Hank: And so we had a few experiments we did, one of them we called Project Thunderstorm, because these things, I agree, they got to have fun names. 

Gonto: Yes.

Hank: Code Yellow, like you're changing up, all that stuff. You got to make it a little fun because, yeah. Project Thunderstorm was we were trying to make it rain without getting struck by lightning.

Gonto: I love it. 

Hank: Gonto, it's hard times right now. I don't know if you've been looking at your stock portfolio, but... 

Gonto: I try not to because not great news and I freak out.

Hank: Yeah, it's bad. But we have stories, I think some similar stories on times when it's been hard times at companies we've worked at and we've pulled through. We also have some failures. I don't think we'll get into that as much today. We want to talk about the successes more. You alluded to one previously, Code Yellow, but I'll tell one first about when I was at Neo4j and we lost our CMO.

I was like two months out of college and the CMO came…

Gonto: What was your role?

Hank:  I was in this nebulous role. It was like a 50, 60 person startup. I was hired as sales ops, but on my first day I was also handed marketing ops. This is back before rev ops and a combined, unified system was a thing. So I kind of got that opportunity.

And my COO and I also eventually built out... Part of this challenge eventually led to me creating an inbound SDR team and a bunch of stuff. What happened was the COO came to me and said, “Hank, we are not going to increase our lead volume. We're not going to spend more. We're not really going to do top of funnel marketing. And I'm making it your job to double our pipeline without giving that constraint.” And I was like, oh, okay.

Gonto: There was no CMO. Did you have a team to help you with that? Or was you basically just having to... 

Hank: We had a field marketer and we had a marketing generalist who kept the ads running and there's always kind of PR stuff going on. And you know a quarter or two later, we did get a new CMO and started building out the team, but the gauntlet was thrown.

And then to make it worse, one of the AEs took a short leave of absence, like a two or three week vacation. And so then the COO also told me, “Hey, you're in charge of his inbox.” Basically, you're an SDR.

And so this is where my first breakthrough in the company came through. Well, there were a few, like given this and my constraints and my new control over all CRM and marketing automation systems, I thought, okay, we can probably figure this out. First, there's probably leads not getting hit.

So I reassessed all of the lead scoring and workflows. Basically, I threw out all assumptions and dare I say, approach things with like a first principles look. It's like, “okay, who aren't we talking to?” There's got to be good stuff coming in already.

And so I freed up the pipes and made it so everybody felt like they were getting more leads even though they weren't. And then when the salesperson went out of town, I was like, there's no way I'm managing an inbox for an individual sales guy. Like I don't want that job.

I'm not going to be an SDR again. So this was before outreach existed. There weren't sequencing tools, but I hacked Marketo to make it a sequencing tool.

And so I got his email in there set up the right way so that I could send emails from him to everybody who was assigned to him. And I did a two-step sequence, one with resources and then the follow up that was like a run on sentence. So it felt very human.

And when he came back, he had more replies than he'd ever gotten before. Basically, he's like, “Hank, you're doing a better job prospecting my leads than I ever have. Can we keep this going?” And I was like, oh, light bulb, this automation thing seems to be interesting.

So I turned it on for all of North American reps, because the European reps did not, they don't want me messing with their leads. One quarter later, they wanted me messing with their leads because suddenly North America's pipeline doubled. And so they came back to me like, “Okay, Hank, we'll try out this automation thing.”

And that was like a huge unlock. Then we got SDRs to manage all of those inboxes. And so they were even faster and better.

And we really created this specialized system. And we just kept looking for ways to automate in this mid funnel. And so in that first job, I really didn't learn much about top funnel marketing.

I just had to figure out the middle funnel. And it was an interesting challenge. It wasn't related to the markets or anything.

It was just related to being at a weird niche database that couldn't get too much funding and couldn't attract the best, couldn't retain marketing leadership and had to make do with what it got. It was interesting. 

Gonto: I think the interesting thing about it is that you've got a lot of restrictions or things that you couldn't change, which is true, basically, for any hard time.

And you were creative with those restrictions, because now everybody thinks about, oh, yeah, we'll do automation, we'll talk and send automatic emails based on what the AE or SDR receives. But before it wasn't a thing. So I think thinking outside the box or thinking creatively is something that happens more on hard times because of the restrictions that are set.

