TEA 250: We Stopped All Work Every Friday To Learn AI (Was It Worth It?)

We went all-in on AI as a company, pulling our entire team off work every single Friday for six weeks, and it cost us thousands of dollars and a completely flat revenue month to pull off. That same experiment is now saving us well over six figures a year in automations alone.

In this episode, Josh and Dylan break down exactly what happened during "AI Fridays," what blew their own minds, what it cost, and the four lessons every founder needs before throwing their own team into the deep end with AI.

Inside this episode:

• The four levels of AI users, from "I just chat with it" to running parallel agents (and how to climb fast)
• Why the entire team abandoned ChatGPT for Claude, and what actually makes one AI "built for business"
• The live dashboard Dylan built in Claude to track the exact dollars and hours the team saved every single week
• How non-technical team members built customer support bots, client health scoring tools, and full landing pages with zero coding background
• The AI sales coach that grades every call straight from a Fireflies recording, plus a P&L tool that flags your weakest Klaviyo flows against industry benchmarks
• The personal app Josh built for his wife in under a week (and why "I'm not technical" is no longer an excuse)
• What "usage," "context," "skills," and "connectors" really mean, and the $1,000 overage lesson nobody warns you about
• When to use Claude chat vs. Claude Coworker vs. Claude Code vs. Claude Design, and the one thing you should never let AI do for you

This wasn't cheap and it wasn't comfortable, but the speed the team moves now made every dollar worth it. If you feel behind on AI and have no clue where to even start, this is the episode that shows you how to catch up fast.

Loved this episode? Drop us a rating because we're going for #1 ecommerce podcast in the world and every single rating moves the needle.

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

► Visit Our Website For Training and Resources

► Leave Us An Honest Rating, Email An Image Of Your Rating To team@theecommercealley.com, We'll Send You A $10 Amazon Gift Card As An Appreciation Gift!

► Learn About Our Mentorship Program For Ecom Brands Making Over $10k/month

► Checkout Our Software, Breezeway - Never Second-Guess Your Meta Ads Again

►  Follow Josh on social media:
YouTube | Instagram | Facebook | TikTok | 

Transcript

SPEAKER_01 0:00

We dedicated every Friday for six weeks of our entire team using AI. And we called them AI Fridays. And Maxwell is about to uh abandon us on this podcast.

SPEAKER_00 0:12

He was sitting just fine until we started recording. Until we started talking. Yeah. He's like, I'm not actually interested anymore. Yeah.

SPEAKER_01 0:19

We spent six weeks though. If anyone's listening, Maxwell's sitting on my lap. If you're watching, you see what's going on. We spent six weeks where our entire team was focused on learning and implementing AI inside our business. And what we want to talk about in this episode is what we did. We want to talk about what we kind of found out, and we want to talk about what we're doing next and some things that we learned along the way. If you feel like you are, you're probably in like one of a few buckets with AI. Like number one, you're not using it. I think that's the minority of people. Number two, you're using it, but you're probably just using it for like basic stuff like chatting, right? Where you're just prompting it and going back and forth and having it do a little bit of things. Level three is where you're integrating some degree of automation. You're using skills and you're not rebuilding everything from the ground up every single time. Number four, you're probably running like parallel agents, you're building multiple things at a time, and you're very, you're like a power user of it. And you integrate a lot of different tools based on the capabilities you know of each of those tools. And I would say when we entered, I was probably, I was like kind of like that level, maybe two, where it's like I was prompt, we were all prompting it and using it, but we weren't using it super in-depth, and I knew that there was more that we could do. And to be honest, we were really only on Chat GPT until we abandoned ChatGPT, mostly abandoned. I still use it for some things. But then we moved to Claude as a company. And and if you're listening and you're like, hey, I'm on some of those things, or maybe you're on Gemini. I don't know anyone who really uses Gemini on a daily basis outside of like certain things. I mean, I always wondered what's the difference? Why would I use Claude over Chat GPT or Gemini? What's the what are the benefits of those? What's the benefit of tools like Higgsfield and Nano Banana? And you you you learn of all of these things and you're like, what should I use? Whisperflow, why should I use it, etc.? And I was kind of on the fence of I had heard of all those things, but I never actually used them. I didn't understand how automations worked. I heard of a tool called N8N or NATON, I don't know how you actually say it. I heard of all these things, but I never actually understood them. And man, these six weeks forced us to learn. Yeah. It's crazy.

