Gonto: I'm working a lot with customers now on focusing on the new era of developers, which are maybe 20 to 27. I've done a lot of interviews with them, and I hear so often that they just search stuff on TikTok. Like they don't go to Google, they don't go to ChatGPT, they just search on TikTok.

So I think doing these kind of things is pretty interesting. And Cursor, I know because a friend works there, told me that they are doing ads on Instagram Reels and everywhere. So if they are doing them, why shouldn't you?

Hank: This week on Code to Market, you're not doing enough on Facebook, all your SaaS is dead, and a billionaire is scared of Twitter. It's going to be a great episode. You ready, Gonto?

Gonto: Let's start with the first one. You go. Tell me all the first ones you found.

Hank: Okay, so somebody sent me, shout out Cynthia, she sent me a Facebook link of LeeRob on Cursor's account. And it's an ad, and it's LeeRob doing, yeah, we've talked many times about the like, great dev rel is expanding people's imagination. It's this, you know, this short video, a reel, but it's on Facebook, of him basically building a game that he could use a USB controller for with Cursor.

So he's showing the product, expanding people's imagination for what's possible. Here's why we brought it up. This thing has over 8 million views.

It's got hundreds of likes and comments. And just made me think, man, this is all the AI companies, except apparently Cursor, are obsessed with Twitter. Everyone's obsessed with Twitter.

People won't even post on LinkedIn. But here we have Cursor, which is now at 2 billion ARR, posting stuff on Facebook. So it just made me think people need to get over their pride on what channels they're marketing on and do more.

That was my thought.

Gonto: Yeah, a couple of things on it that I think are interesting. Number one is the style of what was recorded is prepared for mobile phones. Like it's clear, it's like a reel.

So I'm sure they didn't just post it on Facebook. But also that they posted it on Instagram. 

Hank: And TikTok and YouTube.

Gonto: Exactly. And the point is that it's not just Facebook. It's rather than if you build a short form video, just post it everywhere.

Like you have nothing to lose. You post it on TikTok, on Reels. You post it also on Facebook and you just send it everywhere.

And I think I'm working a lot with customers now on focusing on the newer developers, which are maybe 20 to 27. I've done a lot of interviews with them. And I hear so often that they just search stuff on TikTok.

Like they don't go to Google. They don't even go to Chachabitty. They just search on TikTok.

So I think doing this kind of thing is pretty interesting. And Cursor, I know because a friend works there, told me that they are doing ads on Instagram Reels and everywhere. So if they are doing them, why shouldn't you?

Even look at Anthropic. Anthropic has done a fantastic job on Reels on Instagram. Not just beat by them, but also sponsoring people.

And something I heard that I think is fantastic is Anthropic doesn't just pay people sponsorship for doing a great Reel. They would actually pay up so that they can find better locations, better cameras and stuff like that. So the video that they do with that influencer looks better than the rest.

And therefore, maybe they get more attention or something because of that. So if Anthropic is going all in on this, why can't you just at least post something?

Hank: Yeah, for real. There's so many little other things here too, which is... So at Vercel, because everybody...

I talked to so many founders who were like, we need to find our Lerob. And Lerob was really good at creating content at Vercel, but he creates so much more at Cursor. And I caught up with him recently and we catch up every once in a while.

The reason he's able to be so much more prolific right now is because he's basically an IC. He does have one person on his team, he's pretty autonomous. But there's an interesting thing here of when you have a really productive person, sometimes you push them to build a team and start running a huge function.

But there's the other path of like, no, just keep that person on their IC path and help them just be more productive and remove distractions and friction and meetings and everything from them. Because this video is like... I mean, I think just this video alone is like a better little piece of content than almost anything he did at Vercel.

Gonto: In general, first of all, it's more catchy. So I actually think he did a good job at Vercel with the posts and stuff. It was a different time where people read more blog posts, people don't read as many blog posts.

I think it's a different vibe. But I do agree that you shouldn't row your teams as much with this new AI era. And I'm going to use this to hook us to the next topic, which is that Roj G, Toby Sherman from Vercel, did a tweet that I've been thinking about so many times, where he talks about the SaaSpocalypse.

And he says it's more overstated and understated by people. And his whole point is that it's basically overstated because the key system of records of storage, Salesforce, Snowflake are all there, but understated because the software we're generating is more beautiful, more personalized, basically fits better our business. And this is something that I think is important for our listeners from two points of view.

One is how much should your marketing and product team be using and building new AI apps and how they think about it? And the other one is if everybody's using you from that new point of view, what are the things that you need to change from your marketing perspective, from your product perspective? I have a lot of thoughts on this topic.

I want to hear first from you. Let's talk about the internal side first. Like what are your takes on the SaaSpocalypse on how marketing and product teams are building and using new AI apps?