Hank: Yeah, you have to, I've told teams before, it's like a haiku. Haiku, it's a very restrictive form of poetry, but it's still just as expressive. But you have this tight box.

And I think to put some fine points on it, like, yeah, I was able to look cross-functionally and gather more resources than the teams before me had because the sales ops team and the marketing ops team before me hadn't been looking at each other's systems. That's more common now. But there's probably ways going into these hard times to think, hey, what can we look, you know,  across the room? What are the channels I'm never in Slack? What are the teams I'm not normally interacting with? What data and systems do they have that could probably be used to my advantage? So I don't know.

Those are some big lessons from there. I'm really curious, unless there's anything else you wanted to say there. I'm really curious to hear more about the Code Yellow at Auth0, because you've alluded to this before. You talked a little bit about it. But this seems like the ultimate hard time story. 

Gonto: It was crazy. Like, we were at 40 million ARR beginning of the year. We needed to double. So we need to go to like 80.

And it's been, I remember exactly, it was like four or five months in on the year. And we were only like five, seven millions. 

Hank: And had you doubled the year before? 

Gonto: Yes.

Hank: Like had gone from 20 to 40? 

Gonto: Yeah.

Hank: So it's not only that you weren't like doing the double pace, it's that something like stopped everything. 

Gonto: Exactly. Like we, it was doing so much worse.

Looking at the data, what we saw was we weren't growing as much in contact sales. We weren't growing as much in gated content. We weren't growing as much in signups and our conversion rate had lowered.

So one of our employees, Jared Waxman, came up with a crazy idea that at first I thought it was undoable, but then we decided to go for it because we didn't know what else to do. So it took us some time to accept the idea. And his idea was we're going to do something that's called Code Yellow.

It wasn't red because we still had some time to do something, but it was Code Yellow because like we need to consciously think and do something about this. 

The idea was at this stage, the team was pretty big, maybe 40 people. And Code Yellow's idea was, okay, we're going to get half of the marketing team. We only focus on four projects that I'll share in a second, which focus is to drive more signups, more gated content, and more talk-to-sales. And the only exit criteria for those four teams' full-time dedication was that we hit a specific percentage growth on talk-to-sales, on signups, and on gated content. So there was no exit time just based off of dates.

The thing was that it requires two things besides just setting up those teams. Number one, the other half of marketing was doing what they were doing before. But if they got an ask from Code Yellow, they had to stop what they were doing and actually help them out.

And then I went to the executive team to pitch them that we needed the same from other teams. Meaning that if we needed some help from CS or from product or from sales, they will have to stop what they are doing to do Code Yellow. Before going to that meeting, I remember I chatted with Dave Willner, who was our CRO. He was on board. So we both pushed for this idea. And without getting the executive buy-in, it would have been impossible.

We got the executive buy-in mostly because the founders were freaking out that we're not going to hit the milestones. So it was an easier sell, to be honest. I don't think it was my ability to sell that got them to say, like sure, we'll do this.

Hank: Okay. I want to highlight something here that's interesting. A hard time is the best time for innovation because people are more open to the crazy ideas.

Gonto: Exactly.

Hank: And we see this with companies that we advise. I posted about this maybe a couple months ago on our little Code to Market blog, which I write on sometimes, about this PLG is poison. And all the time founders come to us and they're like, “I'm ready to scale up my company.”

And we're like, “wait, you don't even have half a million in sales.” A lot of times they don't. They'll come to us and they don't have any sales.

And I'm like, you need to go do the hard thing. And if they have fresh funding, they're like, I'm not going to DM a thousand people on LinkedIn to sell the thing. And then they'll come back six months or 12 months later, and then they will do it.

And then they will like hit the Code Yellow and they'll go all in on the crazy unscaled thing. 

Gonto: Exactly. Like when you're freaking out, it's the best chance to try something different.

In this case, we've got four projects that basically this guy who proposed it, Jared Waxman, was the leader of Code Yellow. So he was the leader of the four projects. They reported to the marketing team and actually to the entire company weekly on what was worked on, what did they ship, did they get any results? The idea of that was to give accountability to the rest of the company on it.

We're going to do this. We need the help from everybody. So look at what's going on. So having that visibility, I think was key. 

The four projects we had were dress it up, go wide, go deep and change it up. We tried to make it a bit like funny or like easy to remember the names.

