← Marketing Against The Grain
Marketing Against The Grain
The AI Strategy That Doubled His Email Conversion Rate (Step By Step)
00:37:45
Loading summary
Transcript89 lines
- [00:01]
Kieran
Hey everyone, if you are looking to double your conversion rates from email, we got you on this show. We're going to tell you how to do that all with AI. We are then going to tell you about an email newsletter that's full of the best AI news that nobody's heard of, that's going to transform how you learn about AI. And then we're going to show you how the future of content in an AI world is going to change dramatically and give you some really cool ideas.
- [00:24]
Kip
On what to go build.
- [00:25]
Kieran
Let's get to today's show.
- [00:29]
Kip
Hey guys, real quick. You know we love building custom GPTs on the show and we love sharing.
- [00:33]
Kieran
It with all of you.
- [00:34]
Kip
Well, we wanted to kick that up a notch. We just developed this free guide that teaches you how to build your own custom GPT on chatgpt. We've taken the guesswork out of it. We've got templates, we've got a step by step guide to design and implement custom models. So you can focus on the part that's actually fun, the part we love actually building it. And if you want it, you can grab a link in the description below and go check it out now. Now back to today's show.
- [01:13]
Kieran
So Kieran, on the show today, we wanted to talk about something that we've.
- [01:17]
Kip
Seen a ton of value in, which.
- [01:18]
Kieran
Is AI, especially as AI as applies to email marketing. And email marketing is something every business does. And you recently caught up with our old coworker Dan Walchenk, who's over at Reforged now. And he was sharing some pretty dope AI email flows that he'd built.
- [01:36]
Dan Walchenk
Right. He's got a incredible step by step guide on how you can really personalize email using AI on a one to one basis that will drastically increase your conversion rate. He actually doubled his conversion rate with the flow that we're going to go through. And the other cool thing is you can not only replicate what Dan does, but we're going to give you the tips and tricks of how you can actually improve on that again. Right? So maybe you're just not happy with doubling your conversion rate. You want to triple or quadruple your conversion rate. We actually tell you how to do that on the show. And then the other thing we did, Kip, was we actually, this was fun. Have given the breakdown, the readout of how content is going to work in an AI world and how you can be taken advantage of AI today to do things in content that no other company is doing, that no other brand is doing to Drive. Incredible results for you. Yeah.
- [02:28]
Kieran
I think we gave people the blueprint of what email newsletters of the future might look like. And as part of that, Dan shared an email newsletter that neither you or I are on. It's an AI email newsletter that it's sick and you're just going to want to stick around, if nothing else, just to subscribe to that email newsletter. But we're then going to talk about how that email newsletter turns in and kind of starts the way for a lot of changes in email newsletters like we'll all be making.
- [02:55]
Dan Walchenk
I thought the coolest thing about that AI newsletter is how we took a thing that they are doing with AI that is unique to them and extrapolated it and talked about how you could do this across all of your content. And I really think if you want to do something cool with AI and and you're thinking like, what shall I do? We give you several ideas, some of them so good that Kip and I have already started to slack about how we are going to do some of them. So you want to try to beat us to the punch and do them first. So let's get into today's show. We're trying to talk to people who are actually using AI in marketing in go to market. I think a lot of the talk around AI is specifically in products, like how AI has been integrated into products. And we hear less around how businesses are driving transformation across go to market teams using AI. I know you've done a bunch of cool things. We've caught up before and talked about AI. So maybe why don't you talk about the first use case that you and I connected on email and we can just kind of go from there and riff on it.
