Loading summary
Transcript168 lines
- [00:00]
Akash Gupta
Foreign.
- [00:10]
Wes Bush
Welcome to another episode of the Product podcast. Today I'm going to be talking about a very exciting topic that everybody should be thinking about right now. If you're listening to this podcast, which is AI and plg, because right now it's like the best combination of the two, I think plg, when it was first introduced, there was just not a lot of AI tools out there that were actually useful for PLG companies. And so in this episode with Akash Gupta, he's got the biggest newsletter. Almost 200,000 people in the product and growth space. So he's just a fantastic guy who really does an amazing job at just articulating the forward thinking parts around how do you do product and growth at the highest level? And so you're going to get a ton out of this episode. I know I did. It really breaks down into two key pieces of one, are you using AI for all of the key things that you do to become more effective in everything you do, whether that's just using simple transcribing tools or just creating AI agents for every single piece of your day to day? And even for myself, like we use AI a ton at Product led, I was still like, oh yeah, there's, there's a lot of room to grow here and just get more effective at the tools you can use. So what I would recommend though is if you're listening to this on the podcast, definitely hop onto our YouTube channel at Product LED where you can actually visualize a lot of the tools. It might just hit the brain cells a little bit differently. And then the second piece of this is just wh, what are all of the ways you can use AI? Whether it's for thinking about how do you design your free model, how you approach pricing for AI and PLG companies, and so you'll go through all of those specific pieces and by the end you'll know exactly how you can leverage AI a lot more as a product LED company. Hope you enjoy this as much as I did enjoy learning from this episode myself. Cheers.
- [02:10]
Akash Gupta
There's really two interesting parts to AI and product LED growth. So the first side is how it's going to help you with PLG itself. So we're going to talk about that. But then sometimes the more interesting side for people and more impactful, so we need to talk about as well, is how AI just helps you do your job in succeeding at plg. So let me explain that in a little bit more detail here. So let's zoom out a little bit further. So if we take a look at plg, there's a few key layers that everyone focuses on in plg. Some people call it a seven layer system. I personally like to look at, you know, a specific number of layers for whatever company I'm talking to. So activation. Right. We were just talking a little bit about activation. Very, very important in the bucket of onboarding. A lot of people talk, spend a lot of time in PLG on pricing, on their free tiers. Right. So whether it's free trial or whatever you want to call it, model essentially, then a lot of people spend time on expansion, PQLs, everything related to that. Then people, you know, spend a time on their marketing motion, their marketing and sales. So basically like their GTM work. So these might be like just some specific buckets that we can talk about. And so what we'll do is we'll talk about how AI has impacted all of those. So that's one area. But the sometimes more interesting area is actually the other side of the equation. So let's draw these lines here and then what we'll do is we'll take a look at AI in your job. I'm curious, what LLM do you spend the most time using?
- [04:05]
Wes Bush
Wes, usually chatgpt. If I'm just brainstorming like every day, then Claude for whether it's like landing pages, copy emails, I use that quite a bit. I have not found much use with Gemini. Sometimes if it's a really deep thing, I'll like just blast all of them with like their deep thinking of like I'm really thinking about this question and I'm like, you know, gemini, go through ChatGPT and also Claude. And that's actually been the most interesting is like when you see just from different perspectives, it's like talking to five different people. How would you approach this problem? And you get some, usually some interesting new insights that I have not thought about.
- [04:45]
Akash Gupta
Yes, totally. So let's, let's dive deeper into that. I'm going to dive into a couple different categories of LLMs. So you basically mentioned like two of the three of the core LLMs and I'll, we'll talk about those. I think that using a ChatGPT or a Claude is powerful. I think that there's a couple other interesting types of LLMs out there that people should be thinking about. So one is like soul context. So we'll talk about that. The best soul context LLM out there right now is Notebook lm. Have you tried that out?
- [05:17]
Wes Bush
Yes, I did it like when it was first coming out and it was okay. But I absolutely have to get back in there because they keep updating it. Yeah.
- [05:25]
Akash Gupta
So basically like with sole context LLMs, what you're looking at here is not hallucinating. So if you're, if you have a task like you're writing a document for your execs or your internal team and you don't really need it to have the ability to go search the web or something, you have all the information and you're going to be feeding all that context in, then it's much better to use a sole Context LLM like NotebookLM. And most people, NotebookLM has kind of a bad user experience, honestly, because they really emphasize this like podcast that they can create for you in the user experience, but you kind of just ignore.
- [06:05]
Wes Bush
That many people using that.
- [06:09]
Akash Gupta
Yeah. Actually what you use it for for is just loading up tons of context. So that would be the first thing I say to every PLG person listening.
- [06:19]
Wes Bush
Right.
- [06:20]
Akash Gupta
Check out NotebookLM. Use it. If your company has access to Google Workspace, you probably have access to this tool. So it's also something that a lot of people have access to. Some of these other tools I'm going to talk about in a second, you may not have access to, you might need to be requesting them from your IT team, which is also cool. But you know, a lot of people have access to Notebook. So with Claude and chatgpt, I'm curious, do you ever use the projects feature?
- [06:44]
Wes Bush
Yes, I love projects.
- [06:46]
Akash Gupta
Yeah. I call this your co pilot. It's a specific project that you train. So you train it with some like very important data. So if you're somebody working in a company, you know what matters. Like maybe not as much for Wes and I, but your performance reviews. Right, super important. You train it with your performance reviews, you train it with your one on ones from your boss and maybe your boss's boss, your skip level. You, you train it with, you know, emails and context from your leadership team, like your strategy and your vision. So you create these really amazing high context projects and this is how you really get the power out of these LLMs. Sadly, when you use these projects you're going to burn your tokens a lot faster. And so you might have to go on like a Claude Max or a chatgpt higher plan than you're currently on. But projects I think are a huge unlock. What projects do you use?
- [07:46]
Wes Bush
So my favorite use case for them is I have everyone on my team take we like the Colby A assessment but you could do like any sort of personality assessment and you just learn like hey, how could we work better together? Like, right, how could Wes and Akash work better together? Give us the one pager and also whenever. Hey, Akash raised this, like, interesting issue with me. How should I approach it? And it's like, write a better response. And because it knows you, it's so accurate. And so I use it as basically like my chief people officer of like, hey, here's how I should manage the team. Work better with other people. And so, yeah, for me, it's. It's game changer for that part.
- [08:24]
Akash Gupta
Exactly. And then in terms of a core LLM, like, I'm curious, is there anything you think ChatGPT is better than Claude at?
- [08:32]
Wes Bush
Yes, you get a lot more value out of ChatGPT than Claude because as soon as you have a thread that's really long, you burn your cloud credit so quick. And so then you got to upgrade to the next plan. So I think if you're looking for best value, value, still chatgpt by far.
- [08:49]
Akash Gupta
Yeah, I would say value. But for you guys, like, you guys should get your companies to get you on a cloud max plan so that you don't worry about cloud usage. Like, that was the only thing Wes could come up with. Generally, people who use both, yeah, Claude is literally better at everything than ChatGPT guys. So use Claude instead as we create this workflow for you of how you're improving your job with AI. These are the tools, right? So you're gonna have your Claude projects, you're gonna have Claude, you're gonna have NotebookLM. These are like the top tools for PLG practitioners in AI. So this next bucket, I'm curious, what, have you played around with many agents yet?
