SaaStr 831: How We Use 20+ AI Agents for Marketing & Go-to-Market with SaaStr's Chief AI Officer and CEO & Founder

26 Nov 2025 · 54 min

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

Podcast Summary: SaaStr 831 - How We Use 20+ AI Agents for Marketing & Go-to-Market

Podcast Overview

  • Title: SaaStr 831: How We Use 20+ AI Agents for Marketing & Go-to-Market
  • Hosts: Jason Lemkin (CEO and Founder of SaaStr) and Amelia Lerutte (Chief AI Officer at SaaStr)
  • Focus: Part two of a series on leveraging AI tools for marketing and go-to-market strategies in B2B/SaaS.

Key Themes and Discussions

Importance of AI in Marketing

  • AI Tools in Use: The hosts discuss the implementation of over 20 AI agents across various functionalities, particularly for marketing and sales strategies.
  • Enhancing Marketing Efforts: AI is described as not just automating tasks but improving the quality and volume of marketing efforts. Tools mentioned include:
  • Replit
  • Opus
  • Agent Force
  • Artisan
  • Qualified
  • Misconception of Automation: The hosts emphasize that while AI can automate up to 95% of tasks, it does not endorse lazy marketing practices. Effective marketing still requires human oversight.

Current State of Marketing with AI

  • Underutilization of AI in Marketing: Many organizations are behind in adopting AI for marketing compared to sales and support functions. Common uses include:
  • Messaging
  • Research
  • Creative analysis of marketing campaign data
  • Phase Two Use Cases: The episode stresses the need for marketers to engage in deeper analytics and creative uses of AI beyond basic functions.

AI Tools and Their Functions

  • Specialized Stacks: SaaStr has created a highly specialized stack for marketing, using tools that are hyper-focused on specific tasks.
  • Key Tools Discussed:
  • Gamma: For creating polished marketing presentations and collateral.
  • Reeve: For image generation, particularly useful for B2B marketing, avoiding generic outputs.
  • Claud and Opus: For content automation and video content generation.

Content Creation and Repurposing

  • Tripling Content Output: Jason remarks how AI helps triple output by enabling faster research and content creation.
  • Use Cases:
  • Extracting insights from multiple articles.
  • Creating compelling content by amplifying existing ideas.
  • Repurposing Strategy: Tools like Get Recall and Opus Pro are utilized to extend the life of content by generating clips and identifying key segments for social media.

Marketing Automation and Email Campaigns

  • Hacking Sales Tools for Marketing: Jason discusses adapting sales tools for marketing purposes, revealing a convergence in the functionalities of marketing and sales tools.
  • Targeted Email Campaigns: They discuss how personalized email strategies using sales tools have shown promising open rates and engagement.

Outlook for AI in Marketing

  • Future of Personalized Campaigns: The hosts express optimism about the future potential of AI to create highly personalized marketing experiences, including custom newsletters that adapt to user interactions.
  • Convergence of Tools: They predict a future where sales, marketing, and support tools will converge, allowing for a seamless experience across channels.

Key Takeaways

  • Investment in AI: Organizations must actively invest time and effort into leveraging AI tools for productivity.
  • No One-Size-Fits-All Solution: While many tools exist, finding the right combination that fits an organization’s unique needs is critical.
  • Orchestration of AI Agents: Effective use of AI requires human orchestration to ensure the quality and relevance of outputs.

Conclusion and Future Directions

  • The conversation emphasizes the importance of continual adaptation and integration of AI into marketing strategies. The hosts advocate for a proactive approach in leveraging AI technologies to enhance marketing effectiveness and establish more personalized customer interactions.

Episode Sponsors

  • Salesforce: Highlighting their platform for automating business processes for startups.
  • HappyFox: Showcasing their AI-first support stack with pre-built AI agents.

---

This summary provides a comprehensive overview of key concepts discussed in the podcast episode, focusing on the application of AI in marketing and sales strategies within the SaaS industry.

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:01Welcome to the official Sastr podcast where you can hear some of the best Sastr speakers. This is where the cloud meets. Up today on the Saster Podcast. All these agents are getting better. We have 20 in production. Here's the thing. This includes Replit. This includes Opus. This includes Agent Force and Artisan and Qualified, all the tools we love to death. They automate a lot of stuff. Like they might even automate 95 % of what you used to need a team to do. They don't automate everything. Here's the problem. As great as these tools are, they do not enable lazy marketing. What they enable is better marketing and much more of it.

0:45Everyone wants an autopilot. Everyone wants to buy a tool and disappear. It does not work that way. You can have crappy output. You can have crappy LinkedIn posts. Hey, everybody. It's Saster. Connect data, automate busywork, and empower teams like nobody's business with the one platform that grows with you. Every step of the way. Learn how Salesforce works for startups at salesforce.com slash SMB. That's salesforce.com slash SMB.

1:18Hey, Sasser, imagine having agents for every support task. One that triages tickets, another that catches duplicates, one that spots churn risk. That'd be pretty amazing, right? Happy Fox 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, Copilot, and Autopilot working as one. Check them out at happyfox.com slash faster.

1:52So thanks all for joining us here. We thought we would do a part two to what we started last week. If you didn't see it last week, that's okay. We'll cover some of it here and do a refresher. We primarily focused on sales and specifically outbound all of our AI SDR agents. So what I thought we would do today is start with marketing and then we'll kind of circle back to a few things we didn't get to in broader go-to-market. just in what's working now for us, at least, as our AI is running our go-to-market teams and we have more agents than people here at Sastr. So I thought we would go through that sort of there.

2:32One thing I'll preface for everyone, if you haven't seen this yet or you missed this last time, if you do go on sastr.ai slash agents, this will reference a lot of the things we talked about last week and this week in the two parts for go-to-market. it. So this is our agent arsenal, as it sounds. A lot of these tools are third-party tools. Some of them have a little bit of proprietary stuff in them. There's the proprietary apps we've vibe-coded. But this has the whole guide there. So if you're looking for certain tools we message or drop in or show screenshots of and you want to refer back to them, they're all there just to make it easy for you.