- [03:53]
Unnamed Speaker
Sure. So Reforge is really well known for having educational courses, some of the best in the business when it comes to product marketing and growth. And we've been expanding beyond that, building on that to help people actually apply the concepts and the things that we teach in our courses as people are actually doing their work. And so we've been launching tools to help people go from just learning to application. One of the things I've been working on is a Chrome extension. And the Chrome extension helps draft documents, it helps give you feedback, all grounded in the experts that teach our courses on our platform. And one of the things that I've seen in building this tool is that even though people work in technology, even though they're surrounded by these things all day, every day, they actually struggle to make the leap from here's this tool, how can I use it? Exactly the thing we were Talking about with OpenAI, we have this fancy model. How should you use it? And so it's just so interesting that OpenAI isn't grounding their examples in tangible things. They're trying to impress people with these esoteric ideas and concepts. And I've personally noticed this even more now that companies, us included to this day don't do a good job of saying, we have this new thing. Here's how it's immediately applicable to you. It's grounded in, we have all these new fancy bells and whistles, all these new features, but there's a huge disconnect and all the onus is on the end user to figure out, okay, how do I actually put that into practice? How do I actually use that? And the thing that we started to do was how do we not make someone make that leap all on their own? How do we make it much more actionable and concrete whenever we do that? Outreach?
- [05:24]
Dan Walchenk
Yeah, I want to just give a little plug to the Chrome extension just to make sure people realize what that is. I thought it was a really cool AI use case. So Reforge has the kind of best experts in the business teaching all of this knowledge and you have all of this propriety data because you have like a backend community as well. One of the things there's a company I talked to recently, there's a guy called Ishan, he founded a company called Rocks. They kind of went pretty viral. They are an incredible AI app. And he had in his breakdown packing, McCormick did a really great breakdown on Not Born. And one of the questions that Ishan was asked, and it was like a really interesting question because I think about this a lot, which is, where is the arbitrage opportunity for AI? Let's imagine a world in five years time where everyone has the best AI tools, right? And so where does anyone get leverage? And the thing that he said, which I believe is true, is where you get leverage is in the uniqueness of your data and how proactive you are in using that. And so the reason I'm bringing that up is because like we Forge are one of the companies that do have proprietary data, because you have all of this lock data around the expertise that are being taught to that community and then the community feedback. And what you did was you built a Chrome extension so you can kind of access that data whenever you need it. Like one of the use cases could be, you're in a business setting, you're trying to write a memo and it's about plg. And you're like, hey, I wish I could really run this by Alina Averna, who's one of the people who teaches there and a real expert in plg. You can basically click on the Chrome extension and get like Elena's feedback on your memo, right? Because you've kind of taught it might not be specific to a person, but she's like, her expertise is included in that. And so it's like you have the two things, which is the propriety data and then the proactivity of like, when you use that data, it comes to life in your workflow. So I think that is like a pretty cool use case for AI in general.
- [07:12]
Unnamed Speaker
Yeah, that was a great description. I think that if you want to have differentiation long term, you need to have something else that's just not just text generation. And so one of the things for Reforge is we have the proprietary content. And the idea here is that, sure, you could go ask any of the other many tools out there to generate a PRD for you positioning Doc a Persona. But it's nice when you know where it's coming from. It's grounded from experts you trust and content. And you can also then click into it and you can see some of the best content anywhere out there to understand, here's how someone did it, here's how they did it at Miro, at Dropbox, at SurveyMonkey, at name brand companies, you know, and respect. So that's definitely one of the things that we're relying on as we think about going forward. But to your point, the idea is you don't have to go somewhere else, you don't have to build some other habit to say, oh, let me go to some other website. A lot of people actually have already built that habit with ChatGPT and Claude. But the idea is how do we make it something like Grammarly where it just shows up natively in the tools that you're using so that you don't require hardwiring some new habit in people.
- [08:15]
Dan Walchenk
Right? Yeah. Grammarly is one of my favorite PLG business for that very reason. Okay, so you were going to talk about then the email. So part of what you were doing was trying to get more users to use the Chrome extension and you ran an email campaign where you integrated AI into the email campaign. Could you maybe walk us through that campaign?