Dress it up was mostly UX and UI changes. The idea was we're going to try a shitload of experiments on the web, on different places, just change color, change, put on names, change, put on size, et cetera, to drive some increases. That project did drive some increases, but it was small changes.

However, those small changes allowed us to have time for the other three projects to work better. Second project was go wide. Go wide was increasing traffic.

So a lot of that was based on doing more content, but more specifically doing things we were not doing before. Doing some specific use cases and case studies that were technical. Started to do more videos that we've never done before. Started to use specific campaigns for review sites or specific campaigns for specific case studies. And in general, how do we increase the traffic to the website and for people to come? 

Go deep was more options to engage. So that meant, for example, the biggest thing in go deep was we added drift. So we added a chatbot, both to the dashboard, as well as into the web, so that people could engage us in other ways. 

We also added a concierge program, which was very similar to your product advocates that were technical people that were helping people out on the trial. And then we also started to do weekly live demos and Q&A and specific things as well related to starting to automate based on what people did on their website.

Last project was called “change it up.” Change it up was big bets. We knew each big bet would take half a quarter maybe to implement, in some cases a quarter. And we were going to do crazy shit basically on that one. Some of the things that we did, for example, was what if we remove the entire signup flow? What if we try to remove the onboarding? What if we do a really new onboarding? What if we do single sign on our gated content forms? So we avoid double entry and we know when people are entering it. What if we do pricing calculator? And we had multiple projects there.

Again, the team started to report weekly on the changes and it took us actually six months to hit the results. But because the dress it up and some of the go wide allowed us to actually get better data, get better numbers, get better things. We basically started to see slow increases, slow increases, slow increases, and then big hikes.

Once we saw some of our big bets work, the chatbots did great. Not actually the concierge program that did OK, but the bot on the pricing page or on solution pages and connecting with people with humans there did fantastic.

 And then some of the big bets like adding single sign on to our form, so it was easy to basically change. And completely changing onboarding for something else made a huge difference. The other one that was a big bet that worked was immediate response to form fields where you would enter your phone and email and you will get in under five minutes somebody reach out, which took a lot of operations work. But in the end, we ended up hitting the 80 million ARR by the end of the year, even though we thought we were not going to.

And when we did that, we stopped doing a lot of things. We didn't do any sales collateral. We didn't do any battle cases. We didn't do specific white papers that the sales team and others were asking. We didn't do a lot of like customer expansion programs. And in the end, I conceded to the pressure and we dismantled the Code Yellow teams.

I always regretted doing that. And I asked myself, like, what would have happened if we actually continued with Code Yellow and stopped doing those things that every company does? But if they didn't really help with revenue, should we do it because everybody else does it or because people are asking us internally or should we not do it? 

Hank: What's the internal pressure for the stuff that you had given up, like the sales collateral, the battle cards, the stuff that's very basic and common? 

Gonto: What was the pressure? 

Hank: Yeah. Where was what was the pressure? What was it coming from? What does it look like? It was coming from the directors of sales saying that it was really hard for them to talk about the new features that the competitors were adding, or it was really hard to them to close some of the deals. And then the CRO and CS was pushing as well. But we basically had pushing all over from everywhere. 

Hank: Yeah. Interesting. That's like the number one thing from directors of sales is asking for materials. And we've talked about that before.

Gonto: Exactly. And it was just one thing. But in general, my question was like, we stopped doing things that every other company does. If we hit the revenue, maybe they weren't that needed. And that's we had this whole episode on sales collateral. My contrarian views on sales collateral and how useful it is actually comes from this experience where, yeah, I don't know, maybe they weren't that useful.

Hank: Yeah. My view comes from having directors of sales join and say, “hey, where's all this? We need this.” And I'm like, well, we got here without it.

You know, did we need it? If like we just hired you because we're growing so fast. And anyways, I'm curious about let me tell a quick, interesting failed story. And then I'm curious of your big bets. What happened? 

So at Vercel in the summer of 2022, that was when interest rates started going up. There was kind of the slowdown on VC. Like everybody was starting to get scared. And so we had a few experiments we did.

One of them we called Project Thunderstorm because these things, I agree, they got to have fun names. Code yellow, like all that stuff. You got to make it a little fun because, yeah. Project Thunderstorm was we were trying to make it rain without getting struck by lightning.