SPEAKER_00 2:20

Yeah, I I really enjoyed these six weeks that we did because it was I liked the force timed, like the force time to do it, because I think a lot of like AI can look really scary from the outside, and it's like such a wide arena of things that you can do. And I know we're we're gonna talk about this later in the episode, but you can kind of get stuck in your own mind of like, oh my gosh, what do I do? And when you force yourself to spend time actually doing something and learning something, it's kind of incredible what you can actually get done. Like some of the things that I was able to do, and honestly, I think I can say this on behalf of the whole team. Some of the things that we were able to do kind of blew our own minds. Like, oh my gosh, I didn't even know this was possible. Yeah, and that was really cool because it was it was things that initially I don't think the team would have thought they could do. I know I didn't think I could do it, but when you have the force time, now it took a long time like to learn and figure this stuff out, but then it became kind of an empowering tool, like no, you can do it, and that was that was really cool.

SPEAKER_01 3:26

And it was expensive. I mean, to take out the entire team every Friday for six weeks was I mean, that's a lot of that's 20% of time. You multiply that times the salaries of everybody on the company. I mean, like, that was those those days cost us many thousands of dollars. So, like it it was very expensive up front. And and I don't know how you feel, but I feel like our revenue took a hit from doing that. You think so? Yeah, I mean you go you go look for trying to think of the timeline if we did this because we're recording this podcast after we've already stopped doing the AI Fridays, and we'll talk about how we think about it and do it now. But our revenue is almost it was flat for pretty much the month that we decided to start doing it. Yeah. So this is like April, May-ish, I think, somewhere in that time. Yeah. So it's like, I was between March, April, May. I forget those weeks, but April was like a flat, it was a flat month for us. Yeah. And I'm not saying that that caused everything, but I'm like, man, you take, you take your sales guy off for 20% of the time. You know, you take, I mean, there's so I I think that there was like some shrinking as a result of it, but I'm so glad that we did it because now the speed that it feels like we're moving is so, so fast. So why don't we let's set let's kind of set the base here. Number one, we decided to do these. We didn't know how long we would do them either. We kind of started out, we're like, we're just gonna do AI Fridays. We don't know how long this is gonna go. Maybe it goes indefinitely, maybe it goes a month, maybe it goes two or six months. We don't know. We had no agenda other than to say we want our team to learn the thing and become consumed by it to learn it as quickly as possible because we we we saw, we see where it's going and we wanted to really catch up with that. And man, we learned we learned a whole lot. There was so much we went in that we had freaking no clue about anything. Yeah, like not only what tool does what, but even within them, there was terminology. There was stuff that is like, I don't even know what's a skill, what's a connector, what's context, what is like any of that stuff.

SPEAKER_00 5:20

I think that was one of the biggest notes that I had from doing the AI Fridays was it it gave us an opportunity to talk to the team and educate ourselves, but then also educate the team on how AI works and and what all like that you just said the terminologies of like what is a skill, what is context, what is usage? That was a big thing that came up was like okay, it Claude's telling me that I hit my usage limit. What does that even mean? I don't even know. And there's like context limits and stuff, and what tools do what? Like there are some AIs that are better at one thing than the other, what skills are. I mean, we the the team built a lot of like actual skills that they're now able to use day to day.

SPEAKER_01 6:03

Physical skills and claude skills, two different types of skills.

SPEAKER_00 6:06

I'm talking claude skills right now, but we will get to actual like personal skills because I think that those were built as well. And then learning like Claude co-work, how when to use Claude chat versus Claude Cowork versus Claude Code, and what the what each of those can do, and then even looking at additional tools like Higgsfield, Gemini, ChatGPT to figure out what their strengths are so you can use those when you're up against a problem that you have to solve.