- [08:34]
Unnamed Speaker
Yeah, sure. So the feature that we were launching was we were launching the ability to generate 15 different document types, the classic document types that PMs and marketers have to create could be a PRD, could be a positioning doc, could be a Persona doc, a retention analysis. Lots of classic style documents that are target Persona. Our target user was creating on a very frequent basis. And in creating this email, it just felt like this email came in and said, hey, we have this new feature, you should try it. But a huge thing that I've seen throughout the product development process is that unless people see it on their own data, on their own company, they can't make that leap. And it's interesting to think about the death of Envision and how Figma prototypes have kind of taken over from that. I think it's very hard to test a new product with something like a Figma prototype for the same reason people can't make the leap. They say, this looks like it'd be interesting, but unless I can see my own data in here, I can't evaluate whether it'd be helpful or useful to me. And I had the same feeling, gut feeling, when I was thinking about this email campaign where we wanted to say, we have this great new feature, it can draft your documents for you. I thought, why don't we draft the document for them and send it in the email? And so that's exactly what we did. What we did is we took a list of users from our system, we exported them as CSV. And then what I did is I took their title, I took their company, and I also took some information we had in our system about what their company did, and I exported that as a CSV. The whole thing that I wanted to do was I wanted to show people the value that the product could do. I wanted to show that it could generate PRDs using the templates that reforged talks about in our courses and how compelling and how useful they might be. And so I'll show you what the end result looks like first. So this is an actual email, is a test email. But the thing that made it so interesting to me is that because the company description was in Japanese, Chinese, I don't even know what character said this is, to be totally honest, but the LLM just said, oh, the company description is in Chinese, it's in Japanese. And so I'm going to render the PRD template in this language. I just remember having this whoa moment of this is the kind of thing that would be so much more compelling than a standard email that our team would send saying, hey, we just launched this new feature. You should click through this email to go check out the results. And so how do we do this what do we do? The simple process was we had the LLM generate an idea for a new product. And how we did that is we had, I exported a CSV from our BI system that had a list of someone, email address, their title, their company, and the information about their company. We don't even get into the tools that give you that information. Oftentimes you already have it. And so it was really easy to get that CSV. And then to start off, you could just put this in a spreadsheet and just try it out one by one. Have the LLM generate an idea for a new product.
- [11:29]
Dan Walchenk
Idea for new product. You mean you want to have a fictional product for that company that you can target.
- [11:35]
Unnamed Speaker
So oftentimes a marketer or a product manager, a lot of their work is let me think about what I'm going to do next. I want to build a new feature or I want to have a new marketing campaign. And so I'm going to write a document that describes what this product is supposed to do or what this campaign is supposed to be. And so oftentimes you need an idea. What is the new feature? You know, HubSpot's coming out with the HubSpot app for the Apple Vision Pro. That's a thing that someone I'm sure is thinking about somewhere inside HubSpot at some future date. And so the idea is you need that concept, you need that idea to think I need to create a document for that idea. And so the whole idea is, okay, great, let's think of that idea. And then once we have the idea, then we can go and generate the PRD for that idea. And there could be lots of different document types. It could be a marketing campaign, could be a Persona, it could be your ICP for a product or a tier new offering. Ultimately there's different document types. And so when we productized this, we created many different document types. But in just experimenting with it, we said, let's just do it for one. And so we generated a PRD using the template that Reforge has in one of our courses. Once we had that, we would say, great, generate a compelling subject line. And the idea here is we wanted Dynamics, a subject line that would stand out that was hyper tailored and customized to them. I'm more likely to open something that references Reforge new features for reforge problems, opportunities for reforge than something that's just generic to some company or product that I subscribe to. Once I had the subject line, then I would also generate the HTML required for the email. And so then I had a whole bunch of different properties, new columns in my CSV that I could then just take and import into my email system.
- [13:16]
Dan Walchenk
These were separate prompts. Dan, just really quickly, on the subject line, did you have a prompt per subject line and then a prompt for body copy?
- [13:22]
Unnamed Speaker
Yes. One of the things we didn't get to yet is trying to collapse these all into one mega prompt. You'll see this a lot with AI stuff is if you, if you think step by step, if you have it, break it down into discrete tasks, you're ultimately going to get a better result. And so that's part of the reason I did this. One, it's easier to debug. And two, I think you're going to get better results given the state of the models of where they are today. Because things like generating a good hook, a catchy subject line kind of different from generating the HTML for a PRD in an email. And so you really want to just make sure that the result is as good as possible for that particular task. And so that's why I broke it down the way I did, just to make it easier to debug. And ultimately I thought I'd get better results.