3:12So with that, let's dive into marketing a bit, just because we didn't get to it. And I feel like it's an important topic. And before we get more tactical with it, since we have a little bit of time, Jason, I want to get your thoughts on what you're seeing here as well. I think this is a study I referenced last week as well. This is from Rory and Skill venture partners is a state of AI go-to-market study that just came out. It's a pretty good one. I wanted to call out a few things in marketing. I think they're interesting to me slash us because I feel like for marketing in particular, folks seem to be more behind than they are in sales or support.

3:55I actually anecdotally don't know why that is. I don't think it's because there's a lack of tools, which a lot of folks will point to. I don't think of that as you'll see in our next few slides, there's plenty of tools we're using in marketing and that are working well. So I think it's just a little bit interesting here to see in the data that, you know, most folks are just using AI in marketing for things like messaging. Okay, fine, but that's everyone's done that now, right? Yeah, TBT. Doing research, I feel like, yep, that's a little bit more in the weeds, maybe a little bit more specialized, but most people have not done these phase, what they're calling phase two use cases, which is actually analyzing, you know, marketing campaign data and spend, whether it's in-house or with third party agencies, actually using it to be creative.

4:42I think, I don't know where this stigma has come from, but that the AI is not that creative or that all the AI content comes out the same. I feel like those are also two schools of thoughts that we've kind of myth busted on. And then for things like enriching and scoring. So as we talked about, a lot last week and just how we're running some of our go-to-market through these agents with sales in particular, a lot of that is still relevant to marketing. So I think that's maybe a conversation we can have at the end of, you know, what does this all mean for the two separately and then together? But I think it's something where so many of these platforms and AI, maybe they were built for sales, but you can actually use for marketing, which I'll show.

5:24All right. So let me show you guys a few things of what we mean. And then, yeah, So we can dive into it a bit more on what this might mean for you all. I would say right now in marketing, we have the most specialized stack. And what I mean by that is we have the most amount of tools in marketing, maybe that we have for sales. A lot of our agents that we've deployed in sales, I'm able to deploy multiple agents on single platforms. And so that's helped a lot in terms of how we train and tune all the agents has been different in each of those platforms and in each of those single instances. But I'd say for marketing, we probably have the most specialized stack right now.

6:02We have hyper-focused tools that do one thing really well. And so in this, it's been a little bit of trial and error of us trying to find, you know, what are the very best ones out there? And then not only which ones are the very best ones out there, but which ones will we actually stick with using and also use for our actual go-to-market, right? which ones are just okay we kind of had fun played around for the weekend on and then never used again which ones do we actually need now in our day-to-day so there's two i think that have become core for us it's both from a this one's both a sales and marketing perspective but i've used it for both these cases this is a really great one at making collateral so just go to market collateral whether that's sales collateral marketing collateral the two apps we use the most right now are going to be Gamma and Reeve.

6:58I think the next slide is Higgs field. We use those a lot heavily. I put this slide here. This is an actual file I sent Ripley. But this is a custom... You'll see that I picked this slide because I have a few custom photos on it. The photos were made by Reeve. So where you see these guys like vibe coding outside. I told Reeve, you know, hey, I'm thinking about doing like a vibe coding camp for Summit or something, Vibe Coding Lab. It also came up with that name for me. Vibe Coding Lab at Sasser Annual next year. This is basically a pitch I wanted to do with our marketing team. I gave it actual photos of Sasser Annual.

7:37I was like, here's what our setup looks like. Here's what the tents look like outside. Here's what people, you know, kind of look like in Sasser. So just so you have a grounding, I would say, for the agents who know what to output on. That also will help it not be as generic. I feel like where Reeve kind of kicks butt over Nano Banana or some of the other, or, you know, ChatGBT or Sora is, you can upload an image to those. And I feel like most people are just Ghibli themselves or maybe like a Pixar version of themselves. Reeve is really good at doing mock-ups. Like, I don't think it was intended to be a superpower.

8:15It's really good at doing stuff like this where, like, I used to have to ask her, desire okay can you do a quick mock-up of you know the replet logo in its head kind of make it look real like people are sitting in it one that takes really long when a manual designer does it two because i gave it reference images reeve was like this was a first shot image i was like that one's pretty good i was like i like that one it looks like our tent that we have outside there's people they're vibe coding on the screens i told it make sure they're vibe coding on the screen and so that's pretty cool and so i had to do a couple more i was like okay what are other things i want to conceptualize for this potential space that replet's going to do it's faster um i said okay i wanted to be an exclusive footprint so it came up with this you know here's the rest of the outdoor tent with the replet logo i did orange umbrellas because it pitched that from their brand colors that's prevalent in the logo and i said okay now the last thing i want is i want to include some sort of like outdoor stage and space i gave it photos of past outdoor stages we had with speakers on it and it came up with this and i was like cool i actually think it kind of crushed it on these mocks i wouldn't have been able to do otherwise and it's just such a way in which like when i started presenting this to the replete team they were like that's cool like they can visualize it they can see it yeah yeah just to add the inside tip for guys that might not you know for reeve.art it is a team that built their own image LLMs.

9:41It's not the same stuff. Everyone else uses, you know, there's a variety that they might use. But when, you know, when you build, when you build a lot of images and they all look like cartoons or they look silly or they look goofy, it's not your only asset for B2B. Try Reeve. Upload something that's similar to what you want. And in plain English, tell it what you want. I think for B2B use cases, for real business stuff, the 95 % chance it's going to be much better than what you get out of ChatGPT. I started using this. I told Amelia to use it. She didn't. And then it's just, this is the kind of real stuff that you want, not goofy cartoon looking stuff.