SPEAKER_01 6:33

Yeah. So let's give some some context as to like what we did. So pretty much every Friday we decided uh we're gonna pull the team off of anything else you're doing. We're not gonna take any meetings, and all we're gonna do are what we call AI Fridays. Full eight-hour days. Everyone gets in here around 8:30, 9 o'clock, and we just spend the entire day working on AI. At the beginning, there was no there were no rules. It was like, get in, you want to build an app, build an app. You wanna do something for what your job does, super cool. If you want to just explore, if you want to watch videos, you could research. There was no rules to what we could do. And and then we would do like a show and tell kind of at the end of the day, to say, like, hey, what did you build? What did you learn? What did you find out? What like it was just so that's kind of how we ran them. And we would cater lunch every week that we did it. So around noon or whatever, we would just have lunch, we would cater lunch to the office. And so we'd always kind of start together, but real quickly, everybody learned we got to go to our separate spaces to like think a little bit better, I would say. And then as we're going, we start learning about tools like Whisperflow, which is a way that you can just literally talk into your computer and it it makes it three times faster. It's pretty crazy. Um, and so what we did is every Friday, that was it. We would start the day, we'd take no meetings, other than like we did have some coaching sessions, and we always we're not gonna bail on our clients. So we did coaching sessions, but other than but that didn't involve the whole team. That was just like two people. And so we did these every single day with the objective of number one, just learn how this stuff works. And then level two is how do we apply this to our different roles and positions. And in doing so, our hunch was that we would learn really fast how all the tools worked, what did what, like the difference between Claude Code, Claude Cowork, Claude Chat, and then all the other tools, we would learn how that actually all apply. And then what we did was we monitored how many projects we completed, the estimated time savings of those projects. And then, of course, I got super nerdy and I built an I used Claude to build an entire artifact, live artifact that basically would connect a notion and it would track and tell us how many dollars per year based on the average time savings of each of those projects that were completed by the wages of the actual person, how many dollars per year we were saving along with how many hours. And so we had a whole visual dashboard on it, and I mean it was pretty crazy to see what happened. We did this every single Friday. As the Fridays went on, though, my observation was I don't think we have to do these every Friday because they were it was a long, full day, and then it got to the point for me where I'm like, I don't know what else to build. I have automated literally like 15, 20 hours a week off my plate. Like, what else do I do?

SPEAKER_00 9:11

I I mean for me, I I still make doing AI projects all the time. Yeah. Um, I I liked the like dedicated day to do it, but I I think it did come down to the the whole team probably didn't need the the dedicated day past a certain point. Now you're equipped with what you need to know. So when you are trying to solve a problem, you're trying to do something, you now have in the back of your mind not only the the technical skills of like, okay, how does AI actually work? Like we talked about the context and stuff like that, but then also what tools are at my disposal and and what the limitations are of AI to accomplish that project on a day-to-day basis now instead of just on Fridays, essentially.

SPEAKER_01 9:55

Yeah, we I mean we were able to do things like QuickBooks automation. We were able to create like a sales coaching, a sales coach. So like when call a sales call was done, it would take fireflies, it would shoot it, it would coach him, and it would it would drop in the sales feedback slack channel. We built land, we'd redesigned entire landing pages and funnels. We built software that could calculate all kinds of stuff. Like we had the uh I built a profit and loss calculator that would like you'd up, you'd upload your profit and loss, and it would break down all of the areas you need to improve, how to fix them. It would integrate with Clavio and it would say, these are all your weakest flows against the benchmarks and your industry, and here's what you should do to improve them. I mean, we built so we built tools, we built, I mean, databases in Notion. We built customer support bots, we bought, we built like coaching recap tools, client health scoring tools so we could monitor health of like health of clients that we needed to check in on. We I mean, there's so much. Yeah, and what's it was automated.

SPEAKER_00 10:54

And what's interesting about every single thing that that Josh just said, a majority of them were built by different people. Yeah. Uh, I think probably like each of those projects, there are probably like two projects per person on the team in that list. And I think that's what's really cool is that I think some people think, oh, I'm not technical, so I can't use AI, I can't figure this stuff out. AI is able to teach you how to use it. So if you have an idea of like, I think I can do this, I think this is something that's possible, I'd like to see if it is possible, you can ask AI, and that's what we encourage the team to do. It's like, hey, give if if you're a client, you have the energy audit, give AI your energy audit and say, okay, here's all the things that I do. Look at the things that I enjoy doing and the things I don't enjoy doing, how much time it takes, and help me figure out what of what of this I can actually automate. And that's essentially what we had the team do on AI Fridays.