- [09:26]
Wes Bush
We have, like, we built some in, like Lindy and Adam as well. And so those have been like, super helpful as well. But I think the big bottleneck there is just training the rest of the team to use them. Like, Nadan still is not the easiest to use. So.
- [09:42]
Akash Gupta
Yeah, yeah, a hundred percent. So I think that NAN is very hard to use, but it's probably the most powerful agent platform out there. So what I'm going to recommend to most practitioners, it depends who you are. Like, if you're a marketing ops person listening to this, then probably N8N is great for you. Like, you're already a technical person. You're very comfortable with these kind of things. Boom, go for N8N. But like you mentioned, I think a Lindy or have you ever tried Zapier?
- [10:12]
Wes Bush
Yep, yeah, we use Zapier. We were using it before, but Then some of the team was like, well, let's use N8N because you can do a little bit more. But I think Zapier might be at par. I don't know what your take is. Yeah, yeah.
- [10:24]
Akash Gupta
I personally think, like, Zapier is a little bit more powerful, but because it's a little bit more powerful, it can sometimes be a little bit harder to use than Lindy. And so we can basically just hire. Create a hierarchy here. Like, the more technical you want to be, the closer you go towards a zapier or an N8N and the more functionality you need, then you have to just go to those tools. But otherwise something like a Lindy I would also put in here. There's this bucket called Relay.
- [10:49]
Wes Bush
Oh, yeah, Relapse.
- [10:51]
Akash Gupta
Yeah, yeah. So that's another one that's in this bucket. That's quite good. So you have your agent platforms. And by the way, in the nan bucket, also there's make.com, which is basically as good as N8N. So and in Zapier bucket, you have Airtable. So basically you have all these companies that are producing pretty good products. And what are your favorite use cases of these for PLG practitioners?
- [11:14]
Wes Bush
Right now we're mostly using this internally just for different workflows and everything else. So we have not actually experimented much with like, okay, here's how to do better onboarding, reviews or different pieces of the puzzle for plg. Like on that, right, that you have for the.
- [11:30]
Akash Gupta
Yeah, let's go to the right side later. Let's stick to the left side. So like you just said, internal processes. This is where these things really help you guys.
- [11:37]
Wes Bush
Right?
- [11:37]
Akash Gupta
So I think a couple things that you should consider is like a competitive analysis agent. So this agent, like anytime your competitor makes a pricing change or they launch a specific feature, it can ping you in Slack. Right? And this can save you so much time. You can even. I've created some really nice prompts for myself where these agents will literally analyze it vis a vis my strategy and it'll have my insights on my pricing strategy too, you know, so you can just make these agents extremely powerful from a competitive analysis point of view. I think another one that's really powerful is an ea. Do you have an ea?
- [12:15]
Wes Bush
Yeah, definitely.
- [12:16]
Akash Gupta
So you pay for an ea, right? Like, I think you might. I don't know, sorry, ea who's listening to this? But you might be able to replace your EA at this point. Like, AI has gotten so good. And I'll show you guys a really interesting Jacob Bank Yay. I had this guy on the podcast, the CEO of Relay app. And so he created this 12 agent executive assistant, which is just awesome. And it really covers like a $1200 executive assistant. So we're talking about meeting briefs, adding events from Gmail to your calendar, following up for meetings, summarizing them to Slack, reminding guests to rsvp, showing when you have unusual events, taking out cold sales emails, giving you stats on your email, summarizing all the newsletters that you get, auto replying automatically to simple questions, scheduling your tasks, helping you prioritize. So it can literally you can build these agents to do so much. Totally so powerful. I don't know if any of these use cases in particular you think we should highlight for folks or are they all good?
- [13:19]
Wes Bush
Those ones are all pretty good. I mean like that's what a basic executive assistant should do. And I think even if you do have one, if they can get that automated, which like I trained my EA to just, you know, use Relay as well and a bunch of the other apps like Lindy and that's like, then they can focus on making decisions for you, which is still super valuable much doing the basic stuff. But did they RSVP to that event? Like totally automate that.
- [13:45]
Akash Gupta
Yeah, exactly. So make your EA a 10xea. Go from just a regular EA to a 10x EA. You know, with, with these types of automations by helping your EA with all of this.
- [13:57]
Wes Bush
Ooh, I got a fun question to add into this that I just got from a conference. But asking your team every week, what are you doing now that you could probably automate with AI and like that, like those things on that EA list? I think there's a lot of things they're like, did you rsvp, like did you check in manually? And it's like, well, that could have been automated. And it's like great. Every week just have something you can add to that list of like one new workflow or something like that to really get your entire team thinking about.
- [14:25]
Akash Gupta
It a hundred percent. The next one that people should be thinking about is marketing. I'm sorry marketers, but your job is AI agents are coming for your job. You know, it's unfortunate, but it's true. So basically what's happening, right, is if you think about it with, let's say you're like a product manager. If you're a product manager building a PLG feature, you probably still want to make a social post, right? About your feature. Hey, we just launched it. Or you maybe you want to run a Webinar series or maybe you want to automatically schedule meetings.
- [14:56]
Wes Bush
Right.
- [14:57]
Akash Gupta
So this is where you can create these agents to do all of your marketing for you. And it's kind of wild how much they can do. So I'll give you another example. Same from Jacob. Since we've just been looking at his other agents, let's look at his 40 person marketing team. So this is his 40 person marketing agents. LinkedIn post drafter, LinkedIn comment reply or YouTube description generator YouTube script generator LinkedIn postwriter, LinkedIn lead enricher, LinkedIn influencer post researcher LinkedIn post tracker LinkedIn comment helper tweet tracker, Blue sky monitor. Look at all these different things you can create. Blog post writer, blog post updater, LinkedIn post writer for new blog posts. SEO Ranking Tracker feature Launch tracker Integration tracker when you have a newsletter letter subscriber. When you get unsubscribe requests. When you need to onboard somebody with lifecycle marketing, track your customers, churn customers. When you need to schedule people for events, you need to do onboarding for your cohorts, you need to create reminders you need to discuss with people on Slack. You can literally automate all of this. Your partnerships, your partnership application review, your partnership finder, cold emailing, more partners, checking out your integration. Right. You can just do everything these days with in marketing, with agents. And so I think everybody should be thinking about how can I do more, right? Like 10x more than I'm currently doing on the marketing front because of AI. There's a huge arbitrage opportunity right now for people who are using AI to do marketing.
- [16:28]
Wes Bush
Totally. Yeah. Because you can't compete with them in a way. They can get so much more done than anybody else.
- [16:36]
Akash Gupta
Yeah, like people are pretty amazed by Alex Hero's $100 million book launch for a nonfiction book. Biggest book launch ever. Right? So how did this guy do it? This guy releases like 500 pieces of content a week. There's something to just doing a lot of marketing and tools like this will help you do it. So that's agents. Look at this. Look at how many use cases we've just built for you guys, right? There's more, there's more. Wait, but there's more. So dictation. Do you have a favorite dictation app?
- [17:06]
Wes Bush
You know what, I don't do much dictation other than in like Google Keep. But that's not even AI powered. That's just when I have something on my mind.
- [17:15]
Akash Gupta
So it's. I definitely recommend you doing it, especially when you're Prompting.
- [17:19]
Wes Bush
Right.
