.
7. JLJason Lemkin
  Yeah, and we're gonna ... And we're great, and, and we know that when we're ... The three of us are, we're different. The three of us are really d- It, it's gonna crush, and we don't need any AI. Maybe one of us will take our emails and run it through Claude real quick just to make it better, right? Or, or what I do is I do it for more research. Like, I write the world's best email and then I say, "Claude, how could I make this better?" Do a little research. It will still be better. So that's an AI boost Jenn should be doing. I love Jenn, but she should be d- she should at least be making it better. But for our 45, 50, 50 best ones, we don't need it. What if it's 5,000? Her approach just doesn't work.So yes, a lot of the stuff we're talking about lends itself to higher volume sales. But as everyone gets bigger, it's all higher volume. It's all... There's just so much volume as you scale, right? So yes, if you're tiny and you have three prospects and you're, you're just getting into Y Combinator, maybe you don't need these tools, but we, we graduate out of that more quickly. And, and the bespoke thing for Jen I think will work for high dollar value enterprise, but is... Outside of that, I don't know, man. Uh, it's, uh, you just can't touch enough people. Humans can't touch enough people, and humans don't want to do the work. They don't want to talk to the mediocre leads. They literally d-... I'll tell you an ex-... Like, when I was in London, I wanted to buy a $10,000 product, okay? And I'm literally in London, I, and we're doing SaaS when I don't have any time, right? And I, I get confused with the time zones, Lenny. I don't know if you do. I don't even know what time it is in, in the Bay Area. So I just emailed this rep. It's the end of the year. I'm like, "Just send me the contract. I wanna buy it. But I have two questions." I have two questions I want answered." And I told him these two questions, and they weren't even about price. Took him three days to get back to me, and he introduced me to someone else on his team. It wasn't worth his time. 10 grand wasn't enough because there's not enough commission for him, right? So he introduced someone else to me and the other guy said, "I can't answer your questions unless you'll get on the phone." And I said, "I'm in London. I'm traveling, if you will." Ordinarily, I would've ended this, but I'm, I'm... It's a journey. I'm like, "If you answer my two questions, I will buy your product for 10K." He's like, "I need to get on the phone." Like, AI's better than that. This is not Jens and the whiteboards thing of doing it. So, um, it will f-... It, it at least it will fill the... Even if Jens' process is right, AI can fill all the gaps. What about all the sponsors? We... Uh, th- the leads we didn't follow up with that we got a 70% response rate, right? I mean, the Jens are a diamond in the rough, in a di-... Whatever the expression is. A diamond... There's not that many of them. There's not that many of them. So, uh, I, I love her and I love what she says, and I agree with 99% of it, but h- here's a p-... Uh, um, a related point. Most of us are... Don't have the hottest brands, and we don't have the most elite CROs running them. So we end up settling for not the best sales team. That's the truth. S- most... 99% of the best sales reps wanna work just at the hottest brands. And the minute you're not, the minute your star fades just a little bit, they don't wanna work. They, they immediately wanna jump to the next one. It's just... There's a lot of reasons why, so bear in mind, 99% of all... Of the world cannot attract a team of Jens or better. (laughs) It's just practi-... Even I can't. Even you couldn't. Lenny, you're so great, but even a lot of folks that would want to go work for you, if you wanted to hire someone, they'd be like, "Well, Len-... I love the pro-... But what do I have to do? I've gotta sell newsletter spon-... Like, no, no, no, no, no, no, no, no. I wanna be CRO at, at Lovable. I, uh, I love Lenny, but is that really gonna get me there?" Right?
8. LRLenny Rachitsky
  Yeah. Yeah.
9. JLJason Lemkin
  So we... AI can, AI can beat those.
10. LRLenny Rachitsky
  Hmm.
11. JLJason Lemkin
  But AI can't beat the enterprise thing. AI... I, I have no idea how AI is gonna do in-person sales. I, uh... Someone smarter than me is gonna have to answer that. But, um, a lot of... I... You know, you have such a huge audience, Lenny, but I still think most of your folks are in tech and doing tech sales. Tech is the largest segment of our economy and growing, right? So for the most part, these tools will work for tech sales. Tech sales is over the Zoom, over the phone, over email. We're not... We should knock on more doors. We should do more in person. All the data I've ever collected shows everything closes at a higher rate if you go in person. It's just not... But in tech it's basically a- as much automation as we can get away with
