Hank: Yes, I agree on all of that, except emails are not dead.

Gonto: They are still dead.

Hank: Emails are still one of the highest ROI things you can do. I think cold email is deader than it's ever been. Like doing cold outreach.

Okay, I agree with you. I think that's our common ground. Okay, welcome.

Big announcement. We're running back the Code to Market Summit in 2027 and Gonto will actually be there this time.

Gonto: I need to learn how to ski and snowboard, so please, other people who suck at skiing or snowboard come so I can learn with you.

Hank: Yeah, we'll do like a group lesson for the noobs. We're probably going to do some like snowcat advanced stuff for the more advanced people. People were asking me about that.

Yeah, and we're going to get, I sent a survey out to the people who attended last time and a lot of them are planning on coming to the next time. It's a bit early. It's not even June and this is going to be February 2 through 4, 2027, but trying to work it out.

I'll have early bird tickets up and we'll figure it out, but it'll be a good time. We're going to limit it to probably like 40 people so that you can still get to know everybody and it's still going to be a very exclusive group, which was probably the highlight last time.

Gonto: And it's all going to be developer marketers. Yeah, when is the landing page launching with the early bird and everything? Like, give us a date.

Hank: I don't know. I guess there's no reason it can't be up by the time people are seeing this. Like, everyone's a developer now.

I'll have a page up. If you're hearing this, there's a page up at ctm.fm slash summit.

Gonto: Are you going to use Codex or Cloud Code?

Hank: Oh, I'm Cloud Code all the way. I'm kind of a model monogamist. So I just pick a model and I just stick with it.

I don't do all the switching around. I know you, you're a bit of a model philanderer.

Gonto: But I actually, so I work with factory AI. They are my customer and I actually just use Droid where it's very easy to switch from one model to the other, but I switch with the vibes mostly every week or every two weeks. I actually enjoy it.

And I can get it like one to build something and the other one to review it. But you have to try everything in life.

Hank: Nah, I'll let others try for me and when the tide shifts. I mean, I was at GPT. I was all in on GPT until I started using Cloud Code and now I'm there.

So speaking of, I built, this is my first like actual marketing use case that's working with AI that I feel is fun and interesting. And I built completely on my own. I built a Slack bot and it's gonna sound like, well, why would you build this?

But I'll get to the reason. Like it's having positive effects. So I built a Slack bot.

I just named it agent email. It's not creative or anything yet, but.

Gonto: Agent May is a product that exists.

Hank: Whatever. Okay. It's just a Slack bot.

It's not published. You can't go and install this though I could create a publishable version. Point is I can tag this thing in a channel and what it does is that invokes a cloud API.

So it can like reason and think and all that. It's got an application and database hosted on Laravel cloud. So it can store my contact data and then it's got a recent API.

And the reason I did this is because there's a lot of small email plays that me and the events person, Abby at Laravel wanted to do that just take a lot of sifting through CSVs and then we would have to take that and re-upload it to HubSpot and like take all these steps. And also I just thought it would be interesting and fun to build this. So let me give you some examples.

So one of these plays was, oh, we should look at who came the last three years to Laracon but hasn't registered yet for this year. And let's send them an email from our CEO and it could be really personalized. I'm like, hey, you've come the last three years.

Are you coming under a new email or is there some reason you're not coming? How can I change that? And like, that's one list.

It's a small list. It was only like 70 people. And then there were other iterations.

We were like, oh, who's come to the meetups and site events of last year's but haven't registered yet. And we came up with like seven different plays that if we had to do the work in the spreadsheets and then draft a whole new email with the stylings for each of these groups it would have been a pain in the butt. But this way we were able to do them all very quickly and just chat through it.

And the next phase is we're going to go ask it for more ideas. Like, hey, look at the list again and come up with some different plays for us.

Gonto: You had the objective of using AI this year. So congrats on having your first app. I think a lot of the things you're doing now could be extracted.

Like I'm doing a lot of what's not on the email part because I think emails are dead but I'm doing a lot on the data part on it. For example, one thing we did with Factory AI was like, hey, let's look at people who signed up eight months ago and are still active and tell me for those people what was the feature that they used the most on week one, week two, week three, week four and how are they different and why? And we actually building that what we did was we got all of the data to change all of our lifecycle emails and then we changed all of our lifecycle emails which again, I still hate because they are emails but we got the data by just asking questions and then the first reply from the AI was meh because the features were too broad.

So he's like, no, give us like more specific features. This is too broad. And then he did the job.

So this idea of you can ask, ask, ask and re-ask and use the data. I think it's one of the most powerful things we as AI marketers can use AI for.

Hank: So yes, I agree on all of that except emails are not dead.

Gonto: They are still dead.

Hank: Emails are still one of the highest ROI things you can do. A lot of people are like, like once you have a trusted audience I think cold email is deader than it's ever been. Like doing cold outreach.

Gonto: Okay, I agree with you.

Hank: I think that's our common ground. But once you have the trust and this is part of why I wanted to use Resend because they're very good about that and with the domains and monitoring the bounces and even being proactive about shutting down the user aggression before it gets out of hand. But yeah, here's another play that actually we emailed out today related to this, by the way.