- [14:06]
Dan Walchenk
Yeah, yeah, we found the same thing, that the emails perform much better when you separate out the prompts.
- [14:11]
Unnamed Speaker
Yeah. I got so many email responses to this email campaign. People said, this is the best piece of email marketing I have ever received.
- [14:21]
Dan Walchenk
They said people don't usually say good things about email marketing.
- [14:24]
Kieran
No, people don't really hate getting most emails.
- [14:26]
Unnamed Speaker
Right. And multiple responses where people said, wow, this is amazing. This is fantastic. There's just a fundamental difference between an email that has this in it, where this is the kind of thing that would show up in a sales email. Not even a really good sales rep would do this in a email campaign. But you have that power now to generate things like this for your customers. Even better than a personalized sales email, in my opinion.
- [14:50]
Dan Walchenk
Right. What was your metric for success then? Like when you ran this campaign? Really interestingly, one of the like, aha moments for how good AI could be in personalization was when we were doing something similar at Zapier. Actually we basically had a list of people's apps and then their role title and we could put together like a jobs to be done on things you can actually do with all of these different integrations. And we had similar results in that we had people write back to that email saying, wow, this is the most useful email that I've ever gotten.
- [15:23]
Kieran
Right.
- [15:23]
Dan Walchenk
Because it was so personal to that person. And prior to AI, you would need to have someone go and actually write out that guide for them individually. And now you have AI can basically do that. Like, it's a pretty smart person that can go do that at scale. And so, yeah, I'm curious, like, what was your metric for success? I know you got the ad hoc feedback that this is really good, but how did you know that this was much better than a typical email campaign you would all run?
- [15:49]
Unnamed Speaker
Yep. Yep. So I think some of the things that are most important for an email campaign are the open rate. And so we saw generally very good open rates. But then the other thing that made a huge difference was the click to open rate. So removing the subject line, which is probably the most important thing in an email, is the subject line of the people that open it. What percentage of people click through on the email? So they click to open rate. The click to open rate was double than emails that didn't have something like this. And that felt huge to me. It felt massive.
- [16:18]
Kieran
It is huge.
- [16:19]
Unnamed Speaker
Especially for something that this kind of thing is easily generalizable, where you can say, once you get a bunch of really compelling information about what's important to them, once you identify those data sources, you can take this process and run it on tens or hundreds of email campaigns that leverage data and build context about what really resonates with people or companies.
- [16:41]
Dan Walchenk
Yeah. Was there anything unique about the data that you used that other people wouldn't have access to? Like, did you use any of that proprietary data to craft the email? Or was it just like the normal data you would have around a contact in the domain?
- [16:55]
Unnamed Speaker
So a lot of this was taken from my time at HubSpot, where we collected a lot of data about what people are doing, what apps they had, how they're using the tools, and taken that a lot to reforge. And so we track what courses you're taking, we track the percentage of the course you complete, we look at the templates you're clicking into, we have information about your team, about your company. And so we have a rich data set that's very actionable that we can then pipe into our email tools to say, which courses have you taken, which live events have you attended, what's the content that you come back to frequently? And so all of that stuff can easily be piped into our email automation system, or we can pull from our BI system to make it for a very compelling and ultimately helpful email campaign. So if someone's taking the growth series and they're really focused on the retention and engagement section. We can pull out the material from that section, apply it to their unique use case, and that can be automated in a really, really helpful way. Instead of saying, hey, we have a course recommendation for you that might be interesting. I think today, and I think a lot of companies would even struggle to do something like that. I think the really next level thing would be this is the content that you looked at most. We want to help you apply it in your job today. We could potentially take that whole content, pipe it into an LLM, and say, here are the ways in which you could apply this today, next week, next quarter.
- [18:19]
Dan Walchenk
Right.
- [18:19]
Kieran
But what you're really saying is it's like it's the behavior data that those people have with your case reforge. But anybody watching the show has some behavior data. What you all have done is just been very smart about, hey, I want to be clear on what I'm tracking, and I want to have a good breadth of that behavior data, because the more breadth of that behavior data, the more interesting things we could do in terms of helping solve those problems for our customers or prospective customers or community members. Is that basically the way everybody should think about it?