I love it. So that project specifically was we had just done our first quarter, our first full quarter of ads on a couple of platforms. So we had our first CPLs and cost per opportunities and cost per like we didn't have much data on cost per wins, but we had just enough that the COO and VP of sales, like when I shared the data, they're like, wait, Hank, you can spend a dollar and get this much out of it.

And I was like, well, yes, but that's not exactly how ads work. Like they're tough to scale SEM because only so many people are typing in the keywords. And they're like, “what if we go nuts with this?” And I was like, hey, I'm I'm down to try.

So we did go nuts with it. Basically, we came up with a program of, hey, we're going to try and spend not double, not five X, but like 20 X that we did in one quarter in the next quarter. And we set it up so that every week we checked in, we had all the stats, we laid them all out in the sheet.

We weren't relying on some BI or data analyst, whatever. We made sure we could pull all the attribution data. And we saw pretty quickly, OK, this is actually really hard to scale up.

We had two people where their full time job for a couple of weeks was basically typing in keywords and negative keywords. It didn't work that great, but it did get us some interesting discipline on ads. It did accelerate the ads program, which now is working.

But it didn't have the results, you know, that everybody had hoped for. And I got in writing before like, hey, normally this is when you fire performance marketing teams. So can I just make sure that if we do this experiment, you guys won't fire this team because it's not their bad.

So that was interesting. 

Gonto: Not every experiment works. And I think Thunderstorm project, the idea was doing wild and crazy things.

 And some things will not work in the beginning. Like for Code Yellow, it took us six months to get the projects that actually made a difference working. So you're going to try shit. And the weirdest, craziest, most different things that you try, just the more likely it is that you're going to eventually succeed. 

Hank: So how do you determine a winner and a loser in that? If you have to wait six months for us, you know, we gave it like a month and then we could tell. Because it's ads. So it is very performance based.

You can tell relatively quickly if you're generating. For us, we could see if we're generating self-serve deals. That was the first indicator and then opportunities.

And then we couldn't see closed won enterprise deals in that time period. But we had an indication whether we're creating opportunities that could close. So that one we could tell kind of quick.

But for these ones that you're saying where it takes six months to see the results, how do you have the patience? 

Gonto: It didn't take six months for all. It was more about like we try one thing. We actually tried four things at the same time.

We waited two weeks. None of them worked or one worked, but slightly. OK, let's try another four things for two, three weeks.

But what I meant is it took us six months of trying multiple things to actually hit the objective. And it wasn't just one thing that worked. It was probably, as I said before, like three or four things that were like, holy shit, like these things are working. That basically made the difference. 

Hank: Yeah. Marketing is a compound interest discipline.

Gonto: Yes. 

Hank: As a side note, like things add up and build up over time. You know, no one piece of content is normally going to make the difference. No one campaign usually does it. But I like that. That's interesting. That's insightful. What else you got? Other hard time stuff? 

Gonto: Another hard time story that I have is on Retool. Retool is the internal apps platform.

I joined them because they were doing always fantastic and absolutely great because their PMF is so strong. But they didn't do that much marketing. Like they had a marketing team, but they weren't paying that much attention until they missed a quarter.

And when you miss a quarter, but you're close to 100 million ARR, you're like, holy shit, like what do we do? We need to change things fast. And that's the time that I actually joined them. And I basically joined them to fix their marketing team in three months, ideally.

That was the objective. I stayed working with them for a year, but that was the main thing. So it started with an offsite.

Like I did an offsite in person for five days. The first few days were just with the leaders of marketing. But eventually I met with everybody on the team in person, some were calls, but most in person for at least 45 minutes just to ask them questions.

So what did I do? First was I needed to learn about what they knew. So I asked them a lot of questions and got data on it, both from Salesforce analytics on it. How many viewers do they have on their website? How many signups? How do they think about activation? Who are their customers? How do they find their customers? What was the different user path? What things did they try from marketing from demand gen? What things did they try from inbound? What for outbound and why? After that, I basically sent homework to each member in marketing.

And the questions were more about like start, stop, continue for yourself, for your team. What do you think marketing is getting stuck at and what feedback or what things would you like to try and do based off of that? So I sent them that homework and then met with everybody. Some people came with slides, other came with docs, other came with nothing.

After doing that and those two things, we basically did two things. Thing number one was assessing the people. We ended up firing half the team after that because they weren't good, it wasn't working out.