We made a little gimmick site and we emailed it out to people today. It's kind of a joke like, hey, here's a letter to your boss. And it's at send, there's dashes instead of spaces.

Send me to laracon.laravel.cloud. And it's like a little joke site that our event marketer was able to spin up and deploy on her own. It uses the cloud API. You can like adjust this letter.

It's kind of a joke letter. It's dear sir or madam, I kindly request for funds to attend this year's Laricon. Like, and these are the type of like, this is fun.

This is like a fun email that people respond well to. And by the way, like those email plays. Yeah, we got a lot of replies on why people aren't coming.

We got a lot of tickets sold that way. Oh, I just haven't done it yet. And I think to your earlier point, the reason why email and certain types of email are dying is because people are putting less thought into them instead of more thought.

And to me, the whole point of AI is that you can actually put more thought and creativity into stuff because you can get rid of all the other stuff. The data analysis, you don't have to spend as much time on now. You can spend more time being creative.

Gonto: I agree with that. Like one use of AI that I've been using a lot is this idea of like using dead time to use AI somehow. And for example, one thing we're doing with a lot of the companies that we work is most of the founders have a drive to the office like 30 minutes going and 30 minutes back.

And we ask them to whisper flow their random thoughts every week, every day on it. And from them, like using our own built bot for ChatGPT where we enter what's their voice, like what to say, what not to say, what's the company context. It just builds like tweets and blog posts for us, which then a human, of course, will adjust and make them better.

But it has the POV of the founder. And what was fascinating about it is that as we started to do this daily, because it was like 30 minutes every day from the founder, the AI started to find patterns. And they told us like, hey, this founder is always talking about what he's reading about at least three times a week.

Let's do a thread every week on what the founder is reading or something like that. And I love the idea that it also helped find patterns, which then we could use for something else just from the rumbles of the CEO. But I think using that time with WhisperFlow is one of my favorite things that you can use AI for.

Hank: Yeah, I like that a lot. That's just another good use case, especially like any flows that you set up, making sure to go back and say, okay, like what are the themes and like doing the meta analysis on the stuff you've set up.

Gonto: Exactly.

Hank: Feels important.

Gonto: Since we're talking about AI, I'll switch up topics. But I actually think that Netlify is doing AI wrong. Netlify just shipped two websites.

They have netlify.com, which is for human, and then netlify.ai, which is their new website for AI. In reality, your main website should be both for human and for AI. And I think eventually it will be an agent first.

But for now, it should be for both. And I think if you launch a separate website, it's because you're too scared of changing. And if you cannot change in the AI time, like now, you're fucked.

Hank: I agree wholeheartedly. And it's confusing because, I mean, one, you're just sacrificing your SEO and your AEO, GEO, whatever you want to call it. What do you call it?

Are you GEO or AEO? I'm GEO, but I don't know why, to be honest. It seems like you feel strongly about this for no reason.

It's because you like the letter G because that's how you name it. Yeah, it's Gonto, baby. Gonto Engine Optimization.

So anyways, you're sacrificing that. Also, the website, it's got the human agent toggle, which is funny. And then, shout out to the team that you work with at Clerc.

They seem to be doing this right. They've got this build with agents dropdown. I haven't looked to, like they've got like, and that has like a skills thing.

I don't know. What else are they doing? Or is that the main thing?

Gonto: Yeah. No, so we're actually changing that a bit. But first of all is, we're actually trying now, and it worked really well of this CTA.

Instead of signing up to the dashboard, it's like build with agents. And for now, it's copy a prompt or skill. But we'll actually change it so that you can copy a prompt always, where in some cases it's for skills, but in other cases, it's for a CLI, depending on what you want to use.

But the next test, I think, is making the first CTA, the build with agents, and then the secondary signups. But other things we're doing, for example, all of our docs pages now support both a markdown format as well as regular human format. And what's interesting is that the human format has like step one, step two, step three, step four.

But if you go to the AI format, the markdown format, it doesn't only have the step, but it also has rules based on learnings on what the AI does well on implementing and what it doesn't. So you'll see something like always do this, never do this, deprecated. So we note what things are deprecated and we keep on improving the AI pages with things so that the AI becomes better.

And all of that is implemented with this idea that some bots like Cloud Code, all of the bots will send the right user agents of who they are. So for humans, it will be Chrome, for an AI, it's going to be a bot. So it uses that.

But the other thing it uses is Cloud Code, for example, and Codex now as well, it will send the accept header saying that they prefer text markdown instead of text HTML. So then in those cases, we just reply to the markdown for them because we know that's what they prefer and we know they are an agent. And then the agent can actually be better at using those docs besides having a sitemap.

But I think we're putting a lot of effort with the clerk specifically working with Drew Singh and Steve on this. And I think every company should have at least a team that is thinking about how you are improving your signup flow, your CLIs, your skills, and your docs for AI on your same site, not on a different one.

Hank: Yeah, gotta get the gains all together, aggregated. Synergy, that's the word. We love that word.

Yeah, I like it. No additional notes.

Gonto: I don't think we have anything else for today's episode. It was a short episode, but all focused on AI because we're in the AI era. So if you want us to talk more about AI, please tell us on Twitter.

If you want us to talk less about AI, please tell us. Either way, just fucking talk to us. Thank you.

Hank: Bye. Good stuff.