SPEAKER_01 11:50

Yeah, we had we had everyone do what's called an energy audit where you basically say, Here are all the tasks that I do. Do they bring me joy? Do they not? Am I good at it? Am I not? And then just we said, hey, what can you automate for us and what what could you simplify? And it and Chlaude was able to really help us with that, which was pretty crazy.

SPEAKER_00 12:05

I thought something that was uh also kind of interesting about AI Fridays was it wasn't all AI. Like, yes, it's called AI Friday, but I think a lot of the learning and understanding and everything came from uh not just AI, but what AI then allowed you to kind of open up. I wrote a few down, like like we were able to get in, I was able to get into web design. I I've never done like I've never been able to do web design to that level before.

SPEAKER_01 12:33

And if any real fast, if anyone wants to see what the website Dylan built, go to attendacquire.com, which is our in-person event. Um, I think like three-fourths of all the tickets are sold out for it, but that's our event down in Orlando. We needed to revamp and update it. And Dylan built it.

SPEAKER_00 12:49

It's beautiful, it's like interactive, it's like I don't know if I want to know how much that website would have cost. Oh my goodness. 10 years ago. $20,000. $20,000.

SPEAKER_01 12:59

We did web design that long ago.

SPEAKER_00 13:00

Yeah, yeah. 20K. It'd be insane. And I was able to do it in a few days. But it allowed like the AI Fridays allowed me to now do web design. Graphic design. A lot of the people on the team are now able to create their own graphics for things that they're working on with Chat GPT or or with an AI tool. I was able personally to learn GitHub how to I didn't even know what it was. I knew about it, but I didn't know what it was. Or I've downloaded some things from it, but I was able to learn how to actually use it. And then I was even starting to run apps on a server. So the apps like like the sales grader was actually being ran through GitHub on a server, things that I never even knew I knew were possible, but I had no idea how to do it. I was able to to to do that and completely. I was learning like Slack bots and the team was like automating things, sending in the Slack, which was super cool. And then just on a higher level, like deeper automation and connections with Claude. Okay, what can I connect with Claude? How can I use this connection and this connection to actually make something work together? And then automating with like Claude co-work or things like that. It was it was like kind of setting a technical foundation of AI moving forward.

SPEAKER_01 14:12

Yeah, and I think I think we learned some lessons along the way. Number one, if you do AI Fridays with your team, you're gonna have a lot of usage, is what it's called. And you're I mean, I think we spent the first month over a thousand dollars in overages just on usage, and we have like the Claude Pro accounts, whatever it is for everybody and then on top of our existing Claude. Yeah, on top of it Yeah, so it's like so we had over a thousand, which you know what doesn't matter, it was fine because the speed, the the amount of savings we have is over six figures a year now, just in like automations and skills and so forth. Yeah. And so, number one, what I kind of learned is uh understand what usage is, it's you're gonna go over if you're if you go really deep in this, but I think it's still worth it. Where if you want to get really good at something really fast, you're like, I don't know enough. Like if you're listening right now and you feel I'm behind, I feel I feel like there's so much out there that I don't know, and you're not allocating the time to do it, just do it for a month. Take every Friday for four weeks and say, I'm gonna do or pick a day, I'm gonna do nothing but AI. I have no clue what I'm gonna build. I'm just gonna start the day and be like, here's what I do, what should I work on? How can I learn? What's the next thing that makes sense for my business? And if you don't know, you could just ask, which is the second lesson I would say I learned. It's like, if you just don't know, ask. I built a so from Mother's Day leading up to it, I really wanted to build something special for Kelsey because our son William is 18 months old, and she has like this note running that she built, she keeps track of like words he says and foods he eats and things like that. And we have his whole, his whole little uh bedroom is all like national park themed, and it says like coffee park or something. It's like the national park sign with coffee on it. And and so we love the outdoors, we travel a lot, we and we have this this mission to go to every there's 63 national parks. We want to go to all 63 by the time he leaves the house, which means we we gotta get moving. There's a lot we have to go to. We've hit three or four already.

SPEAKER_00 16:01

How many is that a year?

SPEAKER_01 16:02

Oh, it's like three. Three a year? Yeah, three three to four per year. That's reasonable. Depending on when he leaves the house. That's true. That's true. If he's 30 years old, and you know, we got a long time.

SPEAKER_00 16:10

You got a lot of time.