10:13Or, I mean, that has its place if you want to be fun, but for B2B use this and it's going to crush. Like this looks real. It looks great. There's no excuse for those, for that goofy cartoon stuff, unless that's what you want. And I don't think the, you know, the world has moved past that for B2B. You don't have to settle for crappy text, which we'll talk to you. And you don't have to settle for cartoony images. you in fact should create great pitches for your prospects yeah yeah so how this one came about in this use case was i yeah we've honestly had this journey with replet but then i started to get to know their marketing team they wanted more marketing and so they're a sponsor of our learning event already but we wanted they were like we want to do something bigger probably something custom at staffster annual and so we started thinking through this vibe coding lab think i would say reeve also is best at you can this logo i gave it that logo right like sometimes when you put it in the other generators or llms it will come up with a logo or it will change their logo reeve i think kind of quietly crushes it at the text being good and if you upload logos it keeps the logos so that's one where i to jason's point i really like the professionalism Some of it, it looks polished, right?

11:29So this is something that I think is tendable to folks. I like it for that use case. Another one is just related to marketing. You know, we're doing all these assets for Sastor London in 10 days or whatever this is. And I put in to read like my original concepts of, okay, here's what I'm kind of thinking for all the stages. You know, we need like some backdrops. We need a bunch of signage. and I had Reef giving you know all these different outputs I'm like now that one looks too dry that one is too overdone that one's too boring it's really good for storyboarding which I feel like it's a good superpower of it too because it's going to generate all these really realistic images I give it images of past London events like I gave it the venue I give photos of the venue like it can actually help you picture and visualize and storyboard what you want and so when I started going through that process it made this really cool like portal looking thing I was like oh I like that or like that plus the staff straight AI would look pretty sick and so then we actually we ended up I upscaled it in Adobe Photoshop as a little bit of AI so you can take it out of if it's you know it's fine enough digital quality but if you're doing anything print you do need to have usually higher quality higher DPI and so I put it into Photoshop upscale that gave it to our designer as a 4k asset all generated in read and I was like this is the portal you know clean it up, add SaaS or AI, and we got a pretty sick looking brand.

12:56That is also another great use case more specifically for marketing, kind of related to events, but even if you have a small event, this is a good way to kind of quietly up-level it as well and make it look more polished and better. I've used it in emails, right? Obviously, image generation for emails that we're sending out. But just in go-to-market, if you want it to be more tactical, we're using it here for both sales and marketing collateral that we're using to drive revenue so it is something that works really well i then plugged these three images into gamma gamma if you don't know it is a it's a i want to say newer ish tool that does content creation specifically presentations so like you might say oh i could do a presentation cam but yeah you can and like i've done both and I run tests on both side by side.

13:45Usually the gammas are just a bit better, especially for B2B rather than B2C like Canva is. The decks come out a little bit better. You can give it things like, again, I gave it Replets Orange. I gave it... I was like, check their website for fonts that you should use. Also look up how they talk about themselves. Because you can give it the content you want, but you can also prompt it. Usually the output is pretty good. And so I like Gamma a lot because we use it now to send all of our sales collateral, sales deck, support decks. We've Gamma'd. It's easy. They're also easily shareable. So between the two tools together, you'll have to still customize the two.

14:30I don't use all the image creations out of Gamma. Gamma has built in image creation. And I obviously still prefer the Reave images or ones that we have. that are actually real images. We still pick and choose which ones work better. So that's what I mean. Our marketing stack is probably the most specialized. But it does help eliminate a lot of bottleneck and deal cycles. You're getting them something polished. You're not waiting a week for your designer to come in and do this. Personalized decks, I feel like, just elicit a much better conversation or response from folks. This also, not only can this kind of up-level your decks, So you can also send them to more people is what I found.

15:13We started sending these to pretty much everyone. We used to not do that. We used to be like, here's one set deck we have for Saster. If you're interested in sponsoring, here's the one deck. Sometimes that's not the best thing for teams to pass around when they're like, okay, we're considering opting into this. Or if you're trying to go to market with something, more so on the marketing side, again, it's not clean. It's not unspolished. Just you're like one set that's more generalized, obviously versus something that's hyper personalized to them. You've obviously thought you've listened to their names, but using the AI, you can kind of scale that up a bit more so that you can have this sound to cross.

15:50It looks a lot better. You don't need an agency to do this. And then they can share it internally. If you use it directly in Gamma versus exporting it to a PDF or Google Class, it'll tell you whatever people view it. So I'll get pings all the time. I'm a random XYZ person at XYZ company. Do this deck. I'm like, yeah, I didn't send it to them, but they're at the company and they're always like sharing this around. So I think that's what you want is you want to use this to not only improve your current processes of how you're doing these things, but also just to unlock new ones so that you can do it a more hyper-personalized scale.

16:24And for folks that, I mean, Gamma got a lot of, we've been using it for a long time. It got a lot of press recently. They just said they crossed a hundred million AR this year and raised around it two billion. So a lot of press, but I've actually found we've been promoting Gamma for a long time because we use it. A lot of folks, they may not even talk about it. They've never used it. Literally, I was at the other day at an event with the CEO of a public SaaS company saying, ah, Gamma, it's just PowerPoint. And no one will ever use that. I'm like, you know, they're going to, they already got to a hundred million, my friend.

16:53There's something there. But here's the tip. Okay. Most of you haven't used reeve.art for images. It's all on SaaS for diet agents. Go upload something that you use and just ask it to make it better. I bet you're going to find this is your best tool. For Gamma, if you haven't used it, don't overcomplicate this. Literally go into a Google Doc, write five bullets of what you want a deck to be, okay? Take those five bullets, put it in the free version of Gamma, and watch it build a deck for you. I bet it's the best thing you've seen, okay? You can use it. You can make a template you use again and again, which Amelia does, and then upload it dynamically with data.

17:28Become an AI agent expert. But that could be intimidating for folks. If you haven't used it, don't overcomplicate it. take five bullets on anything that you're doing give it to gamma and watch the agent build a deck for you it's pretty awesome and then if you get excited about that go deeper like we have and then you can make dynamic decks for everybody for real not the same crappy static decks everyone else but you can start really simple with these tools they don't have to be intimidating and i think your jaw will drop for both if you invest like five minutes in each keep it simple invest five minutes.