- [18:48]
Unnamed Speaker
Yes. I think there's multiple classes of data that'd be really interesting. I think there's information that you collect in your system. I think there's external information. I think it's never been easier to go crawl someone's website and then leverage that information. So, like, you know, a good example might be, especially if you're a B2B company, you could crawl someone's website and understand what their pricing tiers are, understand what the dollar amounts are for those pricing tiers, and you could help them understand how they can pitch their offerings effectively using your software and you can ground it in their pricing tiers. Yeah, and so there's, you know, if someone's sending a lot of email campaigns, but they don't have a lot of landing pages, you could easily say something like, you should create a landing page that talks about the benefits that's on your enterprise tier. And so I think there's good behavioral data that you could pipe in. You know, I was thinking about how you could even take information from your internal CRM, you know, notes from a sales rep or notes from a customer success manager about what's really important to that customer. You could pipe that into the LLM as well. You know, that's not behavioral data. That's just your own internal information about what's important to the customer to leverage that for these email campaigns.
- [19:57]
Kip
I want to tell you about a podcast I love. It's called Created. It's hosted by John Ushai. It's brought to you by the HubSpot Podcast Network, the audio destination for business professionals. Creating gives an inside look at how to build businesses as a creator, improve your creative process and stay ahead of the latest content trends. John, an ex YouTuber and Instagram employee, breaks down the creator economy with guests Logan Paul, Paris Hilton, Jason Derulo, Jake Paul, George Lopez, and more. In a not so recent episode I loved was with Liza Koshay.
- [20:27]
Kieran
Remember her?
- [20:28]
Kip
She was one of the fastest growing YouTubers and Viners, growing a fan base well beyond 25 million subscribers. But then she stopped uploading. So what happened? That's a smart woman who made a drastic career decision. You'd want to tune into that one. If you're looking for insights on entrepreneurship and creating a powerhouse business, then this is the episode you won't want to miss. Listen to Created wherever you get your podcasts. Hey, it's Kip. If you listen to this podcast, you know how much I love keeping up with the latest and greatest in technology. But very few podcasts actually give you a dose of the future. The A16Z podcast is the exception. It's hosted by our friend and frequent guest Steph Smith. The chart topping show brings on movers with a track record of being both early and right. Like Apple co founder Steve Wozniak or the CISOs of OpenAI, Anthropic and DeepMind. Even the very first CTO of the CIA. From the science and the supply of GLP1s to drone delivery to the economics of deepfakes. Go ahead and eavesdrop on the future. Check out the a16z podcast wherever you get your podcasts.
- [21:45]
Dan Walchenk
This is like actually where people just don't understand how valuable agents are going to be long term. Because in that use case you can have an agent that is trained just on scraping and summarizing a website. Then an agent that is trained on taking all of your internal engagement data and summarizing that. Then an agent that is trained on taking all of the information you have around that contact and is trained on that and all of those agents are doing their little jobs and they'll all come together and give you like an incredible rich source of data. And then you could have another agent that actually interprets that and then provides a first version of like what would be the most interesting ways to reach out and summarize that content for that person. And that's why I was like, at the start wondering, anyone can replicate what you've just talked about, right? Like, there's some creativity, but if you have any kind of data, you can do some version of that, but then like, to not get sucked into where the average will be, it's like you have to have some unique data source, like something unique and a creative way to use it to get results that are going to be above average. I think that's where we are going to be moving pretty rapidly, because the AI tooling to do this for you, like, you were able to put that together, but at some point there's going to be a tool that just does it for you. So the thing I just described will happen. You just deploy a bunch of agents, we'll pull it all together and write the email. And in that scenario, then who actually has any leverage over anyone else? And I still think the way it happens is still like, there's some unique data and then the human is just much more creative in how they use that data.
- [23:11]
Unnamed Speaker
Absolutely. So I think the hype over agents is real, and I'm so excited for that future. And I could see some people saying, why should I experiment with little processes? Because the agents are going to do it for me. And I think the real key is in experimenting, figuring out what works, and then you can go automate it. I mean, a lot of these tools don't work yet today, but getting really good at experimenting and trying little things out.