SPEAKER_01 16:11

That's not gonna be the case. So I wanted to build her something really special and meaningful that was really cool. And I had this idea. I was like, what if I built her an app and I could call it Williams Passport, and she would be able to like go in and have like it have this little AI image of him on the front with his player card stats of like how many foods he's tried, how many words he's said, how many places he's visited, how many memories he's captured, where we have like images uploaded. So when we go on trips, we can log the trips with all the images and videos, and then it has a whole time and then there's a timeline view, and then it it tracks like how much how many lurs he words every month, and he has like these this cool dashboard, and and and also it's a cool place that we can put the foods that he is allowed to eat and he's not allowed to eat. Does he what category of food is it? Is it a grain? Is it a vegetable? Is it a you know, is it is it he's allowed to eat this or he's not allowed to eat this? Because, you know, we're pretty picky. He we eat a lot of Whole Foods and we want him to eat Whole Foods, and so this app can can this app be used by you know grandma and grandpa and other people to watch William. And so I built William's passport and it is Williamspassport.com. Uh so you can all go see it. It's locked, you're not able to get in. But like you'll get to see the at least the screen. You bought the domain for it? Oh, I bought the domain, Williamspassport.com. Go check it out. And so it is a mobile app, or it's a it's a web app because I'm not I'm not trying to get in the app store. But you go to the web app and the desktop version doesn't I built it for mobile, so it's not really desktop friendly quite yet. But like I built I built this thing in a week and it's the coolest thing ever. And there's like a there's a national parks app. You hit parks on the bottom, and it has like a little plus button in the center. You hit plus, and you could add a memory, you can add a food, you can add a vocabulary, a new word, you can add like all these things in there. And he has tabs at the bottom. I mean, it's just really cool thing. So it was her mother, what it was one of her mother's day gifts. Yeah, and it's just really cool because now we can log and we can document his journey. And I built this app in less than a week. And it was and I had no idea a thing about a thing. And I was able to build this out and make it go live, and so Kelsey loves it, of course. And so we I can go back in there and I could literally look at oh, when we were just in Zion National Park, Utah, there's there's a trip logged, and I can see all the photos and videos, and it like live plays the videos behind it with no music as you're scrolling through it. Like it's just and you can click in and you can re-download things, and it has a timeline of day one of the trip. Here's all the photos and stuff uploaded and videos. It's it's really neat. Yeah, that's cool.

SPEAKER_00 18:35

I I think I've I played around a little bit too with like just playing with different things. Mine always ended up turning back around to work. I don't know why. I I couldn't find any personal projects to to work on, but I was working on like ads manager where this was like really it's not great yet. It's there's so many moving parts that it has to like, I need to spend a lot more time on it. But it looks at your ad account for the last seven days. Then it looks at it uses our breadcrumbing framework to analyze your ads based on amount of spend and like purchase volume. And then I took the one once it pulls the top four ads from purchase volume and like CPA, it sends those ads to Gemini, downloads them, sends them to Gemini. Gemini runs through a whole like analyzation of those ads, then it figures out why those ads were successful, and then it writes scripts for you for your next ads. It's not great, but it was a long like that was like almost a passion thing for me because I was like, man, if I could figure out how to do this, it would be insane. I haven't yet because it's really hard. But it was just a fun project to work on.

SPEAKER_01 19:47

Yeah, well, and then every Monday for well it on Mondays it would like spit out here are the five scripts of what you're doing. Yeah, I don't know if you've read the scripts. They're mediocre. They're mediocre. Well, I read I I skimmed them. I didn't I don't recall the.

SPEAKER_00 19:59

Yeah, I don't I wouldn't, I wouldn't actually run them as ads. I need to, I need to spend some time like figuring out where it's getting lost. But just the fact that I was able to get all those pieces working together, it sends a Slack notification, it writes it in Notion, it uses the Gemini API and the Claude API and the meta ads API, and it's like a whole thing. It was just cool to get all of those things working together at one time.

SPEAKER_01 20:22

Yeah, yeah. It's I only did one personal project, and that was one of them's passport. Everything else was business for sure.