17:58Yep. Okay. Let me go to the next one. In terms of content and automation, so we have a lot of specialized tools in this bucket. You saw on one of the previous slides, this is where most folks are using AI in their marketing stack. I would say, I don't think it's necessarily a bad thing that these are all specialized. Again, like the way that I use gamut and read together it's like it only takes i would say 60 extra seconds to generate an image and read versus doing a native brain gamma that a lot better in my opinion comes out a bit more customized and i port it over so i don't i don't think you should be scared by the amount of tools on this one side we do pick and choose like the very best things and outputs and also a lot of these are off the shelf so i would say for particularly in marketing and how we're using this so far i think the next one is one of our more like enterprise tools and solutions is you know these are all off the shelf so like claude replet reeve higgs field recall which we use for locker content opus we're adding mosaic for video like these are all ones where they're like you know user subscriptions like anybody on your team slash yourself could just go sign up you know whatever, 20 to 30 bucks a month or 10 to 30 bucks a month and try these tools and see which ones work for you.

19:26So I think what's working well for us now is, yes, obviously we do use it for some of our content. We have a lot of content at Souser. And prior to this, that was a lot of Jason and my day was writing content. It's still a lot of Jason's day. It's still a lot of our time. But now the agents, I don't want me to speak for you, Jason, But I do feel like with this agent mix you have, specifically in content creation, you get there a lot faster. So you're able to post more advice to folks. Yeah, I would say what we've been able to do is triple our content output with AI. I think that, and I think the mistake a lot of folks make is they use a lot of crappy social media tools that regurgitate content again.

20:10So they want thought leadership. So they spool up a tool. It examines someone else's feed. It often examines my feed. I see folks rewriting my content and it sort of works because LLMs are good at copying and regurgitating. So if you just want views, I guess, if you think that's going to move the needle for your business, then generic AI tools are fine. Okay. But what we do, which is the real powerful thing is, is you architect the content, you create the content, our videos, our sessions, our posts, our ideas, and use AI to turbocharge It should do more research for you to go deep dive. You know, we used to, that's what you really want to do is to, is for AI to be your steroids for content creation, not to create the core nucleus that becomes trite and boring, but that's how we've tripled it.

21:01I, I write all the, if you've read Sastra for a long time, I've written most of the content for more than a decade. Okay. But if you compare an article today, this year versus two years ago, you'll see that it's pretty similar, but there's a lot more data. I can do a lot more research. And it's not just asking Claude to do the research. I can take multiple articles, three or four articles off the internet. For example, today, just before this, I wrote an article on how venture capital has changed for funds. I took two different data sources, one from Crunch Race and one from an LP, an investor in top funds called Venrock that analyzed 1900 funds.

21:38And I added that data to my post. Okay. I added that data with AI and it makes it so much better. You've got to check the work. It does take a few minutes, but AI can let you take great ideas and make them 10 times better. That's where you get, get the leverage. Yep. I would say too, just on this here, we do use a few different tools here. They all work fairly well, but yeah, to, to that point, right, We are using it to amplify existing content that we have or even build up on that content. Like one thing I did for this webinar, actually, for last week's and this week is I gave it the last three or four sessions that we had done on this topic of AI, our agents.

22:22I put it into Clive because I thought better. I was like, hey, given what we've already covered, I don't want to keep regurgitating things. Like I want to show data. and Claude did not have my data. And I would like, these are screenshots I took. I put these slides together, but it gave me a pretty good outline. It was like, okay, based on the last few talks you guys have done, based on what you have coming up, based on where you're at the six month mark, here's what I think you should cover. That is like SaaS or style actionable content. So it gave me a really good outline. It like gave me this outline, like this format of like, okay, you should go into each of the different segments of go to market.

22:59We should talk about what's working and what's not on each slide. You should call out the tools. And it gave me ideas. I'm like, okay, here are maybe some answers. Like it would ask me like, hey, is this what's working? Is this not what's working? Okay, like then maybe show this as like the next slide. And so it gave me a really good, it was like a really good co-pilot for me. I'm like, just putting my thoughts together because obviously now that we're doing a part two, I was like, I have a lot we could cover and go off on about AI and go to market. But I want to try and make it super tactical for folks so that it's helpful and useful.

23:31And it was really good for holding that. And it was like, okay, I was like, what about this topic? It was like, no, like that could be its own topic. Don't go too much off the engines. And so, yeah, it's gotten, I would say really good at that. So I'll do this outline. And then I obviously did, I still wrote all the slides and did all the, I could use Gamble a bit. And then I pulled all of our screenshots and of our data. And, but now I'm also using it for some of the stuff we're going to do in London. And again, I'm going to give it this one after today. I gave it last week. I'm like, okay, now build on it again, knowing that some people in my audience would have seen these two faster AI lives and some people wouldn't have.

24:10But also make it shorter because now we're together. This will be like the two-hour mark. And I think the opener at London is only 45 minutes. So what are the most valuable points I should hit on? And it's gone pretty good at that. So I think it's great for that too because that saved me a lot of time. Like it still took me a while to obviously make the stack and put all the points together for you guys and put all the screenshots and the data like that, that still takes work. But it did help me organize my thoughts a lot more. Yep. On the content repurposing, not to spend, I know you hit it. A lot of folks, especially founders and marketers that are watching or will watch later.

24:52I find we're all generating so much more content these days, especially video. There's so much more pods. There's so much more everything going on. And people are not using AI to extend the life of that content. There's a lot of disgust on it, but I see a lot of lazy approaches. And let me tell you what we do. It's pretty simple. Amelia has it on this bullet point, Opus Pro, Get, Recall, and Descript. I don't think we touched on the first two, did we, Amelia? No. But okay, so you just did a great interview. and then you stick it on your YouTube and it disappears. Okay, great. I hope it performs.