- [17:19]
Akash Gupta
So we all know that the more context in our prompts, the better. And we've all learned due to the myriad social media posts, the basic prompting techniques act as do this task. Here's some stylistic guidance. Here's my context. Here's an example of doing it right. Here's an example of doing it wrong. I want you to pay attention to this. If you do this really well, I'll give you a reward. My life depends on it, I think. What did Palmer Lucky say? Like, you're like a fugitive and doing this, My entire life depends on it or something, you know, so you're like using all of the techniques, but typing those up takes a long time. You type. We type like good typists type around 100, 120 words per minute. Average person is talking like 200, 220 words per minute. Myself, I talk at like 280 words per minute. Right. So you can speak way faster. So especially when you're prompting, using a dictation tool will save you a lot of time. Not even that. But how many times do you lazily prompt right with just like one sentence or one word? With the dictation tool, you can actually go through and do like a full prompt. And the cool thing about these AI dictation tools is like they'll add in all the formatting the sentence. It's like 100x better than Apple dictation.
- [18:35]
Wes Bush
What are the. The best ones? So super Whisper, that one's for Mac os. Is there any good ones for Windows?
- [18:41]
Akash Gupta
I wonder if Whisper Flow is for Windows. So that's awesome.
- [18:47]
Wes Bush
Interestingly enough, for just mostly macOS that I've seen.
- [18:52]
Akash Gupta
That's crazy. Yeah, I don't know. Maybe there. Maybe this is an app idea for one of you guys out there.
- [18:59]
Wes Bush
Yes.
- [19:00]
Akash Gupta
Build this.
- [19:01]
Wes Bush
We found the gap in the AI tools.
- [19:03]
Akash Gupta
The next bucket for any PLG practitioner. Right. If you're plg, you need to be making changes to your product and to your marketing on a regular basis. Otherwise you're not succeeding.
- [19:11]
Wes Bush
Right.
- [19:12]
Akash Gupta
To succeed, you should be doing a lot of tests on your life cycle marketing, on your activation, on your expansion. All the flows we are about to show on the Right. So you need to have a prototyping tool. You need to get out of documents and specs and like actually prototype stuff. Do you have a favorite AI prototyping tool?
- [19:28]
Wes Bush
No, I, I don't. Since I don't do a ton of.
- [19:32]
Akash Gupta
Prototyping anymore, I feel like everybody should be doing prototyping all the time. Like instead of suggesting A future idea. You just prototype it, you know? So, like, one of the best ones I think, right now is Bolt.
- [19:43]
Wes Bush
That kind of prototyping. Yes. Use Lovable.
- [19:46]
Akash Gupta
Yeah.
- [19:46]
Wes Bush
Yes.
- [19:47]
Akash Gupta
Yeah, Lovable. Okay. We can suggest lovable, you know, whichever one you want, but, like, it's so powerful. Let me just give you an example, right? So let's go to product led, your site, right? So we could take a version of your site. We could say, hmm, I like this, but, you know, I want to change up my site, right? I want to. I want to send my marketing team an example, too, of what I want to change it up as. So maybe I'll take a picture of this and I'll say, hey, this is my site. I want a version with left, right, top hero. The left is what we have now. The right is a video sales letter. Do you know what a video sales letter is?
- [20:26]
Wes Bush
Yep. Definitely.
- [20:28]
Akash Gupta
Yeah. So embed for me. So, like many, many service businesses, this will be how they organize their homepage. So instead of just telling this to your team, you can prototype it. And once you prototype it, what we're gonna see in a second is it's gonna reveal, like, the gaps in our prompt, and it's gonna cause us to, like, think about some extra details. And when we see it, we might say, oh, you know what? Well, obviously, I need to make sure that I have a really good thumbnail for my video sales letter, because it's just gonna be static, or you're gonna think, oh, do I want my video sales letter to Autoplay? And so when you create a prototype, you get that much more into the details. And let's go ahead and see how Bolt does. I'm really curious to see.
- [21:13]
Wes Bush
Yeah, I love this. So Bolt New is your favorite one for prototyping right now?
- [21:18]
Akash Gupta
Well, I'm actually. I actually launched a whole site, so a whole site like this, my latest project, this is all built on Bolt New. So last night, I was just vibe coding, like, some free tools, and we literally built, like, all these free tools. Like, if, you know, for SEO, like, free tools are really good. Check out all these free tools. I vibe coded last night, and it would just launch.
- [21:41]
Wes Bush
That's six. That's awesome.
- [21:43]
Akash Gupta
And, like, we wrote all these articles together, and we wrote, like, this thing on the curriculum. We wrote. We decided, oh, we probably need pages for each of the instructors. We probably need a page for the curriculum. I built all this so fast, all in Bolt. And look at this. This is your prototype. Do you like it?
- [22:01]
Wes Bush
Yeah, I. I like it. I think it definitely needs a video for sure, now that I think about it.
- [22:05]
Akash Gupta
Look at that. So we just helped upgrade Wes's website. And look, when you saw this, right? You see this embed, you're like, now I need to think about do I want it on YouTube or do I want it on a homegrown video player? And, you know, we were just looking at the homegrown video player on substack. It was kind of bad, right. I couldn't click to pause. So we'll think about that. Then I say, ooh, these case studies. Maybe I want to add in a logo of Boomi and Zuzo so I could do something like that. And then this is where your dictation comes in.
- [22:32]
Wes Bush
Right.
- [22:33]
Akash Gupta
So I want to make sure that the video embed placeholder shows a thumbnail and give me some inspo on what that thumbnail should be and the title so we can finesse that. Then I want to add the Boomian and Zuzo logos in here. And now I think the text add 1 million plus in product led revenue in 12 months without extra headcount is a little long because it's kind of squished to the left. So improve that. So I've just used Super Whisper to dictate my prompt to Bolt, and now Bolt is going to go make those changes. And so I think this is the power of these types of workflows. After just a few minutes, we'll have a nice prototype that you can send over your team that's like 10x more actionable than just you saying, hey, add a video. Right.
- [23:12]
Wes Bush
Definitely. And I guess for building web pages now, what do you typically recommend? Because I guess you just launch that all on Bolt versus using like a webflow or WordPress and stuff like that.
- [23:25]
Akash Gupta
Exactly, yeah. I mean, it's just, It's. I have WordPress and Webflow websites also for other projects I've done.
- [23:31]
Wes Bush
Yeah.
- [23:32]
Akash Gupta
And the Bolt website is a hundred times better. Like, on those websites, I couldn't have created these tools that I created. Right. How would I have created those tools? Impossible. These tools are like tapping into LLM APIs to, like, give people questions and things like that.
- [23:46]
Wes Bush
Yeah. I mean, it's interesting to see, like, for anybody who has like a SaaS web, typical SaaS website, it's like, is that just going to be completely the norm or is webflow going to eventually get their crap together?
- [23:59]
Akash Gupta
Yeah.
- [23:59]
Wes Bush
Innovate and make it easier. But yeah, definitely.
- [24:02]
Akash Gupta
And look how much better this homepage already is added, the logos added, this little play button added the title. So now you have a sense of, hey, now I can craft my video sales letter script. And look at this. Like, it made the text look better. It has a nice CTA to book your growth session. It looks like it's in your brand. So AI prototyping. Everybody should be using it. I think let's go ahead and say, like, Bolt New is our recommendation, but you can think about any other tools. I also like, like you said, lovable. Also like magic patterns. I like if you're trying to build an app, like let's say you're trying to build your own business, not just a website, then definitely don't use magic patterns because they don't focus on backend at all. But you know who does? Base 44. Have you heard the story of this company?