- [23:37]
Dan Walchenk
Exactly. Yeah.
- [23:38]
Unnamed Speaker
Is I think, where the real value is, figuring out what the proprietary data sources are, how to use it. And once you can demonstrate that this is really compelling, then you can hand it off to someone more sophisticated, or you can hand it off to an agent to go do it. But I think the key thing is understanding your customer, understanding what's truly valuable to them, and then saying what is the proprietary data source that no one else is going to get yet. Yeah.
- [24:00]
Dan Walchenk
You have a big background in PLG or early working in the early team at HubSpot. And one of the problems you solved is like the number one problem for PLG companies is the cold start problem. It's like, I can't actually envision how this freemium product works because I don't have any data. I'm not going to import my data because I don't know if I'm going to use it. And then this is like that kind of loop. Anything you learned from doing this that makes you think there's potential for AI to solve that cold start problem for some SaaS companies.
- [24:29]
Unnamed Speaker
Absolutely. I think there's so much information out there now. And so I think previously it was there's nothing out there and it's really hard to understand who someone is and what's valuable to them. And I think now it's how can you be really helpful? And so now I think it's all about how can you source compelling information that's out there and provide some real value? How do you stand out? All these companies are out there sending you emails with generic value prop and not differentiated. I think the companies that are really good at understanding what is important to someone and sourcing that information and sending it to you in the right way. So as an example, one of the things I've started to do is I've started to download podcasts from throughout the web and I download the audio, I transcribe them using software just on my machine so I don't have to pay for anything. And then I'm able to take transcripts from those podcasts and I'm able to mine them for insights. And so I can find snippets that are really interesting to other people. And so that's the kind of thing I don't think anyone else is doing that everyone's using something where they say, I want to go and take a YouTube video and I want to create clips for social media. But no one's saying, I'm going to look at all these different podcasts, I'm going to find the clips that are interesting to individuals or to different people and to say, I thought this would be really helpful for you. There are examples, people that are really relevant, really helpful of what's the thing that's going to help you succeed? How do you get ahead? What's something that you're not maybe seeing? And I think that's the kind of thing that would be differentiated now, where you can source information from lots of different places, but serve it up in a way that's really compelling and interesting. One of my favorite AI newsletters sources information from many different discords, from Reddit, from all the different social media sites, and it summarizes all of it automatically for me. It's called AI News, and it's one of the emails that I religiously open and consume every single day. And because there's so much out there, because it's so hard to keep up with all the news, it's a fantastic resource. And that's the kind of thing that like, that is the issue today. How do you keep up how do you know what's interesting? And so I think that is a big opportunity today of taking advantage of the explosion of information and distilling it down into something that's really compelling and interesting and applicable and feels actionable to the person. So this email newsletter is highly recommended by Andre Karpathy.
- [26:45]
Dan Walchenk
Okay.
- [26:45]
Unnamed Speaker
Yeah, he's like the OG. He was in early OpenAI. He was at Tesla self driving cars. He has all these YouTube videos. I mean, don't listen to me, listen to Karpathy. He loves it.
- [26:56]
Dan Walchenk
This is super interesting. Yeah, I've thought about this. You'd be interested in this, Kip. So I've been thinking about building a small little app that does this. They sign up to all of the discords, they have all of the Reddits, they have all of the Twitters, and they put it all into a newsletter that is actually a really cool use case of AI because it's taking unstructured data and then making it incredibly valuable. I'm actually just signed up. I had not heard of this before.
- [27:19]
Kieran
What's the website again? Artificial intelligence?
- [27:21]
Dan Walchenk
No, it's buttondown.com ainews.
- [27:25]
Unnamed Speaker
All right. It's fantastic. I've actually thought about doing a side project and HubSpot should just do it because you could do it way better than me. But the thing I want, I want the thing that listens to different podcasts and social media and says these are the amazing things that people are doing and applying AI. Yes. I want a speed of the amazing things that people are doing sourced from all the different places. And I want it delivered to me and not just summarized. Right. I don't just want it summarized. I want you to go one step further, just like I did in my email campaign, to say this is how you should apply it, given what we know about you in your business, what you're working on, all that stuff. So, you know, in my example, this guy did a customized email campaign. Big deal. But here's how we would do it if we were you. Right? You're a B2B software company. You just launched, you know, a bunch of AI features for HubSpot. Like, you could talk about how Breeze helps me as a HubSpot customer, Do X, Y and Z. And it's not just HuffPo, but for AI. It's not just the summarization of what's out there, but it's the how to apply it.