25:30Just a few, the first one's a little, you probably know, or you might not. The second one's a little more subtle, but let me tell you what we can do with this content. The first one is stick your, if you haven't used Opus Pro, it's on our list on Sastre Agents. There's a lot of tools, but it's the leading tool for turning your video into clips. And listen, it can't take boring content and make it great. But what Opus Pro is really good is take the YouTube URL, stick give it to opus clip and it will force rank the most compelling content out of that and find you some clips and then it will just actually automatically schedule it for you on your social media accounts on linkedin on x on facebook whatever you want it's pretty good it will take the if you just did a 40 minute webinar it will take the best 60 seconds 120 seconds clip it and let you put it right on linkedin where your customers are you should be doing this and you can queue up five, six, eight, 10 of those over the course of the month.

26:21You should be doing that with your best content or you're kind of wasting it. How does it grade it? It uses AI. It's pretty amazing. And there's a free version. This is another one you can literally try in five minutes. Take your YouTube URL and stick it in Opus Pro and see it. It extracts the text, which is actually very easy to do. Actually, YouTube gives you the text through the API. It's not that complicated. And then it just runs that text through some LLM. I don't even know what it is and says, find me the best parts. When Opus first launched pre, I don't know, we were using it in SAS for London a year and a half ago.

26:54It was like magical. I'm like, how does it do this? How does it find the best clips? This seemed like the most magical. And I actually invested in the company. It was so magical. It still is magical, but now that I understand how AI works better today with our agents, it's less magical than I thought. Extract text. It's not much text. Like these videos, even if it's two hours, it's not a lot of text. It's not much for the LMS to process. Give it to the LLM and say, break it up into the best two minute chunks. It works really well. And then just, then it just says, hook in your LinkedIn account, hooked in your Twitter account.

27:22And it just does it all for you. It's magical. So you're wasting your content. If you don't do this, you could also ask it to make three minute clips, which we do too. If you have a little more energy, you can put those three minute clips on X and LinkedIn, like the best ones for deeper dives. You'll see Sastra do that all the time. You'll see these three and four minute clips. I don't have time to make it. Opus makes it for us. Like it's really good. Okay. So what's the next? So that's how you turn your video into clips and short thing. Just do it. Opus is like, there's a free version or it's 20 bucks a month or five bucks a month for the paid, but you can try it free.

27:53Then we use an app. Very few people use called get recall. There's actually two apps called recall. One does like transcription. I don't know. One is an API of tons of folks use the founders at Saster and you'll got a ton of customers out of it's an API get recall actually just transcribes your whole internet world with AI. Like everyone should be using this. It's just a niche app. It'll do everything. I use it for one thing. I give it that URL for the YouTube and it instantly gives me great summarized text. And then instead of Opus doing its own thing, I take that text and I put it into Claude and I tell it the type of article I want to write out of the content.

28:30This is really powerful. Okay. So literally we're doing this hour long webinar right now. Okay. When it's done, I will grab the YouTube URL and I'll go to get recall and give it to it and I'll get the text and I'll put it in Claude. And let's say I only want to write a Sastra post on content repurposing. I will take the transcript of this session and say, Claude, write me a Sastra post just on what Amelia and I said on content repurposing. And then it can write an article just on that. Now, of course, you've got to edit it a little bit, but it's going to be pretty good. This is, and literally if we wanted to, we could take this session and turn it to five or six articles by getting the text from it, from GetRecall and YouTube, and then just asking Claude to write five or six great articles out of it.

Read the full transcript

29:13This will be better than the crummy agency you asked to write a bunch of crappy articles. Take your own content. We watch that. If you have something great from your own events, digital events, world events, find the three or four great themes from it and write articles just about those three or four things with the extracted text and Claude, it'll be great. Like it will be great. And this will return. You can get 10 great clips out of your content, but actually you can get five or six great articles. If it's a really good piece of content, don't waste that great interview. Don't waste that. The best keynote you had don't waste the best session from, from an event you just had.

29:51It's just a waste, right? I mean, no light criticism, no criticism. I just did this great digital day with G2. We did it. Goddard and I did it with the CEO of Zendesk. Okay. It was so good. Now they've decided to hide it behind some WISC and we do love WISC at SAS, we use it ourselves, but to hide it behind a link wall to capture email addresses, I guess, maybe that's the right thing to do. But because I know so much about AI for support, I asked the CEO of Zendesk so many great things. I could write five articles out of that piece of content. We talked about pricing for AI for support. We talked about how difficult it is to train AI agents.

30:29And so how to simplify that. That could be like four different articles that G2 and I, if anyone from G2 wants to help, we'll do the work. We could create five great pieces out of that interview instead of it like just disappearing into the ether.

30:47The other thing I was saying, as you were chatting through all this, a few folks are asking questions related to if there's a way to do this more programmatically in the chat. I think what you guys mean, if we're using, you know, all like, Jason, for this, let's just choose this content example you just gave. Is there a way you can do that automatically or are you doing that manually now? Do which part manually? You know, is there, there's no way right now to automate like Opus grabbing or recall grabbing YouTube automatically. It's a great question. All these agents are getting better. We have 20 in production.

31:26Here's the thing. This includes Replit. This includes Opus. this includes Agent Force and Artisan and Qualified, all the tools we love to death. They automate a lot of stuff. Like they might even automate 95 % of what you used to need a team to do. They don't automate everything. Here's the problem. As great as these tools are, they do not enable lazy marketing. What they enable is better marketing and much more of it. Everyone wants an autopilot. Everyone wants to buy a tool and disappear. It does not work that way. You can have crappy output. You can have crappy LinkedIn posts. Again, where I started, none of this is full automation.

32:10You can automate 95%, but even me, listen, I'm not like, you know, I'm no, I didn't found a lovable or, or cursor, but I'm a somewhat successful founder. I'm still doing this myself at this point in my career. I'm not, I, but I'm getting lots of help. So I'm much more productive, but I still have, you know, this term maybe is over used. I'm still orchestrating these agents and Amelia is still orchestrating her agents. So if you can't put in 20 or 30 minutes a day to do this, don't start the project. Okay. If you want to do nothing, we can't help you. If you want to invest 20 or 30 minutes of actually kind of fun stuff.