52:39 – 53:50
An overview of SaaStr's org
1. JLJason Lemkin
  in GTM.
2. LRLenny Rachitsky
  What I think might be helpful is just, like... Let me zoom out for a second and describe what you've gone through here. So you used to have... I love this visual you had of the desks of the sales folks in your office, uh, where you had inbound SDRs, you had outbound SDRs, maybe a support person.
3. JLJason Lemkin
  And, uh, three or four AEs. A- account executives.
4. LRLenny Rachitsky
  Okay, and three or four AEs who kind of take these leads and then close the deal.
5. JLJason Lemkin
  Yup.
6. LRLenny Rachitsky
  And so now ins-... Just like, instead of humans, there's an agent doing each of these jobs.
7. JLJason Lemkin
  Yes.
8. LRLenny Rachitsky
  So you have this inbound... This outbound, uh, agent that's just sending emails-
9. JLJason Lemkin
  Yup.
10. LRLenny Rachitsky
  ... trying to find potential leads, uh, uh, an inbound agent that's talking to people that are interested, trying to get them, uh, more excited. And then is there an AE agent, or I forget what that-
11. JLJason Lemkin
  No, we have-
12. LRLenny Rachitsky
  ... how that worked.
13. JLJason Lemkin
  ... uh, that's one I'm still learning. We have one full-time AE plus 20% of Amelia's time, so it's 1.2, doing what five or six AEs do.
14. LRLenny Rachitsky
  I see. So basically-
15. JLJason Lemkin
  And no SDR/BDRs.
16. LRLenny Rachitsky
  Got it.
17. JLJason Lemkin
  Yeah.
18. LRLenny Rachitsky
  So basically, all the top of funnels is AI.
19. JLJason Lemkin
  Yes.
20. LRLenny Rachitsky
  And there's one human out that takes all this, all the great stuff and just closes the deals, negotiates pricing, things like that.
21. JLJason Lemkin
  Yeah. Maybe 1.2 just to say it, but, but yeah, one-
22. LRLenny Rachitsky
  1.2.
23. JLJason Lemkin
  Let's call it 1.2.
24. LRLenny Rachitsky
  Yeah, so that's where I wanted to go. So Amelia,
53:50 – 58:37
The role of human oversight in AI operations
1. LRLenny Rachitsky
  so that feels really important, just somebody, not necessarily full time, but just staying on top of these agents, watching their emails, making sure quality's high, making sure they're running correctly. Talk about just, like, how important that part is to this whole operation.
2. JLJason Lemkin
  It, it's critical. It, it's critical, and you know, people are posting on LinkedIn that they want to hire these GTM engineers or... I don't think that role exists today. I, I, I, I worry when I see these roles. I think today... And listen, if we get together in 18 months, we'll, we'll update this 'cause the world's changing so fast, right? I think today, 95% of 100, you've gotta promote someone internally. It's gotta be a nerd, someone that likes marketing and sales and is quant. You know, a lot of B2C people are frankly good at this stuff because in B2C, sales and marketing are kind of the same thing, you know? But someone that's a nerd that loves to sit in front of data for a couple hours a day and, and route data and manage these agents. And, uh, they could come out of product, they could come out of marketing but maybe they could come out of RevOps, but they better be nerdy. Odds they come out of regular sales approach zero. So I would find someone on my team that raises their hand and says, "I've already done this, okay? I, I've already, I've already written 10 apps in Replit and I, and I, and I love Vercel and I did this and I've already tried these ones on my own. Can I please manage these for you?" And then have them be your chief orchestration officer, but it is a new skill set.... it, it really is. And, and ultimately finding someone that's gonna spend an hour to two a day to manage these agents is the new, new frontier for us to figure out. Um, they, they do not, they, they operate autonomously, but w- not without constant oversight and iteration. That's the-
3. LRLenny Rachitsky
  Hmm.
4. JLJason Lemkin
  ... confusing part. It, it... And, and you just, if you just buy one of these products and disappear, you will have zero ROI. So, maybe too long of an answer, but that's critical, and I just think unfortunately, you're gonna... You have to grow this resource at home today. You have to. Even with, between the ver- Vercel that basically grew the resource at home, right? We're not all Vercel, but I just haven't seen it. Everyone's posting for this job, um, but we need veterans. We don't have veterans yet, right? And going back earlier in the conversation, if that is you, you're gonna be super employable next year (laughs) . You're gonna have so many job offers, you're not gonna know what to... You're gonna have to beat them off.