- [28:32]
Dan Walchenk
Yeah, that's really good.
- [28:33]
Unnamed Speaker
And the thing that's not in here is the podcast. There's a Twitter account I really like it has clips from YouTube videos, pulls out all the good stuff. And so like I want that, but I want that for things that I can go execute, I can apply in my day to day job.
- [28:47]
Dan Walchenk
So basically the AI agent listens to podcasts, transcribes them, it parses through them, it has a list of people who are subscribed, it then has data about that and it personalizes the use case to that person and emails you a newsletter that are these use cases that personalize to the data it has about you. This is like where I think all media goes actually we talked about this on the show Kip. This is like one to one. See, all of us will be signed up to the same newsletter but the weekly content we get will be drastically different. It's coming from the same unstructured data, but it's personalized towards us, hyper targeted towards us.
- [29:26]
Kieran
This goes with my one other pet rock on the subject. I agree with everything you and Dan just laid out here. The additional step would be the problem with email newsletters is that no one hits the reply button ever. Ever.
- [29:38]
Dan Walchenk
Yes, ever.
- [29:39]
Kieran
And imagine everything you just said, Kieran, where I got this really personalized summary in this case for AI news because we're all dorks, right? And that's what we're interested in. And imagine just being able to like find one thing on that newsletter that you're really interested in and hit reply and have it send you a brand new newsletter just on that one thing, completely expanded.
- [30:00]
Dan Walchenk
Yeah, yeah, this is great. I have so many ideas here. I have so many ideas.
- [30:05]
Kieran
All right, we'll see you next week guys.
- [30:07]
Unnamed Speaker
I've had the same idea and the thing I love about it is like I don't want this in chat, I want it over email because it's like take as long as you want to go and pull it for me and come back to me with something interesting. That's one thing that blows me away about all the LLMs. Everything is chat based. They don't have a feature that's like, hey, where are you going? To think about it. And we'll email you and you can email back with us. And the thing that I keep thinking about with this is do you know what that would do to your deliverability if all your users are constantly emailing you back? It would be fantastic. And the context you can build about the person. I don't want to go to some app, some other app on my phone or on my computer. I want the thing that just knows who I am to me Especially as a geriatric millennial. I just want to email it back and forth.
- [30:53]
Kieran
First of all, I'm stealing geriatric millennial because I definitely am one of one of those as well. Become like the oldest possible millennial. The other part of this now, now we're just talking marketing, everybody. It's the other things you don't do is nobody links of where you can actually go and use that thing. So let's say it's talking about a new Chat GPT feature. It's like you then go and you figure out in ChatGPT how to do it. And it's like you could imagine like going and applying some of that stuff in certain products and recommending the products, having all the click data of which person uses which frontier model, then customizing. It's like, oh, we know this is a cloud person. Like we can now do this customization based on that and make this content even way better for them.
- [31:35]
Dan Walchenk
Yeah. I think there's something really pivotal in what we're discussing, which is what's happened to Google. All that's happened is I used to have to go through all of the blue links and now an agent goes through all of the blue links and gives me something clear and concise. What we're talking about is basically I used to have to subscribe to all this media stuff and consume the content. Now my agent subscribes to all of the media stuff and consumes the content and gives me a personalized version of it, which is kind of wild to think about because all of the knowledge and content that we consume today all gets moved into just structure unstructured data for the agent. Like there's an agent layer doing all this stuff.
- [32:16]
Unnamed Speaker
You can't unsee this feature once it's clear. And it drives me crazy. The fact that I need to comb through all these sources, the fact that things aren't automatically customized and tailored to me. I have to go read the documentation for some tool that I'm using. Why is it just explained to me in the moment? Customize. And why are we sending out the same email to everybody? Yeah, it feels like a crime. Feels like a crime against users.