32:46I mean, this stuff is fun. These tools are magical. If you want to invest 20, 30 minutes a day so that you could have massive output, use this playbook, but there's nothing you can just click and forget. There is, but it's not going to be good enough. If you're not going to read the text, if you're not going to decide how to take that interview with the CEO of Zendesk and turn it into five posts, it's not going to be great. It's going to be generic. Yeah, there's a few questions in the audience on how we're using it, maybe more so for things like marketing automation style email. So I'll go to this next slide.

33:19when it comes to email marketing i am full disclosure using sales tools to check the books email marketing at rgcm now in november of 2025 or whenever you're listening to this once there is a truly great all-in-one ai marketing platform i will try it i will use it but it doesn't exist yet. So I have hacked the existing sales tool to use them for marketing, but with fairly good results. Now, what we're not able to automate yet, but we actually are starting to get there and starting to kind of like lightly prototype things and replicate is I haven't solved for automating our newsletters. We do these four to five newsletters a week on, it aggregates all this extra content, It's all this really great original content that you're seeing us outputting here.

34:14We put in the newsletters, all of Jason's articles, everything we're writing up on AI, you know, our agents page, everything in there goes into those newsletters. We're still doing that manually for all intents and purposes. So even in today, six months down our AI journey, I haven't found something truly great to automate that process. I don't think it exists yet, but if you know one we should try, drop it in the chat. There's still a gap. Yeah, there's a gap. Automating, making those emails that our newsletter is great isn't yet AI-able completely.

34:51We can come back to that. We can go back to this the next time we do this. Yeah. So it's like early. On Sastr.ai, our new platform, you can actually sign up for a different newsletter, an automated newsletter that shows you stock quotes and news all across AI and the web and everything. that one is fully automated and custom. So I can see it into the future. Like it can do things our tools can't. For example, it extracts the latest Saster video. Like the minute we're off this one, it will go into that automated newsletter with no work. That's so cool. And it will figure out what the best articles are on Saster today, which AI is good at, and customize it to the newsletter.

35:32But the full voice that Amelia does in the newsletter where we exert it and we talk about those learnings that we make it personal. AI is not there yet today, but I wouldn't be surprised. I think we'll find a tool sometime in the next three or four months that is somewhere in the middle, that it can extract enough stuff that if you're not doing good newsletters, it's much better than nothing, right? Because we do have these automated ones on the side, but it's not there yet. What is interesting on marketing as you dig in is you'll find, so much of the energy in AI that we see is in, coding tools, obviously, sales tools, and like social media tools, but a lot of marketing that's still early.

36:15You'd think there would be replacements for a lot of things. I think there will be next year and there's a lot of startups, but as soon as we find something great, we'll do the next, we'll do the next one with even more tools. But right now we're using, in many cases, these are point tools, right? And they don't do as much as we'd love them to do. And we'll, We'll check in 90 days. Yep. Specifically when it comes to email, what I'm doing right now is kind of hacking existing sales tools. While we haven't found a way yet to automate the newsletters, but where I have found some success is actually just hacking the existing AI, I'll call them go-to-market tools, even though they're labeled a sales tool, to send marketing emails.

37:00And so that's what I started, you know, the webinar with the practice of, you know, I think some of this is starting to converge and maybe we'll see more of that. But also, I think we may start to see change in how people interact with emails altogether. So it may not matter in the end that it's all kind of under go to market versus sales emails and marketing email. I'm already starting to see that. I don't actually care anymore that some of these emails that I'm sending for marketing are in sales polls. I don't care. I just think of them as go-to-market tools now. And this is where actually these go-to-market tools are pretty good at this because they were built for sales and because they're built to re-engage people.

37:41Because sequences is basically a marketing drip campaign. So just think about that. If you're doing sequences on an AI tool, you can probably do a marketing drip campaign. Those are the same fundamentals in how the AI is set up and how you want to set up. But instead of you giving it just, you know, sales prospects that you want it to keep sequencing, you give it a subset of your marketing contacts and say, hey, okay, these are specific folks. Maybe I've narrowed them down to, you know, you can't do everybody in your database now. It doesn't really work well. And it can't output that quickly, depending on how big your database is.

38:17So ours is really big. I can't email everybody in the database even with our 20 agents today but what i can do is i can wheel down some of the folks so what i'll do now is like this one that's here this is a screenshot i think from qualified which again qualified is for all this purposes a sales tool but i've been using it for marketing so i'll see in marketo because it's also hooked up it's hooked up to our marketo and it's hooked up to our salesforce it will see when people are opening emails and if they reach for a certain threshold old right now for they've opened a bunch of London emails but haven't bought a ticket.

38:52It will send them this one-to-one email. Now, I'm also still sending them the email blast of, hey, Sasha London's about to sell out. I'm still doing that, but I'm using the agent to actually do one-to-one personalized marketing email for this person. I don't know where Matthew works. I think it says here, actually. But this kind of called some of that up. Okay, I noticed you've been exploring content. It actually pulled up content that they were reading on saster.com. I didn't tell it to do that. It just did that. It contextualized it based on his heart interactions, saw what he was reading on saster.com, related it back to his company, knew that he was in this campaign where he was opening our London emails, but hadn't taken an action and offered him a discount code to saster.

39:38This is where I feel like there is a somewhat of a evolution in marketing and a convergence in go-to-market that why can't you have more specialized emails in marketing like this? Not to replace newsletters, not to replace email blast necessarily, but to be an added layer because you can get leverage out of AI to have a more customized email to this person once they meet a certain criteria. And basically I treat these marketing emails, again, as drip campaigns. And so far they've been working really well. This is fairly early days. I put a screenshot here. This was like 3 ,000 emails I'd sent in this one campaign just on qualified.

40:19That's not counting any of the inbound conversations that had or inbound people it's following up with. This was just for this specific campaign of people who met this criteria. You'll see here, open rates fairly high, right? This is on par with what we've seen in our agent force. Some of these people also go into agent force. I'm doing like an A-B test, right? I'm split testing them just to see which one works better right now. About the same, which is great. Like it's high since they're about the same. But we're getting fairly good open rates. And then the click-through rates will be, you know, obviously a little bit lower.