5. LRLenny Rachitsky
  And when you think of... Is Amelia, would you describe her as a go-to market engineer or do you... Is that a different role? Chief, uh, you call her-
6. JLJason Lemkin
  I would say now, yeah.
7. LRLenny Rachitsky
  ... Chief AI Officer?
8. JLJason Lemkin
  Yeah.
9. LRLenny Rachitsky
  Okay. Cool.
10. JLJason Lemkin
  But she knows the products cold, cold. She knows how the... all the quirks work, how all the agents work. And, and here's a... This is, this is a, a complicated issue, but an interesting one. If you're running multiple agents, okay, someone's gotta segment which of the base the agents are working with or they're gonna have-
11. LRLenny Rachitsky
  Mm-hmm.
12. JLJason Lemkin
  ... tons of conflict. You need someone smart enough and a- any, like, really nerdy demand gen marketer that loves data can do this, but you've gotta segment your base. Otherwise, it just becomes a mess. There are no people on X and the internet talk about these master agents that can manage agents, that can manage agents. We're not there yet, okay? May- maybe, like I'm excited for it, but we're not there yet. So, just even figuring out how to segment your base so you can do inbound, retargeting, remarketing, new marketing, like so, uh, so that, that is complicated. But most badass marketers kind of understand that. They're already doing AB testing, segmenting their bases. This is not new, is it?
13. LRLenny Rachitsky
  No.
14. JLJason Lemkin
  No. Yeah.
15. LRLenny Rachitsky
  It's, yeah.
16. JLJason Lemkin
  But turning it on with zero work is a F. Like there's just, it's just no chance.
17. LRLenny Rachitsky
  This episode is brought to you by Datadog, now home to Eppo, the leading experimentation and feature flagging platform. Product managers at the world's best companies use Datadog, the same platform their engineers rely on every day to connect product insights to product issues like bugs, UX friction, and business impact. It starts with product analytics, where PMs can watch replays, review funnels, dive into retention, and explore their growth metrics. Where other tools stop, Datadog goes even further. It helps you actually diagnose the impact of funnel drop-offs and bugs and UX friction. Once you know where to focus, experiments prove what works. I saw this firsthand when I was at Airbnb, where our experimentation platform was critical for analyzing what worked and where things went wrong. And the same team that built the experimentation at Airbnb built Eppo. Datadog then lets you go beyond the numbers with session replay. Watch exactly how users interact with heat maps and scroll maps to truly understand their behavior. And all of this is powered by feature flags that are tied to real time data so that you can roll out safely, target precisely, and learn continuously. Datadog is more than engineering metrics. It's where great product teams learn faster, fix smarter, and ship with confidence. Request a demo at datadoghq.com/lenny. That's datadoghq.com/lenny.
58:37 – 1:05:40
Advice for salespeople and founders in the AI era
1. LRLenny Rachitsky
  I'm looking back at some notes I took as you were talking of just, like, advice for each function almost of how to be successful in this future that you're seeing. So I'll try to summarize briefly. So your advice for salespeople is use the agents, build one yourself, try to train it, help it run, run alongside you so that you understand these tools and be the person within the sales org that's, uh-
2. JLJason Lemkin
  For m- For leadership, I advise that.
3. LRLenny Rachitsky
  For leaders.
4. JLJason Lemkin
  I don't know that if your average SDR or junior salesperson is gonna get budget for their own agent. If they do-
5. LRLenny Rachitsky
  Mm-hmm.
6. JLJason Lemkin