- [32:41]
Kieran
It feels crazy, right? Yeah, it feels literally insane. What's happening.
- [32:44]
Unnamed Speaker
Yes. In what world should we send the same message to the same people? Why wouldn't it be tailored and customized? It seems crazy.
- [32:53]
Dan Walchenk
Yeah. This is why I think AI native software is going to be so far removed from what we do today. I've been talking about that for two years. But the first kind of swath of AI companies actually all did look like SaaS. Click, click, click, click. The newest cohort, which I would put in the last like six to 12 months are starting to look drastically different from what we think about as software.
- [33:14]
Kieran
They don't look, they're looking very different.
- [33:15]
Dan Walchenk
And that to me is the biggest shift that I think will happen. Like it's happened throughout time. Right. If you, if you went for on premise software to SaaS software it would be completely a different experience. I think the software to AI software is just going to be totally reimagined. Like the actual UX patterns are so different and the actual output is going to be so different. Like it's going to be crazy that we used to do this segment level content. Like people are like what you used to like take segment level based upon the person's role like that. Oh, we took a role and we said you're all the same. Now you all get the same email. I think it's going to be an incredible time to be a marketer. If you are the kind of team and marketers that can actually really dive into the personalization part and get really creative there.
- [33:58]
Unnamed Speaker
Yeah. The same way that you look back at the abacus and you say wow, that's the stone age. I think even the modern day tools we use right now are going to look like toys. When the UI is dynamic and generated by UI and everything is infinitely tailored and customized for, for your unique company and individual.
- [34:17]
Dan Walchenk
I love that.
- [34:18]
Kieran
And by the way, remember, if you're a marketer out there in that world, shared experiences are gonna be leverage. Everybody's getting personalized. Everything bringing people together and having something that they can share outside of just themselves will become a very scarce resource.
- [34:35]
Dan Walchenk
Yeah.
- [34:35]
Kieran
And very, very valuable resource.
- [34:37]
Dan Walchenk
Being able to garner influence on mass would be the most valuable commodity you.
- [34:42]
Kieran
Have be the most precious resource which.
- [34:45]
Dan Walchenk
Is basically the digital twenties. Right. It's the 1920s old. It's like that was the core craft of marketing was the ability to just craft incredible copy. It's like reverting to a similar art. I think this is a good place to end.
- [34:58]
Kieran
I just want to go build this newsletter now, man.
- [35:01]
Dan Walchenk
I've actually. So that's literally all I'm thinking about.
- [35:04]
Kieran
Literally. We're like, well I guess the show's done because we want to go build this right now.
- [35:09]
Dan Walchenk
I'm actually on my notepad doing bullet points right now.
- [35:13]
Kieran
Share that with me.
- [35:14]
Unnamed Speaker
I'm surprised you're not in repl it right now being like go, go Go, go, go.
- [35:17]
Dan Walchenk
Yeah, yeah, yeah. I'm not too far. I'm building a functional spec as we're talking.
- [35:22]
Unnamed Speaker
Okay. Okay. So this is confession time. I had this idea for teaching, and the idea was like teaching a concept. And it was like, I'm a PM at Reforge. Teach me SQL. And the idea was just drop an email to me every single day and actually build a proof of concept where you could email it, it would go to OpenAI and it would reply back within a couple minutes and it would say like, great, here's the thing. And the idea was like, you could continue the thread or it would just take that context and it was stupidly easy to go do. And this was like, I don't know, like six months or a year ago. I can only imagine how much easier it would be to actually get started on something like this with a replit agent today.
- [35:57]
Dan Walchenk
Yeah. And the latest models. This was awesome, Dan. I know that our listeners will get a ton of value not just from the practical use case, but also just trying to think about where AI can go in the future and how people can start taking advantage of things that have not been built yet. So appreciate you coming on and sharing everything.
- [36:13]
Unnamed Speaker
Thanks for having me.
- [36:14]
Kieran
Thanks for being here, Dan. See you.
- [36:15]
Unnamed Speaker
See you guys.