- [24:47]
Wes Bush
I have not heard the full story, but I have heard of them.
- [24:51]
Akash Gupta
So these guys started as one guy. This was a one mil. One guy who sold his company for $80 million. All cash.
- [24:59]
Wes Bush
Yep.
- [25:00]
Akash Gupta
Insane. And so what he did is he vibe coded the company on base 44. Like, literally, he built base 44 on base 44. If you want to build an 80 million solo person company, check out these.
- [25:13]
Wes Bush
That's pretty sweet.
- [25:14]
Akash Gupta
So look at that. Look at what we filled out here. Right. We just completed like half of our master class on AI and plg. And we barely talked about plg. We're talking about your personal productivity as a PLG practitioner. Because this is just. It's so important. Like you yourself want to be a 10x practitioner at what you do. So that we can now move over to this other right side.
- [25:34]
Wes Bush
Definitely. Yeah. I think the. If you were to go under, like, AI job on that end, it's like become the 10x version of you, so you can really have all that time and creativity to focus on the other part of PLG as well. So I love that. And there's a couple other PLG elements. I don't know if you would agree. They might make sense to be there, like on the data side or maybe even the team side. Could be some fun stuff there as well.
- [26:02]
Akash Gupta
Yeah, I think so let's talk about.
- [26:04]
Wes Bush
Both of those and maybe the overall strategy for the business.
- [26:08]
Akash Gupta
Yep, let's do it. So how does AI affect all these elements? Right.
- [26:14]
Wes Bush
Yeah. Right now, like, what we've done at product led is we just built out GPTs, which is still just scraping the surface for each of those, but like to help you get a free model to map out your product LED strategy, your pricing. But there's there's so much more stuff.
- [26:30]
Akash Gupta
That could be done there, 100%. We're pushing Miro to its limit here with these stacking links. The first thing to talk about is probably activation. Like, this is the hottest topic, right? Everybody's always talking about activation. So I think there's a couple different things with AI and activation that you should be thinking about. The first is like you should be getting people to your activated moment. Your TTV should be like 1/10 if you're a regular product and you're thinking about how do I reinvent myself with AI? And at the same time, because your TTV is going to be 1/10 your time to value. And let me explain this just a little bit, right? So like your activation moment, where I worked, where I was a VP of product last Apollo I.O. 2 1/2 million dollar sales tech company. Like the prime activation moment is when after you download an email address from our contact database. So let's say you want to email West Bush, you find West Bush in our contact database. You say, reveal email address. Then you send Wes an email address. That's like our activated moment. So why can the time to value be one tenth? So what we can do is when you log into Aakash's website on Apollo using AI, we can say, oh, it looks like you're from the domain www.news.akashg.com because you're akashakashi.com. so I looked at your website and I think that 10 contacts you should reach out to are this. And I actually think these would be really good podcast guests for you because I saw you're running a podcast. So I've also drafted these podcast emails from you. Wow. Right? So that's how you deliver 1/10 the time to value. But then at the same time, then your activation metrics need to become. You need to have like a ladder of activation metrics. So we always used to talk about setup and habit. That's a good one. So you can use that metric ladder if you like it. But these days in the AI world, I see people often talk about this, right? It's like activation moment one is like AI assisted activation, right? So that's that 1/10 the time value we just talked about. But then the next version is actually user driven activation. So we create this ladder now because the setup is usually the AI assisted. You look at AI assisted and user driven activation and then you look at how. So that's one way this AI changes your activation. Are there any other things on activation you would highlight that people really need to know about more.
- [28:57]
Wes Bush
Like the one thing I'm finding with our clients that are like fully AI enabled is just the profiling questions are really, really important to get the AI assisted activation right. And so that's just one thing where it's like you ask the right questions, you can get way better AI assisted activation. If you don't, then good luck. It could be hallucinating and not that helpful.
- [29:22]
Akash Gupta
Yes, 100%. So how are you doing the profiling? And in a lot of cases, like people I work with, they're not asking profiling questions anymore because they can infer a lot from the domain and who is joining. And a lot of times if this is an enterprise deal, you might even know like, oh, is this a PM or something like that, or you can connect into an Apollo to figure that out. So what people are doing these days is they're skipping the profile questions and then they're doing AI enabled. But then you need to get really good AI evals to make sure that like you aren't making bad assumptions. And you need to see how correctly these are being done. Either good profiling questions or good AI evals. And if people don't know what I'm talking about when I say AI evals, AI evals are these really important thing when you're building AI features like AI activation features that are going to assess the output. So you're going to create, for instance with sending an email, one way to assess the output is did they actually send the email? How many edits did they make? What types of edits did they make? So you might create an LLM judge to analyze. These are the types of edits users made and that would go into your eval suite so that you could really understand and consistently improve your AI assisted activation. So that's activation. Shall we move on to pricing?
- [30:39]
Wes Bush
Yes, let's do pricing. Most people don't price, right. And there's a few things I want to hear your take on with pricing because when I talk to a lot of SaaS founders, many of them when it comes to adding AI into their products, they for the first time they have like pretty big cogs cost of goods sold. So they don't know how much is this AI model going to cost. And so dialing in the amount to give away for free, the amount they give away in their each plan and how they price is really, really quite difficult for companies. So maybe even we just focus on like AI companies for pricing. That could be really interesting too since there's a lot of question marks for a lot of SaaS founders there.
- [31:22]
Akash Gupta
Yes, exactly. Right. AI has COGS. And I think Brian Chesky just made the news because he said he's been using Alibaba's open source Quinn model. Right. He says he doesn't even use OpenAI models in production. So one of the important things to think about is like, what models are you actually using in production? When you're developing something internally, it's fine to use Claude or OpenAI, but those are really, really expensive. Right. And so thinking about optimizing and building expertise within your AI engineering organization so that they can say, hey, for a constant set of AI evals, how do we get the cheapest price? What open source software are we using? So that's one interesting component. But most people here, they want to know like, okay, how do I deal with this? Right? So we have to take examples from the best, using the best AI players. And the best AI players generally in PLG these days are doing either pricing based on usage or outcome. And what we're seeing less and less of is pricing based on seats. If they have seats, there's a hack to doing seats. So here's the hack, right? You charge overages and you have multiple plans. So we just talked about the example with Claude. Sounds like Wes is stuck on the regular Claude plan, so he ends up having to use ChatGPT. So having that max plan is their solution.
- [32:50]
Wes Bush
Right?
- [32:51]
Akash Gupta
Wes starts to see that pop up. Hey, upgrade to max if you want more.
- [32:55]
Wes Bush
So multiple data too, but yeah, it's definitely great. They get you right when you are at the like maximum threshold of a thread and then it's like, oh yeah, I, I'm curious about that question.
- [33:06]
Akash Gupta
You finally trained the AI to understand your intent and then they're like, you can't talk to it anymore.
- [33:12]
Wes Bush
Perfect time to upgrade.