And then the other thing was setting up some base projects on learning about things that they didn't know. Thing number one that was interesting to me is they were starting to do a lot of demand gen, more like outbound and AVM and webinars. However, 85% of their revenue came from inbound and they had no growth team, were doing nothing on inbound.

So my first recommendation was instead of trying things on something you've never done, try to improve the thing that is already working that you've never worked on and it just works by chance. That was thing number one to focus on. Thing number two was you need to better understand your customers.

Some people were saying that they…because I also interviewed the entire executive team, some people were saying the customer was developers. Others were saying they were a director of engineering. Others were saying it was product.

So there was no alignment on who it was and why. They also didn't know the user path and what was going on. So the other part, the second part was, we're going to do real customer research.

Some based on interviews and some based on actually looking at data. They didn't even have like activation and thinking on data on that side. So doing both data analysis, as well as starting to do some more analysis on the customers and understanding them.

And then the third part was, okay, we're also going to start doing and trying some new experiments, but we're going to have an experiments team and focus on hiring that. 

Interesting things that we learned, for example, was most of the team, including the CEO thought that the developer was the main user. However, when we did the data crunching, what we saw is that the first user, maybe 40% were developers, but 60% were directors of engineering, mostly because Retool is so easy to use.

And we saw that a lot of directors of engineering were entering, trying it out. And once they saw that it worked, they would actually include their right hand, the senior engineer to try it out as well. And if that worked, they paid and then they put a junior engineer.

So the junior engineer was actually the user after it was paid. But for the discovery, it was never the junior developer, even though they thought at the beginning that they would focus on junior developers. So it took some time, like all of the research starting to do some growth experiments, et cetera, probably took two, three months, but it all just started to work after we maybe 12, 13 hours a day, five days in a row to do a deep dive into the team.

I actually think it helped them to have somebody from outside that was just poking on things, asking questions, looking at data and giving them an idea. 

What I have to say about Retool is the CFO at Retool was running marketing at that time. He used to run marketing. It's a weird CFO. He was awesome. His name is Adrian and he was a really great partner who he was open always to trying things, doing things that would challenge and then iterating based off of that.

Hank: That's awesome. Especially from a CFO. So it sounds like the way that this worked, they hit the hard times, they missed the quarter and at a hundred million, if you miss a quarter, suddenly as a founder, you're like, we're not going to be able to IPO.

We're not going to get acquired. All of this was for nothing. You have to start thinking of all your options and it means you either cut the team in half and go for profitability or you figure it out.

And that's true of both of your stories. Like fantastic. So you challenged the assumption.

This one, it sounds like this was much more about alignment and the alignment was probably harder than Auth0. 

Gonto: Yes.

Hank: Because at Auth0, it sounds like you kind of knew who you were going for, but you had to, the alignment was getting around.

Hey, we need this secret passphrase of Code Yellow that means we're going to do whatever we think is going to work. Whereas at Retool is about, we have to create alignment on who are we actually selling to? Who's the user? And what experiments are we going to run and where the focus is? Because it sounds like they fell into a common trap of, Oh, we have, I see this all the time. Like, Oh, we have a bunch of inbound. We have to diversify to outbound. We have to go AVM and all that. When it's like, no, there's still so much to gather here.

Gonto: Exactly. They did nothing there. Like it just worked without them doing anything.

Hank: Yeah. Which of course, like I have a bias toward that because of my first story, right? Like at Neo4j, I was told like, you can't do anything else. You have to like basically double from the stuff that already works.

So I kind of always have a bias for that. If I have a bias, it's not going out, but yeah, I love that. Very interesting. Any other lessons or principles you think that are good to distill?

Gonto: I think from this, what you said exactly, right? The other thing I think is like, I think you have to stop and analyze everything. Like in this case, we went deep into what data did we have? Did we know the customer or not? Did we know the user path? And we just shook everything up with three clear priorities. We came out of the offsite with these are the three things we're going to do in the next six months.

And there's nothing else that we're going to do besides that. So I think coming with focus on these are the three things after doing this, cleaning the team to getting the right people there. So people didn't feel like there were other people who were slackers or not doing their job. 

And actually diving in and accepting on, maybe we are a huge company, but we don't really know our customers. Like they were humble enough to say, maybe it's true. We don't know. I'm going to actually do the research. And they were open to learning things from that. Like that, I think is something that was absolutely great in Retool.