40:54That's fine. I don't expect that many replies also on this email. It's giving them a link. They don't really need to reply. If they buy a ticket, they have the campaign. Cool. if they don't talk to me in this interaction they've only talked to my agent that is a okay you know bounce rate fairly low so early days on this one because we've just started to add this into the mix but this is just such an interesting thing just for all the background things of you know it being this is technically a sales tool but you know sales tools can do really good sequences so why I was like well I couldn't do a really good drip campaign if it has all the right inputs And so this is where I found kind of a way to fill that gap that Jason was talking about in the meantime, in the interim.

41:39This is what I'm doing right now to still get good output in marketing and kind of fill that gap for now. Yeah, I want to keep getting through the content for one more time. But we hear a lot of folks when we talk about our AI stars saying, oh, no, those are marketing tools, right? Their AI is leading to convergence between sales and marketing. But in fact, it is even leading at some level to convergence between sales, marketing and support as the agents can do more. They the lines get blurry. So today that means Amelia is hacking what is nominally a sales tool, I guess. Right. To do marketing emails, to do personalized marketing emails instead of a marketing tool that we said of Marketo or HubSpot.

42:28Right. That's a hack today. It feels like a hack, but it won't be a hack next year. These will converge. And like, I do know I've done a lot of investments in the SaaS side of e-commerce where there's high volume and already tons of AI and already there, there is no difference between sales, marketing, and support for a variety of reasons. We're going to see the same thing because imagine you want to go buy a necklace and you go to a website. You don't want to talk to support versus sales and you, and marketing might need to market to you instantly. So you get a discount so you don't leave. And so these are all blending.

43:00Your skills are going to need to blend and your tools are going to blend. So that's a, for right now, hack the tools that you buy. Don't. And if it's a sales tool and you think it's marketing or it's marketing, you think it's sales, that's a new world. These lines are going to get blurry. I'll say the interesting thing is too, is because it's blurry, it's almost better. for the output of it it's actually better for me that some of this is in a quote-unquote sales tool now i just call it a go-to-market goal because it already knows all these things right i'm like okay art qualified if they're on the website it already knows that like i don't have to be like okay i have to you know back in the day to put the market munchkins on everything to see if they went to your website i'm like qualifying already knows that artisan does that too like the Salesforce knows, you know, and Agent Force knows everything your Salesforce knows.

43:53Like, if you already have all this really great data and you're kind of using these go-to-market tools that already have, again, they were built for sales, but really can be easily hacked and applied to go-to-market, I actually think it's better. I don't mind that it's merging and converging. I think it's leading us to send better marketing emails. again because we're sending so many hyper personalized ones we all have like higher quality conversations with folks too like even when they do reply this one's low obviously artists like artists higher asian persons higher on replies but this one also has a link right so i'm not comparing apples to apples directly because those two are more where we've programmed them where we want them to reply this one gives them a link so they don't necessarily need to reply but the replies i do get just across the board are just so more hyper personalized hyper custom the thing i run into actors and issues i sometimes can't reply to everyone like directly like sometimes they'll be like hey i already bought my ticket but hey what do you think about xyz tool you guys mentioned on the webinar or this or that and then like i'll answer once and then they ask them what's your follow-up questions i'm like oh no real amelia cannot keep up with all of amelia's agents and so that's where actually my biggest fit follow-up is not being able to keep up with all of our agents.

45:10It's not, you know, again, finding the right tools aside and maybe hacking tools for now, not the biggest problem we have. So that's an interesting nuance too, but I don't, I actually don't mind that these are converging. I think it's for the best. I think we'll see it lead to maybe good changes within marketing where, you know, you mentioned newsletters. I would love a really great tool for that, but I also see where, you know, I'd love a tool actually that in a you know in a six month state i could see this happening uh it's a newsletter but it's hyper customized right like why can't a newsletter be one-to-one in the same way that this email is one-to-one it should be yeah it should be right like the way jason was mentioning millions of emails but we should they should be customized they should be yeah okay if you're reading sas .com and i see you already read that article i'm not going to put it in your next newsletter like you already saw it i'm going to put something else related to it that's like a follow-up one like that's where I see this all kind of converging and going.

46:07It doesn't quite exist there yet, but I'm actually excited for it to get there because I feel like actually it will be better to virtually adding value. I feel like this is a good thing. Yeah. I mean, when we have AI for both newsletters and drip marketing, that's what we need is for drip marketing. So if we know whatever types of content you've interacted with, it's not very difficult for AI to customize the further content or the further campaigns you get. Well, there may be tools out there that do it or claim it, but the capabilities are limited, but it's not hard. This should be there soon is your every journey should be customized by AI.

46:42There's no reason for it not to be in every newsletter should be. I mean, we forget about your drip. No one should be sending the same drips, even a segment of drip. No one should be sending like two or three segments of drips. There should be a million drips that draw from all of your content, like dynamically. And our newsletters, we should be sending a million different newsletters, not three different newsletters. We know what you like. If you're a sales professional, maybe you don't want to read about Jason's vibe coding anymore. Maybe it's driving you nuts. If you're really into our AI content, maybe you don't want to hear about all the headaches hiring a VP of sales.

47:14You just don't care, but everyone's getting this mix of content of go-to-market AI, VC funding, and AI should be smart enough to change all that. And then we're all going to see our response rates go way up we're all going to see it go way up we're looking for that one we're looking we're in the hunt we're in the hunt for that we're in the hunt for that i'll do a quick i'll do a quick closing thought and then jason let's hit anything you miss and then we can do some q on a for a few minutes over this is interesting i was at like a salesforce event yesterday up in oakland and just with folks that i was talking to is you know one because folks know we have multiple ages now and i kind of talked about the story they're like are you ever worried that like your agents don't talk to each other and i was like no because even though our stamp is hyper personalized and specialized like you just saw like i use qualified for some things, artisan for others.