- [33:14]
Akash Gupta
Yeah, exactly. So the seats model works either with multiple plans like that or with overages. So that's what Warp.dev does is they'll have like a certain amount on each plan and then they can charge you overages. Actually Lindy does the same thing, right? So they have a certain amount on each. If you go over it, they'll start to charge you that there's usage based, right? Usage base. This from people who haven't studied PLG as much as us. They may not even realize the history of it. Usage base started way back in the infrastructure days, like AWS and stuff like that, because they just had to price on usage. But you know which companies have the best margins and which Grow the most sustainably are those infrastructure companies. And so all the end layer application SaaS companies are now adding in usage. And then the final example is outcome. So the most canonical example you guys have already heard about it, but I'll just refresh your memory is Finn from Intercom. Right. So Intercom had built like 300 million ARR business on customer support software and then they disrupted themselves by creating an outcome based thing that basically says like if we resolve the customer's complaint, you pay us then with this AI. And so that's outcome based. And I think what we're seeing is like to deal with the cogs, teams are moving from seats to usage and ideally outcome if they can define it. So it's like if you can define outcome, go to for outcome. If you can define usage, go for usage, otherwise go for seats with overages in multiple points.
- [34:44]
Wes Bush
What other examples have you seen other than fin that have really nailed the outcome based pricing? Because I can't think of too many myself and I'm like, you know, like you said, I think that recommendation like if you can't think of an outcome, go to usage. If you can't think of usage, go to seats. That pecking order is perfect. But any outcome based ones that you've found that are really, really good.
- [35:05]
Akash Gupta
Yeah, there's a couple different elements. Let me just pull some of these up. Harvey AI is an interesting one. Have you heard of this guy? These guys? These guys I think are, they're worth like two or three billion dollars now. It's like they took the pricing off their page, which is smart. So you have to talk to them about this. But they basically charge for brief drafting or per contract reviewed. So it's like a legal work output, not effort. Another one that people probably have heard of is Jasper. Have you seen this one?
- [35:36]
Wes Bush
Yeah, yeah, definitely.
- [35:37]
Akash Gupta
So for some of these tiers you're going to pay like per published blog post. So it's not per generation but you actually used the campaign asset. Another one that a lot of people probably have heard of but maybe haven't looked at. The pricing is 11x AI. So you'd pay for like per qualified meeting booked, not per email sent or something like that.
- [36:01]
Wes Bush
Okay.
- [36:02]
Akash Gupta
It's a sales agent. So yeah, I think there's a lot of good examples out there. We'll just put those in there so you guys don't forget 11x fin. Like these companies, these are really the ones winning. Like the cool thing about outcome based pricing is how fast you can scale. Somebody like Toyota might try you at 1%, and all of a sudden the outcomes work and they might just ramp you up all of a sudden.
- [36:26]
Wes Bush
Yeah, I think it's like it goes back to how do you align your value with what the customer gets? And it's brilliant because everybody in the company's like, hey, we need to book more meetings for our customers. And it's like, great. At the same time, we're making more money. It's not just boosting usage, it's the outcome they care about 100%.
- [36:47]
Akash Gupta
And that goes back to the core lessons that Wes teaches about product LED growth. So then we go into the model, Right? So there's a couple different things I'm actually curious. How do you think that the AI model, and here we're talking about, like, you know, free tiers, free usage. What is the most important impact that people should focus on?
- [37:06]
Wes Bush
Yeah. So before we, like, ever decide what to give away for free, the user research is really important of like, hey, what is the end outcome? What is the. If we could break that down into, like, three levels, what are those beginner problems that they have? What are those intermediate ones and the advanced ones? And then those beginner problems that everybody in your total addressable market has? That's where I love to focus on that free model. And like, oh, we could give away that stuff for free. Like at Vidyard, the free model is like, hey, make it really easy for people to make videos. Here's the Chrome extension. You record a video, you could send it to anybody, but you'll also get the insights of did they watch it or not? And then if you choose to use the marketing tools and analytics, then that's the next natural step, is like an intermediate milestone.
- [37:54]
Akash Gupta
Yes, well said. So the way to think about these models. Right, is that the reverse trial is really, really common these days with AI. So like we talked about, the cogs for AI are really expensive, but you want to get people to that AI assisted activation moment. And to get people there, you generally need to put them on a premium plan. And so that's where the reverse trial comes in, is you reverse trial them into the premium plan to start, and then you say, okay, you're out of AI credits. So you guys have probably all seen this before, right? They basically give you a certain amount of AI credits and you put those in. So before, we used to always think about, hey, like, I have my free tier, I have my free trial. Now people just think about, what are my AI credits that I'm giving away for free. So that's one really interesting change. Another interesting one is I've seen, I don't know if you've seen this in practice as well is many more people are interested in like free trial with credit card because there is that cost here and if the person isn't willing to put in their credit card or you can even add in sometimes what I've noticed these days is the free trial with qualification. So the qualification would be like they actually registered with a business email for instance, not a Gmail address so that you know that they might eventually actually pay for you. And so basically the cogs have changed the dynamic where our free plans before we used to give away so much like every month we would just refresh more and more and more and more and more credits. Now there's that. Now there is I think a really interesting. So you know who, I'm actually curious, who do you think are like the best PLG AI companies?
- [39:40]
Wes Bush
I mean like I think the LLMs are crushing it. We all know because they have that. But you know the part that I would say is like a big caveat on this is most of them, the ones you feel like, hey, these are the amazing ones that everybody knows about, they are like so unsustainable. Like they are just burning a bunch of cash giving away like amazing, amazing value. And so that's the part where I'm like, you know, the business owner of me is like how long is that going to last until the VC funding dries up or some of them die out and they're just going to have to either put you on a older model that's much more affordable or cost cut on some other ways or what's.
- [40:24]
Akash Gupta
Your take on exactly? Like there's a cool model about like how much margin lovable is making? It's like 18% or something. The higher up you go on the wrapper layer, the less margin you make, the foundational companies make the higher margins.
- [40:36]
Wes Bush
Right.
- [40:37]
Akash Gupta
So I think like if you are a rapper, right let's say like Cursor. So I would say Cursor is probably one of the most successful AI PLG companies out there. So we really want to study their model model they actually give you basically an IDE that's pretty good for free unlimited usage. So let's say I pay for Claude Max. So I use Claude code inside of cursor free. And so I'm a cursor user but I'm a free user. And so they have like hundred, probably tens, hundreds of millions of active free users. And so you create a free plan without a AI. And so all these points we're talking about in the model, basically the insight is AI costs money, Frick. So we need to solve it in one of these five ways. All right, let's move into expansion then, shall we? What would you say are like the biggest impacts PLG to expansion?
- [41:29]
Wes Bush
If you get your pricing right, let's say it's like we were talking about price per outcome. Expansion naturally happens. So I think that's the fantastic part. There's a lot more data that AI can go through and find, okay, which are the accounts that we should expand. That's like the, the basic infrastructure layer. Get that, get that in place.
- [41:50]
Akash Gupta
AI assisted PQLs.
- [41:51]
Wes Bush
Yes, absolutely. And then there is just the natural expansion of the pricing. But what other specific ways have you.
- [42:00]
Akash Gupta
So I think the AI assisted expansion is a really important topic for people to understand. So basically what happens in AI assisted expansion is like you can use the AI to figure out, hey, there's a hundred PMs affirm. Let's say that's where I was director of growth product and 30 of them are using it really well and 70 of them aren't. And it looks like the senior PMs are the ones who are using it the most, not the product leaders. So I can associate that of the 70 people not using it, these 15 are the senior PMs most likely to want to be using it. It not just target them, but then actually come up with the use case. So let's say the use case for my product again affirm.