Everybody was open to learning things. Everybody was open to trying new things. Everybody was open to just shake things up.

Hank: Awesome. 

Gonto: With that, the other thing that to me is interesting about hard times is that now we have something that at least in my retool story and in my outsider story didn't exist, which is AI. Now with AI, you can do so much more with less, with less people, with less time, with less resources.

Toby Lutke, the CEO of Shopify, just shipped a letter. It's on Twitter. It's fantastic about how the entire company needs to focus on AI.

And you were telling me that you had a leadership offsite that you were planning to focus on strategy, but because you saw most of your team didn't do AI, you said like, fuck it, we're going to do AI offsites. How did you do it? What was the focus? What were the outcomes? 

Hank: Yeah, it was an interesting decision I made unilaterally that my, it was my go-to-market leadership team offsite. So I had the head of DevRel, I had two marketers, RevOps/Sales guy and head of CS.

And we're all going to come out and prepare for some launches we have coming up. But there are some workflows, some things that are just workflows and some things that are AI workflows that we've been talking about for months and that some of them hadn't done yet. And then there are also a couple of people who just didn't understand the power of not only AI, but just APIs and that, hey, you can connect these systems and automate them.

So I kind of lectured my team. Lecture is a strong word. I kind of, I ranted.

What I did was I ranted about guys, this is really important to me. We've got to figure out how to do this because everybody's asking for headcount and we're trying to grow a team. And I'm like, how can I justify that if we haven't tried automating and tried saving ourselves time and work, you know, before doing this? So I said, we're scrapping the strategy stuff.

We can do that async. I made them all come to Idaho because I've traveled too much. So I said, you're all staying at a hotel two miles from my house.

We did a couple of fun things. We shot a potato gun. I'll share the tweet.

Gonto: It was fun, it was awesome.

Hank: That's a, that's a very Idaho thing. Making a gun that shoots potatoes out of plumbing and hairspray and a lighter. So aside from that, yeah, basically for two and a half days, all we did was we spun up an automation tool.

We had a couple lined up. One of them is a company that's been trying to sell us and we told them you want to be part of this hackathon, trust me. And they, they blew it.

They weren't ready. So we said, great, we're not using that one. Another one's very interesting, but it wasn't malleable enough.

It's basically an AI chatbot at the end of the day. We're like, that one's going to be too hard to use. There's a ton of tools like copy AI and Lindy.

And we went with N8N. Which is…

Gonto: Which one? 

Hank: N8N. You heard of this one?

Gonto: N8N

Hank:  It's a self-hostable Zapier, but you can also get the hosted version. It's kind of an older tool now.

It's super easy to self-serve and it's got little agent tools. So you can plug into any API and you can plug into any model that you have an API to and just go with it. So a lot of the time was spent.

Gonto: Why not Zapier? 

Hank: Zapier is too expensive. Zapier's pricing just does not scale. And also its workflows are too, I have lots of opinions on workflow tools.

I've worked at a workflow tool, trade.io, which is now trade.ai. So I have lots of opinions, but Zapier is too linear. Like the workflows aren't flexible enough. And it's very like point to point.

Gonto: What I like about Zapier is they were the first to implement MCPs, which I think others are going to implement soon as well. But they did that first, which was cool.

Hank: Yeah. So we made an MCP. Because I had two more technical people, right? My head of CS and my head of DevRel are very technical. And they were like, we want to be able to serve docs.

So they basically made a little MCP with PG vector of our docs. And so we were able to use that. We built a connection to that.

Gonto: There's a great service called Layer. Build with Layer. Have you heard?

Hank: No.

Gonto: Where you can create your own MCP without writing code for your API and for your docs. So you just put all of your functions and it's basically creates automatically an MCP for you, which I've tried and it's fantastic.

Hank: Yeah. For those who don't know, MCP is basically a way of getting a bunch of context or docs or something together that your AI can then use and focus on. 

So for us we were like we would like our docs, the most recent versions available for our AI so that it can reference that and search that before searching just the entirety of the internet right? Is that a good way of explaining it? 

Gonto:I think it's it. 

Hank: Yeah that's you know kind of like the new form of rag I guess.

Hank: So then from your offsite you basically try to look for tools for AI automation. You picked this one. What are you going to use it for? Is it automation overall? Is it for content? Like what is it for? 