SPEAKER_00 20:27

And I'll I'll say kind of tying off to the like meta ads thing. Another lesson that we learned through this whole thing was to not let don't let meta do all or not, don't let AI do all of your thinking for you. I actually recorded a whole episode. It's an audio only episode. So if you're on YouTube, you'll have to go to Spotify or Apple Podcasts. It's episode 247, where a client lost $72,000 by because they trusted AI, basically. It's a it's a really it's a cautionary tale as to what could go wrong if you trust AI too much to do your thinking for you. So I would encourage you if you're interested to listen to that.

SPEAKER_01 21:07

Yeah, so that's lesson number three. Lesson number four is uh iterate as you go versus doing everything on a single day. So we obviously batch learned and and and did things every Friday. However, now that we're not doing it, we are just iterating as we go. So we got to the point where like we all understood it. We understand how it skills, connectors, how automations, how different tools work together. We had this good grasp on it, and it just got to the point where we were able to just say, let's go ahead and just do this as we go. As you're going, build the skills that you need. The goal was to get everybody up to speed with what we need to know in order to really operate faster as a company. We got that, and that was pretty much it. Maxwell's in here whining.

SPEAKER_00 21:47

I don't think Maxwell wants to be down here.

SPEAKER_01 21:49

He does not want to be down here. He's like, Bring me back upstairs, guys. He didn't even want me to cut him. He's like, just take me upstairs. Yeah, he's like, unless you have food or you take me upstairs, I don't want you can't pet me. Um and and I think that so those are kind of the four lessons like number one understand what usage is and know you're gonna go over and that's okay it's just a lesson we learned man claw can get expensive if you don't know the different models like there's like high there's like haiku sonnet opus and they all take a different amount uh the second thing is don't don't trust it to do all or the second thing is if you don't know just go ahead and ask it and it'll tell you the third lesson was just don't trust it to do all your thinking the fourth thing is you know just iterate as you go versus doing it on a single day although I do believe there's a lot of value in just binging and doing it for full days because man if we spread that out over like an hour a week or something it would take us the whole year to to try to catch up to speed and by then it's so far behind. So if you feel behind or something I just recommend just go indulge in it like crazy. You could also just play hooky for a whole week in your business and just go knock it all out in a single week. But hey would you do it would you do it again knowing what we know now? I think I would yeah I would do you think we should do it another six weeks or do you think that we're kind of good at this point?

SPEAKER_00 22:58

I don't know if if it's necessarily the extended because like from Friday to Friday there's no benefit in necessarily that I think the forced time is what's really important. That's that's really what I think my takeaway was so I think you could just you could do it monthly you could do a day a month and you could probably get the same value especially because AI moves so fast just at least catch up on the new things and and don't picture it as you know that if you were to do it a day a month if that's what you were going to do. If you do it a day a month you have to get all your AI projects done on that day. Like you don't think of it that way. Think of it as that's my day to go deep figure out what is possible, what I can do, learn the tools, learn the new things that came out so that in a week or two weeks when you're up against a problem or something that you're trying to do or solve or you have an idea, you now know oh, I can use claw design I can use clawed code I can use Gemini I can use these things to execute this idea but you wouldn't have known it was even possible if it wasn't for doing doing an AI day where you're forced to learning and figuring it out.

SPEAKER_01 24:04

Yeah. And I think that this is good for the team. If you're a founder then one of the greatest as most compounding tasks you could do is spend time just thinking and this is a thinking thing to me. It's like researching and thinking of how you can improve things but there's a point where your team is going to have to learn and just throw them in the deep end and it's a lot of fun. It feels like a cost up front but it's already paying for itself because now the speed we're moving and the amount of things we can accomplish has gone up tremendously. But all that said, we hope you found this valuable I mean hey if you're if you're if you feel behind and you're you're like I just want to really take this by the reins and start to understand something I encourage you you should go do it. Commit to a certain cadence in a certain amount of time just go really deep don't be afraid to explore don't be afraid to spend money either like don't just try to do this all in ChatGPT in fact we are Claude power users at this point. Yeah Claude is the best for business they all have their different use cases we're not going to get into that in this but Claude is overarchingly the best that you possibly could have and if you are enjoying this this podcast please uh leave us a rating we're trying to become the number one podcast for e commerce brands on the planet and it would mean a lot to us it doesn't take very long so low cost to you of time and you know big huge gratitude value to us maximum value and gratitude from us so we hey we appreciate you and we'll see you in the next episode