- [42:48]
Wes Bush
Right.
- [42:49]
Akash Gupta
For a product like a product board, let's say for PMs would be putting in my roadmap items. And the moment is really when you connect your roadmap items to user insights that productboard collected from somewhere else. Right. So then the use case might be, hey, for these 15 senior PMs, I'm going to find a customer insight that would be useful for each of them and I'm going to suggest it to them in a targeted expansion email. Do you see how powerful that is? And I might send this email to the admin as well. And this is the magic of AI assisted expansion. And a lot of companies are running this playbook right now and like literally their expansion will go from let's say 110% as their average per year to 120, 130%. So it's really powerful.
- [43:36]
Wes Bush
You can see that now. Is there any tools you found that are the best at helping out with.
- [43:40]
Akash Gupta
That one of the a lot of the same tools you guys are used to actually like in the PQL space and stuff like that. So there's like pocus and other tools like that. All those same tools are building in this space. Although most of our tools are going to go on the left side. So other thing that you want to think about with expansion. Right, right. Is how do you make these triggers. So basically what you do is you like improve all your triggers. Like most of you guys have a bunch of transactional lifecycle emails and calls that you take, customer success calls you take with the account owner. You improve all of those with these AI assisted expansion. So actually this should go here. Yeah, so there's expansion. Let's move on to go to market. Market. So this is probably the area most disrupted by AI. What would you highlight for folks here?
- [44:30]
Wes Bush
I mean go to market to me is quite broad. So it's like everything you do to grow your market, get more sales, whether it's marketing, sales, even CS as well could be considered part of that. And then I don't know exactly where the specific part and when we look at the AI on the job side I'm like impacts everything as far as each of these, like you mentioned the marketing, the sales should each be AI enabled at the very least. But as far as the workflows, I think there's just a lot more tools that now enable it to be like that's the CSM specific AI tool, the marketing specific one. And but the part where I have not seen and curious to hear your take on this too is any tool that does all of it together where you can get the best like this is your 1 record for your go to market motion and all the tools kind of come in here and you can get a good picture of how everything's going because like I see see all the SaaS tools, if they're going to survive, they, they add AI in some capacity, they make it more useful. But yeah, that, that base layer of like your go to market in one tool that's non existent.
- [45:38]
Akash Gupta
Yeah, I mean you can use Salesforce is marketing sales and support clouds if you want. And that's like the closest you're gonna get to like a single tool that can do all of this. But the vast majority of people I see building this correctly are using like like 17 different tools. You know they have GTM engineers like just totally crushing it on each of these. So like let's start with marketing. Like you can do like 10x more marketing with all those agents we just talked about. So people are basically like absolutely crushing it with the channels that they're doing. So a couple like teams that didn't used to run webinars are now running webinars because they have an agent like setting everything up for them. Teams that didn't used to write SEO. Have you heard of this tool called outrank?
- [46:25]
Wes Bush
No, I haven't, but I'll check it out.
- [46:27]
Akash Gupta
So outrank will literally write LLM article for you every single day that's like fully optimized for SEO with like videos and images and infographics and FAQs and link backlinks from their marketplace. There's tons of other SEO tools out there, but basically you should be doing 10x more marketing. Whether that's webinars, SEO, a lot of, a lot of people are under indexed on employee LinkedIn. And so AI can basically help you like scale this insane amounts because now you can write all your employees LinkedIn posts and track them really easily using AI. So basically you want to do like way, way, way more marketing. Like one of the things that a lot of people are doing right now is they're using VO3 and nano banana to create ads and they're hiring agencies that are like specialists in this to create like insane ads. You could actually create like totally viral ads. Like the biggest companies in the world, Lexus, Reebok, you name it, they're hiring these people and using that strategy. And the reason it matters for product led growth is like we were always in product led growth kind of about that volume game of free users.
- [47:42]
Wes Bush
Totally.
- [47:42]
Akash Gupta
This is like enacting it at scale.
- [47:45]
Wes Bush
I always think about like, what do product led companies have to really master? It's like all the low cost channels like master, master them. Because if you can do that then I mean usually your lifetime value is a lot lower than like a sales like company. And so they're always going to have the like the trade shows and everything else kind of maxed out a lot of times. And so you can definitely own those specific channels before they can. And yeah, I love what you said about SEO, webinars, employee LinkedIn accounts and everything else there too. That's awesome.
- [48:16]
Akash Gupta
Yep. So then sales, like sales is also undergoing like a 10x with AI, right? Just finding leads, writing customized emails. The coolest one I've seen recently is like customized videos. Have you ever seen, have you seen these avatar companies?
- [48:32]
Wes Bush
Yeah, I mean video art's going in that direction as well with the avatars because I think people don't like the way they they look, it's not like a perfectly presented, like, oh, I didn't quite do my hair today. It's like, well, with an AI avatar of yourself, you can always look perfect.
- [48:47]
Akash Gupta
Yeah, this dude is literally getting like hundreds of thousands of follows like every week or so when I look at him. This is an avatar, you know, like this, this bodice.
- [49:00]
Wes Bush
Are you talking to those people or no, all he did.
- [49:03]
Akash Gupta
Nope, he just recorded this audio. Like, not dressed up nice. Not in this beautiful studio. Yeah, he's just like in his desk. But look at this. These are all avatars. Like, look at how successful these posts are. Totally. So you can absolutely crush it with AI avatars. Don't let people say they're not like good enough yet. They're easily good enough if they're fooling millions of people every single day on Instagram. As I just showed you, they definitely work. And so like your outreach can just be so much better. And like what I've seen is that people create AI to analyze their emails and then write more. And so like, you know how before you'd add like a hundred people to your email sequence? Now what you do is you add it to this agent sequence that sends it to like five people. It looks at. As soon as it gets a signal that, hey, somebody opened or clicked at it, it looks at says, oh, it has like another pixel in there to see like what they clicked on in the thing. Oh, they really care about this. So it uses that signal to analyze and iterate on the email and then improve it. And so these are just amazing workflows. You can just get AI to do it all for you though you're not doing it.
- [50:11]
Wes Bush
The best tool you found for that on the sales front, Apollo for sure.
- [50:16]
Akash Gupta
Another really good one is called Instantly and another good one is LEM List. So if you use like those three tools, you'll be pretty good. So now we're going to talk about data. Tell us more about data and AI.
- [50:27]
Wes Bush
Yeah, this is the reason I asked is because there's usually like, I haven't found too many good data platforms out there that will analyze and tell you some interesting insights. Like we, for our clients we use mixture between, you know, mix panel, amplitude, post hog, depending on budgets. But yeah, we, we do all the analysis for them of like, hey, this is your biggest bottleneck and going from there, but we gotta improve that for sure.
- [50:54]
Akash Gupta
Yeah, I think like there are some next gen product analytics tools out there. So you mentioned post hog and amplitude and stuff. There's some more AI forward tools. So Chameleon and Statsig, Statsig was just bought by OpenAI are two interesting ones to think about for your product analytics and experimentation in particular. Like I would say two interesting things around experimentation. So one is prompt based experimentation. Have you heard of this?
- [51:23]
Wes Bush
No.
- [51:24]
Akash Gupta
Okay, so this is.
- [51:25]
Wes Bush
I know exactly what it is.