Hank: Yeah so the goal was learning. The goal wasn't so much about hey you have to whatever… but I told everybody I was like there's not really teaming up here. Everybody needs to work on a workflow, myself included. 

So our head of DevRel he created a workflow that takes our YouTube videos and takes the transcript and then creates a blog post in the style and tone of the person who wrote the video which is great because that's better than the normal AI slop of “hey write an article about Laravel 12.” Instead it's “hey convert this transcript using the same style and tone into an article.” 

One that ended up not using AI but she had to use AI for the troubleshooting of the code, because she's not a coder, was our field marketer. She created an auto discount generator so she can now actually it was very clever so she she used Snowflake the Snowflake connector to look up hey who deployed yesterday and from that list I'll pick winners or something.

We're working out what to do more with this. I told her we can't turn it on because it would be too costly because then she takes that list and she generates a Shopify discount code for our Laravel merch store for however many people she wants and then she emails it to them through HubSpot which after she built it she's like “hey this is great but actually the more useful use case is we're gonna make it so our support team can request a discount code anytime they just want to like thank someone for being patient or something.” So that was cool and she's excited.

There's no AI stuff in that. But one that's very AI heavy is our support guy, he's taking all of the recent tickets and summarizing them into themes so as you would predict from when you broke down our launch a couple episodes ago he immediately saw like, okay as predicted like a certain percent of these tickets are about pricing confusion. And so now it's just supporting the narrative that we need to put more resources into reducing pricing confusion and you know he had other things which are going into like interesting UI stuff. 

Another one that our product marketer did was she gets all of the PRs from GitHub, filters out the internal stuff and then creates a changelog draft because everywhere that she and I have been there's always like the PMs are like, hey product marketing should write the changelog and the product marketer is like, no the PMs or the engineers should write the changelog so we were like what if just AI writes it.

Gonto: Nobody wants to do the changelog.

Hank: Nobody wants to do the changelog but people love changelogs so now she's got it to where every Monday it's just pumping out a draft. 

And there's many other things like one thing I'll tie back to something you said about Auth0 was we have a contact sales form, we don't really have a sales team. So we're almost done with the workflow that's gonna just give people that first reply and like or at least draft the reply and our guy set it up so he can like hit approve in slack if he likes the reply and make sure that it's appropriate. 

So now everybody has this power and it was less about the workflows that came out of the hackathon and much more about we all have a skill now and that can change how we think and approach things.

Gonto: I love the idea of like you focus on the skill and you get people to do it so that they can do it in the future. And I think what I've seen with teams that do something really well with AI is do the thing with AI that the rest do not do. 

What I mean by that for example is most people are using AI to write content. I think AI content sucks. Most of it is not that great or like Google will notice etc. But what if you use AI to research what to write about. What if you use AI to research what people are searching, where people are sending your tickets as you said, or something like that and then you get a human to actually write it but you get the research done which was also time consuming.

So thinking about like how can I use AI, not just for something customer facing, which maybe is not the best, but how can I make all of my other processes more productive so that anything that is customer facing I can still do it but with a lot of help behind.

Hank: Yeah exactly. An example, I didn't talk about the workflow I made. The workflow I made is it does what an SDR should do after someone makes their first deployment on Laravel cloud, it looks up hey has this person used other Laravel products before and what doc pages have they looked at, that you know we have a record from rudder stack of them looking at. And from that, what I'm this is where the MCP that we're working on is going to be most useful, it wasn't very useful in the state we produced in two days, but then I can suggest “hey these are the docs that you, knowing what you've seen and what products you've used before, that you should look at.” And like an actual personalized thing for a new customer feels very, that feels very valuable to be able to do that type of thing.

Gonto: And there are things that I've always wanted to do like a fully personalized onboarding, fully personalized things that they can do. Which before they were hard they didn't work that well and now I think it's so much easier.

Hank: Yeah because the only way before was to have some fresh out of college kid that's cheap enough like do an hour of research per person and that doesn't scale. So that's why we've never been able to ship truly personalized onboarding stuff. But I think we're going to be able to do that at Laravel and I think people like it. They don't care that it's made with AI. Like it's going to be the most useful links for that person and the resources that they didn't want to go find on their own.

Hank: Yeah so we're out of time. But that's it. Welcome to the hard times. Take advantage of it. Later!

Gonto: Cheers!