  ... run with it. The problem, Lenny, is that all these agents that work today, they have forward deployed engineers, they have training, so they're all like 50 grand and up. 50 grand, 80 grand. I mean, people pay more, don't get me wrong, but kind of the entry level point for these is sort of like 50 grand plus 25K for the FDE or 75. Like, they want... Even Clay I think starts at a 100K a year. So if you're a bigger organization, that's cheaper than a human, right? It's, it's a tall, but, but there are no $99 a month, uh, products, which has a lot of it... They're, they're trying and, and I think it's gonna come. They don't auto train yet particularly well. And so I don't know that junior folks are gonna have access to a 100K budget unfortunately, right? So that advice is for the VPs, the folks that are worried. A lot of them are worried that I'm obsolete, right? That I'm obsolete, that I'm not gonna get that role at Vercel or OpenAI. Um, so keep, keep going. My... The advice for the junior folks is to embrace it. You, if you... Whatever tools your organization is using, become the best person at working with that agent, and you will automatically get more efficient. Yeah, is it annoying that, that you walk into work and the agent set up four calls for you and maybe you only wanted to do two of them? Embrace it. Embrace it, 'cause you'll be twice as productive. Right? I was literally talking yesterday, there's a company I'm on the board of called owner.com, which is kind of like AI for restaurants. They're crossing 100 million in, in revenue growing really quickly. He's got a hundred folks on the sales team, Kyle does. With AI, he's trying... He's targeting three to five million in revenue per rep. Three to five million. Honestly, if this was three or four years ago for a similar company, it would be 3 to 500K. That's an order of magnitude more efficiency. Still 100 reps, right? 100 reps. Uh, he's gonna need more to hit the number for next year, but three to five million per rep. So if you're that guy that can work with those tools, you become more valuable. But if you fight it, if you don't wanna do that extra meeting, if you fight it, um, there's a, there's a, there's a tool, another tool we use for RevOps.... there's two we use. The, the, one's called Momentum, one's called Attention. They're both great. They're very similar. And what it does is every single thing a human does is automatically tracked in your CRM instantly. Every comm- real time. There have been tools that have done some of this before, but literally everything. And, and when we rolled it out, and it happened with a few other folks I know, one of the folks on that old team quit that day. Quit the day we rolled out the, the AI rev ops. You know why? He hadn't done anything in 30 days. (laughs) The gig was up. (laughs)
7. LRLenny Rachitsky
  Oh, wow.
8. JLJason Lemkin
  Every day he would show up to our standup and he said, "Yeah, I'm doing outbound and I'm really working on that deal with Vercel."
9. LRLenny Rachitsky
  Oh, wow.
10. JLJason Lemkin
  And nothing would close. And then we said, "Oh, we're, we're..." and he, and he quit that day. So my point is, he didn't lean into it, right? Lean into it. Now you're gonna have full, total transparency on your day. You know, granola is a, you know, a side example for everything, but, but AI is gonna track everything you do. Like, if you wanna fight it, if you wanna fight the future, good luck to you. But embrace it, and you will, you will leap ahead of your peers that aren't embracing this. It will embre- embrace all the transparency, all the leads, all the work it makes you do, because you can't close 10 times as much doing the same amount of work. Can you?
11. LRLenny Rachitsky
  Nope.
12. JLJason Lemkin
  We do more work with AI. I'm working the hardest I've ever worked. That's with all these agents and all the output they create. And Amelia does a lot of it, but even I am working the hardest. But, but it, but it's better. But it's not less work, it's the ha- it's more work. The agents are so productive, you have to keep up. So I'd... So sorry to interrupt. For the, for the, for the managers, buy an agent, deploy it yourself. Don't have someone else do it. Do everything from training to ingestion to orchestration and those terms will be less scary to you. For the junior folks, be the guy, the gal, the person that loves these tools and that is the first person to embrace whatever it is. Don't fight it.
13. LRLenny Rachitsky
  Awesome. Okay. For founders in startups, what I'm hearing is the, um, be that forward deployed engineer, have that person, that sales engineer, that's sitting there helping companies. This is basically an opportunity to, to, uh, compete with bigger players where you're actually there helping them set things up. Like, all the companies you mentioned, none of them are big companies. They're all startups, which is really interesting.
14. JLJason Lemkin