- [51:27]
Akash Gupta
Every like. So we just showed you guys the bolt example, right? Right. What if you could just now instead of hitting publish, you just hit experiment. That's what you can do with prompt based experimentation nowadays and especially when you're doing a front end change that doesn't affect the backend. You just send it over to your on call engineer, they review it in two minutes and it's alive. Like it's, it's just amazing. So prompt based experimentation and then the other thing I would say is like these bandit algorithm, have you heard of these?
- [52:01]
Wes Bush
No.
- [52:02]
Akash Gupta
So what these do is as you're getting more signal, let's say you're running like two variants, A and B. And A is really starting to crush B. It'll more quickly start to allocate traffic to A than B. It'll keep leaving the experiment on. So you eventually get statistically significant results. But as it gets signal it'll even faster shift it. So let's say like you're doing a big marketing push or something like that, like you're paying a bunch of influencers to post about you. You want to make sure that it doesn't take two weeks to shift to the winning variant. Right? You want to swift to the real variant in real time on that day. So a bandit algorithm can really, really help you with that. Then there's the other element of data. So you can use AI data platforms. So like you just said, it's so hard to get out all of the data from whatever people are doing. So what people are doing is they're throwing all their data into like a data lake or warehouse, like SISU or, or even Snowflake has this functionality now. And with AI making it so much easier to just write your SQL queries, you can connect together like 17 different data sources like your CRM, your marketing data from WordPress, your PQL data from Pocus, your sales close lost data from Salesforce on one data platform that writes the SQL for you. So let's move on to team. What should people know about teams in the AI age for plg?
- [53:30]
Wes Bush
I think the big part that you're unpacking here too on the the left side of like AI jobs as well is there is a lot even we just focus on content. Like your average content marketer. It's like, well, do you, you still need them or do you need more of like the strategist that understands the expertise and that's like, okay, that is probably more valuable than somebody to just write content. So I think a lot of the lower level tasks, it's like that can be done by AI now. So you probably have to look at your team and be like, do we have the right people in the right seats? Because could that completely be done with AI? Maybe not today, but in a year or so? Absolutely. So I think that like GTM engineering, that's, that's probably why we're seeing that a lot more. Because it's the engineering of it all that once you set it up you can get 10x more output with a fraction of the time team.
- [54:28]
Akash Gupta
Yes, exactly. So and the teams are becoming a lot smaller, like probably like half the size, you know.
- [54:34]
Wes Bush
Yeah.
- [54:35]
Akash Gupta
Across the board.
- [54:36]
Wes Bush
Back to what you shared with Jacob on his, you know, kind of like org chart with like whether it's the EA or the marketing team that's going to become way more common when you're designing your team. It's like, yeah, that's a seat on a team meeting. It's a vital function, vital job, but it's not huge human. It's the AI agent that's just going to be doing that and we're going to consistently improve it. But yeah, that'll be much more lean.
- [55:03]
Akash Gupta
Yep. So that's the team and then what about the strategy front? What do people need to know there.
- [55:09]
Wes Bush
From like AI perspective? I think it's just becoming way more commoditized like every single market. I think SaaS is about to, despite what many people think, I think it's about to boom because there's a lot more micro SaaS out there. And so how do you stand out and really develop hard to copy modes? I think that's something more and more companies have to think a lot more of and develop disciplines around whether that's just the CEO or the leadership team of like how do we actually stand out? And one thing you mentioned on the other side was like the competitor awareness, like that's, that's important too. But are you solving the problem with the best approach? That should always be the top question. Question.
- [55:55]
Akash Gupta
Exactly. So features are now commoditized, taste is the differentiator and there are hard to copy modes like data and distribution out there that you should really be focused on. So this is the high level view we've just walked you through literally everything you need to know from activation, pricing, model, expansion, gtm, data, team, through strategy, on how PLG is affected and then on your own. So decide how the job is affected. So this is AI XPLG in a slide.
- [56:26]
Wes Bush
I love it. And where would you recommend people start? Because they're, they might look at this and be like, oh my goodness, this is overwhelming. There's way too many things I should be thinking about. And I too like, I try and stay up to date on AI stuff and I'm like, ah, I still got about like five or six tools here I gotta check out after this. So what's your take on where to get started?
- [56:48]
Akash Gupta
Started? Yeah, so on the left side like pick one tool a week that we've listed that you haven't tried yet and add it to your stack. So you know, probably you guys should all start out with dictation because most of you are probably not doing that. Then you should move on to agents. Then you should move on to like a copilot. Then you should move on to like a prototyping tool. So then probably like a sole context LLM that'll be like your five new tools to try over there and then on the PLG side. So you need to look at of these layers, activation, pricing, model, expansion, GTM data, team strategy. What are my weakest and strongest at? So a lot, lot of you guys, you feel like, okay, I'm pretty strong on my activation, my model, my expansion, my gtm, my data, but it's really my team strategy and pricing that I need to fix. Right. And so then you go into those specific areas on your team side you think about, okay, do I have the right people who are using AI on the strategy side, have we built a hard to copy moat? And on the pricing side, how are we going to move down this if we're stuck at seats, how are we going to move to usage? If we're stuck at usage, how do we move to output outcome so that we can optimize our cogs? So that would be my sort of roadmap and starting points.
- [57:55]
Wes Bush
Totally. Yeah. I keep looking at that like time to value 1/10 the time and I think for everyone asking that is like that's going to become the expectation in your market regardless of if you think it is possible or not. Somebody's going to figure out how to get people to value in 1/10 of time. So best be you disrupt yourself. Yeah, absolutely. And so how do you stay up to date on everything on the AI side? What's going on in PLG as well? Because the way I see what you do is like you are just a absolutely fantastic learner. You're always testing new things and really refining your approach on that end, which is really hard to do. Like, I'm sure you spent a ton, a ton of time like going through that and picking up all these new things. But what's your like overall approach strategy to staying at the forefront of this?
- [58:43]
Akash Gupta
Let's just keep it really tactical for you guys. Actually. Let me just keep, let me, let me walk through specific things you guys should actually do. So the first thing you should do is you should follow like really good people on X. Like X is the place that has the best AI knowledge out there. And so there's a couple specific people that do really, really good job. Andre Karpathy, everybody's heard of him, right? Right. But are you following him on X? Even follow like his replies? Even his replies have a packed with knowledge. Then there's this guy, S W Y X. He actually runs the AI Engineering conference. He's just like a legend. He posts amazing stuff. Then there's this guy, Greg Eisenberg. Again, you guys probably heard of him, but are you following him? Are you checking out all his videos? He just showed you how to master going viral with a 55 minute video video on X. So X following some really good people like that. The next place is going to LinkedIn and following some really good AI people. So you have. He is right at the top of my feed Omn. I was literally going to go search for him right now. Amazing AI agent content. Then you got Ruben Hasid, really, really good prompting and tools content. And then you got this guy, Jake ward, really good AI SEO content. If you want like one step above for entrepreneurs. Chris Donnelly doesn't only do AI, but he does a lot of good AI. Yeah, he's ranking the AI models today. Right. So they're generally talking about AI. So you go to LinkedIn, the next place you go to is YouTube. Right. So on YouTube you're gonna get some of the most insane AI content that you'll ever see. And you'll also get Ms. Rachel toddler videos. And one of the best channels that probably people haven't heard of yet is this guy, Matt Wolf. Wolf just does the best. You know, he's just breaking down ChatGPT Atlas in a 30 minute video for you. Then another guy, Nate Jones, he's like the gray beard of AI. Really, really great takes on AI, Very high value takes on AI and Then this is probably the most underrated channel for AI out there. Actually, it's my friend Peter Yang who creates these really cool AI tutorials out there and just has some pretty sick Claude code, beautiful designs, AI agents, cursor tutorials. So YouTube's the next place. The next place you want to go is to substack and so on. Substack, yeah. These are two really good people. Powell Hearn, Nicholas Cole, they're always writing about AI. So you can see my feeds are all tuned in to the good people to follow. And so you're, you're gathering this information from these four sources. That's step one. Step two is you're just spending time trying out the tools. So basically anytime a new tool comes out, chatgpt Atlas came out. I put it on my calendar as a thing. 15 minutes. Try ChatGPT Atlas. And every week I try to pick up a brand new tool if there wasn't some news that caused me to. So I just have a tool list right now. Like personally in my mind I want to get better@make.com. right. I haven't used make.com very much. So the next time there's a week like that where there wasn't a tool based news item, I'll just go ahead and pick up one of those that I already have on my list.