Hank: Well, the teams aren't doing enough yet. Vercel is definitely going to be ahead of the curve here. I know shortly after I left, our friend Trey, who just announced he's at Resend, another good catch for you, Zeno, he helped build, he's basically the PM of replacing Salesforce's quoting and pricing product that Salesforce has.

So they already had a history of like trying to build custom software internally. That was really hard. Like that was a big project.

It saved them, you know, like six figures, but it was a lot of work. And now I think they've just taken that as a lesson of like, oh, we can do stuff if we have the right person, like kind of PMing this and they're building more. And you can see if you follow some of their team members, like this one guy, Drew, is an interesting follow.

And they're trying to build a lot. What's also interesting is the respect, because trust me, from day one, I'm the one who implemented Salesforce at Vercel, at day one. And figuring out that object architecture to make it fit with Vercel was a nightmare.

I did, I think, a pretty good job for a solo RevOps guy at the start and it worked. But as the price kept climbing every year, every year, Cachermo got louder. I'm like, can we just build our own?

Like this UI sucks. It's so slow. And the thing is, like, well, it's a great system of record and it's something sales reps are familiar with.

How would we replace it? It would cost so much to replace it. So many engineers.

The price has changed. And so it's interesting, though, that he's saying, we're still going to use Salesforce as the system of record, at least for now. That way, there's still all the familiar UI.

If you're an AE joining Vercel, that's going to help you do things smoother. But then adding all the extra stuff that's going to be more performant and faster and better suited for Vercel's internals. I don't know.

I think the examples are endless. I agree. One of my customers is Factory AI.

Gonto: They build Droid, which is basically a competitor to OpenCode and CloudCode. And because they built that, they use it internally all the time. And it's the most advanced team that I've seen on using these new interfaces.

In their case, they don't build as many new apps, but they do use it for everything. To give you an example, they have a data droid that basically connects to skills that they wrote about different tables and their connections. And they don't have postcode.

They literally read something that gets events data from the web, both page views and events, send it to Data Warehouse, get other stuff from other services to Data Warehouse, and they just use that. And what they do is they ask it for things, and they build charts with HTML. And when there's one that they want to keep, they basically post it to an internal service behind SSO where they can have those.

They do the same, for example, with blog posts. Every time they're going to ship a new launch announcement, they get droid to read the internal PRD, the internal docs, and everything absolutely from that's internal from an engineering perspective. And then for the blog, they basically just use whisperflow, where they're like, oh, I want to talk about this.

Oh, maybe not, maybe better about this. And they just run it for an hour. And then they get a pretty good blog post just because they used all of that information.

And they have a skill that then humanizes that, make sure that they first understand the docs. And then they use that information to build all of the other things. Last example I'll give you.

Similarly, they still use Salesforce because they're AEs like the UI, but because they hate using Salesforce, like the marketing team and the product team, they actually connect through the CLI in droid to it. And with the CLI, they create campaigns, they create objects and then do everything just from the CLI. And they have a skill that basically understands the CLI, your objects, and basically can get things done.

What I think is interesting from this is it gives new companies an idea of what the future might look like. And I think it gives the opportunity to either Salesforce to change how they think about things where they're like, okay, some people will continue to use my AI, my UI, who are old school people, but a lot of the new ones, we need to build a really good interface for connecting with us. They need CLIs, they need MCPs, they need skills, like they need to build all of that, or there will be a new vendor that will kill them.

And I think the new vendors will be focused on systems of record and saving data and having good interaction with AI more than anything else.

Hank: And RouchG specifically called out legacy systems with poor, slow, outdated, and inconsistent APIs. And by the way, I have a client who never bought PostHog. They built their own event tracking and we're getting the conversions in the paid ad platforms.

And it works better than I thought it would when they first told me about it. I have another client strongly considering building their own marketing automation platform, which is, that's the one if you had asked me six months ago, I would have said, don't touch that one with the email and like, maybe do CRM first. And now I've kind of flipped on that.

Like, yeah, it's so much as possible now.

Gonto: I think the challenge... And the marketing condition, like we talked about Zeno, like Zeno's product is really good at sending emails, creating campaigns and stuff like that. They have the right APIs.

So maybe you can do a good job marketing automation because you use recent to send emails instead of building your own. But you focus on the rest, on the other part, on either building a better UI or chat interface that is basically completely for your business. Before, I used to think though that I needed a UI for everything and that all teams should use UIs.

So we should build custom apps and custom UIs for us. But lately, more and more, I think, skills and CLIs are enough. And I'm actually really happy to just chat just because I like talking as we have a podcast.

And I do so much whisper flow that I just talk to my computer and it gets shit done.

Hank: Here's a fun fact for you that I remember back from when I was at Trey and we were defining target accounts. The top two spenders at any given software company are always the marketing department and the HR department, almost always. They spend so much on so many tools.

And every single one of those tools is basically designed to give a non-technical person easier access to data and to move it around. Exactly.