  It is true. I will say it is... Working with Salesforce the way we are, it's the way I haven't worked with Salesforce since it was the first product I bought. I mean, 20 years ago, dude. Um, when I bought Salesforce 20 years ago, I got everything was hands-on. I bought two seats. I remember back in the day talking to my rep, I'm like, "Don't talk to me. You don't have time. You work at Salesforce. I'm buying two seats." He's like, "No, no, no. This is what we do. I'm here for you. I'm gonna help you get your app on the app exchange. I'm gonna do all of it." And that went away. It's kind of back, right? And Mark's got 2,000 folks doing this at Salesforce now, so we have a forward deployed engineer. I'm not even sure it makes economic sense and maybe in two years someone like you or I would not get an FDE, but, but, but the big companies are figuring it out. But yes, pick a vendor that will make you a success with, with an FDE, right? And, and, um, people throw this term out, but again, it's, it's just someone that's gonna make sure you get trained and onboarded with your product for real. But yeah, so star- startups have a motivation to do this in a way that maybe some of the incumbents really are still figuring out, right?
15. LRLenny Rachitsky
  Which is a big opportunity for startups to actually do this. And, and this actually resonates, uh, very, uh, closely to what Jen suggested for startups, which is sell services initially. Just like you work for them, solve their problem, not with software to get things started, to grow that into a massive contract with them that is software.
16. JLJason Lemkin
  You mean the, you mean land and expand, start with a smaller deal and grow it?
17. LRLenny Rachitsky
  Start, no. Like, like for like a person sitting there doing this work with them, essentially what you're describing. Don't just like, we have software, go try using it. It's like, we will have someone on our team helping you figure this out, and, and maybe there's not even software yet to do all these things, but we'll have them do it for you, and then the software will take on more and more of that work over time.
18. JLJason Lemkin
  I think that's always been great advice, and startups have always been great at this. Like in the early days, you... The CEO does it, is, is doing support. The CEO's onboarding you. I think what's different today is the agent will fail without training and onboarding this product. It, it, it will fail. Like, it will never, it will never work. So it, it becomes an imperative if you wanna win. Like, you need... I, I, I... You... This term's been bandied about too much, but you need a team of humans, we can call them forward deployed engineers, that make 100% sure that when the agent is turned on, it's awesome. That's your job as a founder, make sure it's awesome. Versus in the old days that... Even when I built, uh, uh, one of the first e-signature services that was so easy to use, we'd have some customers take them two years to go live. Big customers. That's just not okay in the age of AI,
1:05:40 – 1:08:08
Forward-deployed engineers
1. JLJason Lemkin
  right?
2. LRLenny Rachitsky
  Mm-hmm. Yeah. And these forward deployed engineers, is that just another name for sales engineer or is that a different sort of background often?
3. JLJason Lemkin
  It can be, and I think... All this nomenclature is confusing, man.
4. LRLenny Rachitsky
  Yeah.
5. JLJason Lemkin
  All this orchestration and ingestion, I think we're c- we, we're confu... Like, Palantir is obviously the single most successful public B2B company at the moment, right? Maybe, maybe Databricks and a few others will beat them in the next wave of IPOs. But their idea for forward deployed engineers is similar, but these are nine-figure deals. Most of us aren't doing $100 million deals. So the idea that you'll have Garry Tan and an army of folks out in the office for six months getting the software to work, that, that, that sort of inspires the idea. But it's really... Sure, it could be a sales engineer, a solution architect, but the difference is the SEs that a lot of us worked with in the days were resources. And often they... Like, th- there was also then a classic resource of like eight sales reps to one SE. So the eight reps would... There'd be a pod and one SE would be responsible and you'd kind of have to fight to get Jason or Lenny's time to help the eight reps. This is inverted. The FDE's number one job is to make the customer a success, okay? I was literally the other day doing a presentation, an AI leader that closed a $3 million deal. Um, the FDE did it all himself. Sales wasn't even involved in the deal. They went on site, they got the deployment going, they tuned it, everything. All sales did was manage it through the procurement process, okay? That's pretty different than a guy answering some questions for the humans, right? So it is a combination of customer success and SE or whatever, but it is, it is really o- I mean, frankly, it's just being a, a classic consultant that gets the