- [62:02]
Wes Bush
Yeah, I love the time boxing. Also having that weekly rhythm of like, I'm going to learn one new thing every single week. Do you have like, let me guess, your own roadmap of like the pecking order of things you need to go.
- [62:14]
Akash Gupta
I just have it in my head, you know, like, these are the tools I still want to try. And you can sign up for like Lenny's product pass or something and he'll give you access to like 30 tools and so that'll give you a list of, hey, I have these tools, I better try them.
- [62:27]
Wes Bush
So yeah, man, I could totally see you doing something similar like that with your newsletter because you guys like AI tools here of like the product and growth stack. I think there's something there for sure. It is.
- [62:38]
Akash Gupta
Yeah, it's in testing right now and we're signing up for final people. But yeah, it's coming out in a few weeks.
- [62:42]
Wes Bush
That's awesome. So when people listen to this podcast, they have to go and check your newsletter out and then hopefully some of those tools are available too. But yeah, it's coming.
- [62:53]
Akash Gupta
Yeah, some of the ones we showed like magic patterns and Warp and V0 will be part of the bundle amplitude.
- [63:00]
Wes Bush
So yeah, yeah, and to like wrap up. One question I have for you to just kind of munch on is what you mentioned. Like the moats are changing. Like everything can be copied a lot more. But taste, taste is something that's kind of vague, but it's something people know when they have it. So can you maybe share your take on that? Like, why is taste becoming a lot more important for any product that founder to really consider and invest in as well up for their business?
- [63:32]
Akash Gupta
Well, let's first break down what is taste? Right, so taste is the ability to only show people like what they need to see. So I'll give you an example of some app that you guys are all using that has some taste. What are we looking at here? There's nothing here. Like this is not like any of the average SaaS apps that you look at. It's just a blank white screen. ChatGPT has decided all they need to show you is a chat box. That's taste. Right? Taste is about actually like, like instead of giving people a hundred things, giving people one or two, creating a clearly sculpted journey through your product that layers in the right elements of your product at the right time. Taste is also having a zero tolerance for bugs. All your SaaS products out there listening are littered with bugs. Littered with bugs. And as you build more and more features, you are adding in more tech debt and bugs. And there are so many because you're not using your product like a power user that you're missing thing. There's like so many. Yesterday I was testing, you know, so chatgpt, they might have some taste on their homepage. They actually don't have amazing taste throughout all of their product teams. If you look at their create image. Right, let's create an image. Like let's create me speaking on stage to a thousand people from left side cine quality. Exactly me. Okay, so maybe I'll upload like a little photo of myself as well. Well, let me pull in a photo here. Okay, so we've added in a photo of me. Let's see what happens like with this. So taste would be really paying attention to what is the user experience when where they're spending the most time. So right now I'm spending a lot of time. They don't have like any sort of loading experience or anything. Okay. Finally it comes up. I'm not sure if I hit thinking, will it skip? Okay, I'll risk it. So now this counter is already at 1 minute 20 seconds, but it hasn't been generating for 1 minute 20 seconds. So what is this counter doing this thinking it isn't showing to me.
- [65:26]
Wes Bush
Right.
- [65:27]
Akash Gupta
So there's all these elements where just because a company is good at taste in one area doesn't mean they're always tasteful. The same is true with your product. You need to find the areas where you are lacking taste, where there are these rough edges in your product, your user experience and you are squashing them out. You know, one of the most interesting companies, they just crossed a billion dollars in revenue is Linear. They're a perfect example of this. They have a zero tolerance for bugs. They basically solve all their bugs. If you as a customer reach out with a bug, they'll like solve it day. Like that's how many engineering resources they keep on bugs. Like even if you're the only person in the world experiencing that bug, they'll go solve it right away. And so they're an example where they have a very simple feature. Their entire product is task management. You could probably vibe code the basic version of their product in Bolt in 10 minutes, but you're not going to replace them. And they're going to hit a billion and two billion and five billion and ten billion. Sure enough. Because they have that craft and take.
- [66:22]
Wes Bush
Yeah, I love that. I think that's the perfect way to wrap up too because it's one of those things where it's. I think if every product that founder really focus on that, that is something that helps you differentiate in your market and that's also hard to copy. Like you said, not many companies, they can copy Linear and their products, but they care obsessively about that on the taste front. So is there anything else you feel like we talked about today that you want to recap on, remind people of to kind of pay attention as we are entering the AI PLG world where you have to use AI if you're gonna stay ahead?
- [66:57]
Akash Gupta
Yeah, just don't even like I know everybody is basically like hates the word AI now. Like, you know, AI is like the computer or Microsoft Excel or do you guys remember when everybody switched from Outlook to Gmail? You know, AI is just the better. It's the next evolution of technology. Instead of thinking like how do I stay up on AI or I'm feeling AI FOMO or somebody's beating me on AI, think about how am I staying on the latest edge of technology. And when you think about that, your answer will always be AI in the end.
- [67:29]
Wes Bush
Totally. I love that. And I think what we'll. We'll also link to a lot of the people you mentioned. As well on Twitter it's X now. I keep saying Twitter But Die Hard, LinkedIn, YouTube and then Substack as well so people can follow and then like you dedicate that time every week to at least learn one new thing, try new one new thing is well to stay at the cutting edge as well. So this has been awesome. Thank you so much for coming on the pod. This has been a ton of fun and yeah, looking forward to making sure we keep you coming back on a regular basis because this was fantastic and there's so many tools here where it's like there's the 2025 version we've got. We gotta do one next year and see what has changed as far as the different tools and the workflows, agents and everything else. So we'll I think start an annual tradition here. So thanks Ken for coming on.
- [68:21]
Akash Gupta
Awesome. Can't wait to compare now next year.
- [68:26]
Wes Bush
And to wrap things up, thank you everybody for listening to this version of the product that podcast. Make sure to rate review this on wherever you listen to podcasts, whether it's Apple, Google, you name it, Spotify. I'm going to read every single one of those views and that's how I know how to improve this. Also, if you want to stay in contact with being and learn what is going on in the world of PLG G and every single week get the best actionable deep dives on product led growth. Make sure to head on over to productled.com forward slash newsletter. I am personally writing each of these deep dives every single week and you're going to get a ton of it. So make sure to head on over there to product led.com forward slash newsletter.