Gonto: So there are a lot of things that you need. From the other perspective, I'll give an example from Clerk, like ClerkOS platform. We're thinking a lot about how developers interact with your services.

And for example, their dashboard team now focuses a lot more on CLIs and MCPs because people see less the UI. And we're focusing with them on, before we had Quickstarts where each Quickstart was for one technology and it had a step-by-step guide on how to implement it, how to add it to JavaScript, how to add it to Next, how to add it to Vue, et cetera. Now it's just one Quickstart per action and the Quickstart is download the skill that you need for whichever tech you're using or use the MCP or use a CLI and just copy this prompt, run it, and it's going to be running.

That's it. And then as part of the prompt, it will suggest you to sign up, but the prompt and the skill are the important ones. And in the prompt, for example, one thing we're building, we haven't shipped yet, is a prompt generator where you can say like, okay, I want to add a multi-factor or I want to add a step-up or what do you want to add?

And then the skill will understand that and just create those things and call the right APIs for that. And it will even help you get analytics. So I think if you're building your product, then questions that I will have is one, do I really want the main CTA of my home to be signed up to my product or do I actually want it for them to just run this prompt in your console?

I think people will prefer the prompt. Do I want or do I need Quickstarts that tell me how to add things step-by-step or do I actually need skills, MCPs and CLIs that will interact with your tool? And I think everything from either a marketing website perspective at Docs is changing now.

And if you're doing a developer tool, you have to deeply think about what are the useful primitives in this new world? Do I even want to send people to the dashboard or do I just need them to sign up but maybe with a CLI using a skill and I don't need them to see the UI? At least food for thought.

Hank: Yeah. One more thing that no one's talked about this yet, but this is always a part of the conversation in the old build versus buy combos is maintenance. The thing I haven't seen people have a good answer for on the like, like it's so much cheaper and easier to build now.

No one's talking about, well, how do you maintain and improve the systems? Who's going to do that? The person who is excited about building a net new thing is usually not a person who's also excited about maintaining and updating the old thing.

And, you know, those are overlapping but different skill sets. That's going to be, I think the next phase of this is as people deal with the reckoning of like, okay like we, you know, we vibe coded or agentic engineered or whatever you want to call what you did our way into internal tools. Then comes the maintenance burden.

And that's going to be an interesting thing that I think people will be facing over the next months and years.

Gonto: I think there will be a new team called AI Ops which will own a lot of this stuff where they will own creating the right skills or working with the teams to create them. They will own maintaining them as things get updated. But I think it will be an internal AI Ops team which will be, I think sort of like getting these product ops and rev ops and stuff like that all centralized in a place and just updating.

In the past, I think we decentralized ops into rev ops, product ops even engineering ops in some cases just because the things they had to build and do were huge. But now if you can just use AI and a chat interface maybe we can get them all together. And because they are centralized they can use the data from each other and make sure that everything works even better in a company as a whole.

So I think we might see more of those things too. Yeah.

Hank: I would love to hear, you know anyone watching this thing, please leave a comment on is there an internal tool your team has built or thinking about building? I'm curious what people are thinking out there. Super fascinating.

Honestly, like I, I don't know when I went back to a full-time consulting back in October, November I don't think I'd be working as much but there's just too much interesting stuff going on. So I'm working more than I wanted to, but it's fun. It's interesting.

This is like such a crazy time. 100%.

Gonto: Last topic. Do you want to talk about it Hulk?

Hank: It's more of a small little follow up because we've talked about our mesh and hub code and, you know, kind of the way open code kind of lashed out at him. So he launched hub code. It's not the fork of, well it might still be a fork underneath.

I didn't look that deeply into it. What's interesting about this though, is he because he had backlash on Twitter he only launched this on LinkedIn and YouTube. He's very good at like doing the demos and walking through but I thought it was interesting.

You know, he's a billionaire. He's kind of like, I don't want to go back to Twitter. Those guys weren't very nice to me.

And people on LinkedIn are going to care about this more. They care about the, they care about HubSpot more. So it's interesting.

It kind of ties back in with our first topic of yeah, if you want to launch something like you want something to get out there you can go multi-channel. Yeah. I don't know.

I thought that was interesting.

Gonto: And to me, like their mesh is one of the best I think and most known founders. Like he could have just posted on Twitter again and that was fine. So I don't understand why he didn't want the trouble.

He didn't want people talking. I don't get it. Like, of course, lots of marketers are in LinkedIn but he had nothing to lose on cross-posting.

So if Dharmesh you're listening just post it on Twitter too. Dharmesh is listening. You know, we've got other ass too.

Yeah. I think that's it this week. That's it.

Hopefully you liked the episode and we want to hear from you on the AI side. Like how much are you using AI internally? How are you thinking about how your products marketing does change with AI?

But any thoughts, please do share. Thank you all.