product done on day one. On, on day one. That's the difference is that when you go live, it works.When you go live, it works. That's what that... And it's a fancy name for a bunch of folks on your team that (laughs) when it goes live, the AI agent actually works. So you have 100% success rate instead of, like, the 5% rate of 2024. So those people. But they do need to be tech- technical. I don't know if they need to be engineers. It can really vary based on, you know. I love it when they're kind of like mediocre engineers that are, like, s- in love with the product. That's my favorite type of FD. Like, their on- they don't really want to code much anymore, but they did code and they, like, just love your product. Um, but, uh, I don't think it could be someone with no product chops, but it could... All different folks can work but they- they've got to know the product cold. But, um, yeah, I would... Startups like... I mean, the term is thrown around. You need four folks that will just make sure that on go live the product works, your Agentic product works. That's
1:08:08 – 1:16:21
What's changing and what's staying the same in sales
1. JLJason Lemkin
  what you need.
2. LRLenny Rachitsky
  I think what might be useful to close out this conversation is to kind of go through what are the things that are changing, like maybe a handful of things that, that are changed that are now gonna be different in the world of sales go-to-market, and what are a few things that are just gonna stay the same?
3. JLJason Lemkin
  Yeah, let's go through it. Obviously, s- support is the first one to have changed. It's r- obviously permanently changed with AI, right? Whatever vendor you look at, 50 to 80% of support is done by AI. And we don't always think about support as GTM, but it is. It's the start of a customer journey. It's very important to the customer journey. So, so if you're skeptical, go look at, you know... Support has changed permanently. So that, that train's left the station. Really, a- as we record this, not much has changed in sales. I mean, I do think the stuff we've talking about is the bleeding edge. I do think the classic cadence-based SDR running campaigns through a, through a tool him or herself will be mostly extinct with 12 months. There is no reason AI can't do a better job than that role. The classic qualifier, qualifying inbound leads, which is a crappy experience for customers, should be similarly extinct within twel- mostly extinct, um, in 12 months. The rest we're, w- we're going to wait and see. I think what we do know for reps, for sales reps, everyone wants to be even more efficient in the next 12 months. There's a lot of reasons people want to be more... A- again, at the bottom end it's, it's, it's cost, it's profitability pressures. At the high end it's cultural. We just don't want 200 reps running around or sell a rep that don't know what the product does. Ev- Just, the, the company owner, three to five million per rep is a lot different than three to 500,000. So you have to adjust as a rep. You still have to still... Every AI leader can't hire enough reps, like we talked about. But you're gonna have to adjust to being geometrically, if not exponentially, more productive with help from AI. So you have to embrace these tools for real. And everyone, a lot of folks, a lot of old school GTM leaders are like, "You know, AI isn't gonna hurt sales reps. It's just gonna make them, give them superpowers." The best ones, yes. The mediocre are just gonna be, like, more mediocre (laughs) . Right? Um, so the, I think the AE is w- we will always have salespeople, but being a people person is not enough anymore. You know how you can tell a mediocre salesperson, Lenny?

Episode duration: 1:42:10

Install uListen for AI-powered chat & search across the full episode — Get Full Transcript

Transcript of episode I-R1bc1rlFs

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome

Introduction to Jason Lemkin

What SaaStr does

AI’s impact on sales teams

How SaaStr's AI agents work and their performance

How go-to-market is changing in the AI era

The future of SDRs, BDRs, and AEs in sales

Why leadership roles are safe

How to be in the 20% who thrive in the AI sales future

Why you shouldn't build your own AI tools

Specific AI agents and their applications

Challenges and learnings in AI deployment

Making AI-generated emails good (not just acceptable)

When humans still beat AI in sales

An overview of SaaStr's org

The role of human oversight in AI operations

Advice for salespeople and founders in the AI era

Forward-deployed engineers

What's changing and what's staying the same in sales

Get more out of YouTube videos.