48:13Our marketing staff we just showed is so disparate of 10 different rules. What's the staff? We have a Jason AI that's Delphi. We have an Amelia AI that's qualified. I'm literally doing a voice and video test with them later. I have a London agent I'm rolling out on Agent Force where you can call it and ask it what sessions to go to. That's freaking cool. And someone was like, aren't you worried that these agents won't talk to each other? I'm like okay let me give you an example of a shift example where I had a worse interaction than talking to maybe one to three agents like I had to call Verizon the other day because the life I had the office was down and the experience was so horrible like I had one I had a call this number there was no chat there was no like AI troubleshooting of hey I saw your you know internet's down and i'm logged into my account you should know my internet are you down by the time i get there and so i got routed to five different people because we had a verizon for business account and it was specialized and it was i don't know what they called fios or whatever and it was you know whatever it was specialized and so i kept getting bounced to all these different departments and finally i think the screenshot on my phone was like 45 minutes later the guy who eventually came on said oh yeah there's just an outage in your area so your internet's down i go great you could have used ai to tell me this when i logged in you could have made the email me that there was an issue in my area i would have wasted 45 minutes on the phone but my point in all this is i talked to five different humans and it was a much worse experience than somebody who has bounced between maybe one or two of our agents or at the max maybe three or four of our agents.

50:01And I will say I know that because in our agents, they do point to each other. So if you ask our Amelia AI right now for SaaS advice or like how to hire a VP of sales, it will kick you back over to the JSON agent. And if you come back and you're like, hey, I actually had a question about London, she'll remember, like Amelia AI will remember that she could use the JSON agent and like just pick up on that conversation. You know, equated to at least I wasn't transferred to a new customer service rep. If you go back to, if you're kicked back to any of our agents, they remember everything. Like they remember your whole conversation, context.

50:36They can see what you were doing and actually get you to the right answer versus, you know, being bounced around by humans. That's still worse, right? It's just an interesting nuance there. Something I hadn't really thought of until somebody asked me, I was like, but this experience is still better than the crappy experience I just had with the bunch of random human beings who actually could not sell by.

51:03Tell me the internet's down. You know, the Verizon story is a great one. Let me just tie it together. You know, the, if we're talking about GTM, right, we don't talk a lot about support, but if support is pretty good, it can be a great lead capture, an opportunity capture tool. qualified you said for example it i mean it does a limited amount of support when you're using that agent it captures leads when it does the same thing right it does the whole thing and so these agents you know i had two bumps myself i had one with mercury which i love which is a banking product and i did it the other day and it said it would take them a day to respond to my issue now you could criticize that you could say why the hell can't you just have an ai agent do the support But when you have that experience, you're missing a chance to capture that lead and get someone into your journey.

51:56If Mercury or Brex, which we both use and appreciate and love, if they all don't really have AI support or can't get back to you, add a basic agent like we started with Delphi. Add something that adds value to your customers today. Put it on your site today. Like some of these tools take a long time to deploy. Delphi you can do in a day. you'll instantly start at least improving experience and capturing some leads and potential prospects support as these things converge marketers and sales folks don't think about support it will become a lead generation tool for you and it already is with our agents it already is they're blurring so make your if you don't make your support great too and your marketing sales will be better thanks everybody we'll see everybody in london we'll do this live we'll do this q a live you can come with all your questions and talk to you guys soon See you there.

53:18Hey, Sasser, imagine having agents for every support task. One that triages tickets, another that catches duplicates, one that spots churn risk. 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 that sits inside the HappyFox Omnichannel AI First support stack. Chatbot, Copilot, and Autopilot working as one. Check them out at happyfox.com slash daster.

From the publisher

SaaStr 831: How We Use 20+ AI Agents for Marketing & Go-to-Market with SaaStr's Chief AI Officer and CEO & Founder

Join us for part two of our series on leveraging AI tools for Go-To-Market in B2B/SaaS. In this episode, SaaStr's CEO and Founder Jason Lemkin, and SaaStr's Chief AI Officer, Amelia Lerutte discuss the importance and current state of AI in marketing, highlighting their key tech stack and tools.   They share their experiences and insights in utilizing these tools for creating personalized marketing collateral, automating content repurposing, and optimizing email campaigns with AI. Learn about the convergence of sales and marketing tools, the future of AI in go-to-market strategies, and how specialized AI agents can dramatically improve your outreach, lead generation, and content creation processes.  

---------------------

This episode is Sponsored in part by Salesforce:

Connect data, automate busywork and empower teams like nobody's business with the one platform that grows with you, every step of the way. Learn how Salesforce works for Startups at salesforce.com/smb.

 ---------------------

This episode is Sponsored in part by HappyFox:

Imagine having AI agents for every support task — one that triages tickets, another that catches duplicates, one that spots churn risks. 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 2 cents per successful action. All of it sits inside the HappyFox omnichannel, AI-first support stack — Chatbot, Copilot, and Autopilot working as one. Check them out at happyfox.com/saastr

---------------------

 

Hey everybody, the biggest B2B + AI event of the year will be back - SaaStr AI in the SF Bay Area, aka the SaaStr Annual, will be back in May 2026. 

With 68% VP-level and above, 36% CEOs and founders and a growing 25% AI-first professional, this is the very best of the best S-tier attendees and decision makers that come to SaaStr each year.  

But here's the reality, folks: the longer you wait, the higher ticket prices can get. Early bird tickets are available now, but once they're gone, you'll pay hundreds more so don't wait. 

Lock in your spot today by going to podcast.saastrannual.com to get my exclusive discount SaaStr AI SF 2026. We'll see you there.

More from The Official SaaStr Podcast: SaaS | Founders | Investors

All 68 episodes
SaaStr 831: How We Use 20+ AI Agents for Marketing & Go-to-Market with SaaStr's Chief AI Officer and CEO & FounderThe Official SaaStr Podcast: SaaS | Founders | Investors · 54 min
Listen in VO