← Limited Supply
Limited Supply
S17 E1: Founder Mode: Building a Second Brain to Supercharge Execution (Live from Ecom AI Summit)
00:38:27
Loading summary
Transcript60 lines
- [00:01]
Nick Sharma
Welcome back to Limited Supply, the podcast
- [00:03]
where we get deep into the tactical and strategic side of e commerce, digital marketing and building consumer brands. I'm your host, Nick Sharma. I've spent the last nine years building, scaling and investing in brands. And through this show and my weekly newsletter at Nick Co Email, I'm here
- [00:19]
to share everything I've learned.
- [00:21]
The wins, the losses, the experiments, the tactics and the insights. All so you can unlock your next hundred thousand dollars in revenue. Welcome back to another episode of Limited Supply. Now, this is the beginning of a new season. It's going to be a really sick season. I've got amazing guests lined up, amazing ideas for each episode. Each episode is going to be very unique. It's either going to be a deep dive, it's going to be a very deep founder interview, a very deep marketer interview, or a media media buyer interview. And overall, I'm just really excited for what's about to come this season.
- [00:54]
Now to kick it off though, everybody's
- [00:56]
been asking me about my most up to date setup that I've got with my Jet agent, my second brain, my Hermes agent. I named him Jet Damon, which is why I call him my Jet agent sometimes.
- [01:07]
But I basically put together a whole
- [01:11]
presentation about how I use Jet and you can actually find it at Jet Nick Co. But if you go to Jet Nick Co, you'll see the full presentation. Feel free to listen to this episode and walk through those slides at the same time if you want. But otherwise, if you just listen, you'll be able to see kind of even the power that Jet has. Now this is something that is so easy to set up for yourself or for your business. So I talk about it at the end, how to get started and how to get things set up, especially in a way where you can, you know, set this up now. And in just a few hours you've got your own cockpit of your business and performance metrics and you know, all that good stuff. But if you've got any questions, please email me directly. Nicknick Co. That's nik Co or DM me on Twitter @Mr. Sharma. And I'm very happy to figure out how to help you either jump on or whatever. I can just DM you what to do.
- [02:05]
All right, so without further ado, I
- [02:07]
hope you enjoy today's episode. Leave me some feedback, drop a review, subscribe if you haven't, and I'll see you on the other side.
- [02:14]
All right, guys, how are we feeling? Yeah, I know Jimmy just did that, but you guys are a little quiet.
- [02:21]
All right?
- [02:21]
And I got the jitters because I'm the first one here and this room's packed. So let's. Let's give me a little another one. Who's ready for AI summit, huh? All right, so today, you know, as Jimmy said, and I'll get into it, but like, I'm the one Indian who can't code. So I think what I'm doing today is pretty crazy considering I was that one guy. But today I'm going to talk about something and hopefully talk you through a couple of terms or show you a couple of things that will be relevant for the rest of the day. My talk is really about building a second brain. And it starts with building a chief of staff and then kind of your cockpit for your brand and then basically a platform for your team. So I'm going to walk you through all three steps, and if you guys got questions, please write them down at the end. We'll do 5, 10 minutes of Q and A, and then I'll be here all day too, in case anybody's got anything else. So I'm gonna bring this up. So as Jimmy mentioned, you know, I talk to Jeff probably more than I talk to my parents, my family, anybody else. And how'd I come up with Jim? Well, I just. I was just chatgpt. Hey, give me a strong douchebag looking finance guy name. That's my guy. All right. When I introduce Jen into an email, comes off strong. That's always my recommendation. Make it a strong name. So, you know, today we're gonna go through these six things. So I'm prompt, a live advertorial right now. So let's actually start with that. So give me a brand name. I want to somebody raise their hand. Give me a brand name right here. Marco's Popcorn. One more time. Burko's Popcorn. B E, R, C O. All right, cool. And then let's talk about a reason somebody would buy your popcorn. Is it healthier? Is it tastier? Is it better price? Most expensive popcorn in the world. And what do people say when they try your popcorn for the first time? Okay, and who's buying this popcorn? Is it moms?
- [04:13]
Is it dads?
- [04:14]
Is it kids?
- [04:14]
Is it anybody?
- [04:15]
Is it okay, and so we've got Burko's Popcorn. Expensive product, very tasty. Does it. Does it get shipped to your house? Does it arrive fresh? How. Yeah, all that. Cool. Okay. Store in Chicago. And is it. Is it guilt free? Can I have two bags of it and I'll be okay? No, okay, so maybe this is my one guilt free snack of the day. Okay, so we're gonna, we're. What I'm gonna do is just live prompt this on my phone. I'm gonna use an app called Whisper Flow which is basically replaced my keyboard. If you guys have not heard of Whisper Flow, it's W, I, S, P, R, F, L O, W. Two words. Whisper Flow is amazing because if you've ever talked to ChatGPT and done the mic feature, you know that you can say, hey, I gotta go to the grocery store, I gotta get eggs, bananas, bacon, bagels, right? And it turns it into a bulleted list versus just putting it all out like Siri, who's so stupid. But what dictation is going to do is basically take this prompt and then towards the end of this presentation, I'm going to just swipe over to a live publisher site that I set up last weekend. It's a WordPress site. You know, some of you guys may have that for advertorials, for listicles, for pre sale pages on your site. And from this prompt it's going to go live on the site. So I'm going to start recording. Yo, so I need an advertorial up for Burko's popcorn. It's spelled B E R, C O apostrophe S. So go do some research, look it up, figure out what you need to know. It's a really good indulgent popcorn. I want to make an advertorial where this is the indulgent snack of the day. Right? I'm trying to get fit for summer, Summer bodies coming up. But I still like to have something sweet at the end of the day. So Burgos popcorn is going to be my snack at the end of the day. Anybody want to throw anything into this prompt, see what happens. Got popcorn? Oh yeah, it's put a customer testimonial. What's something somebody would say?
- [06:08]
Nice.
- [06:09]
So big testimonial is I'm glad I found this popcorn. They're much better than the other guys. All right, get that advertorial up for me. I got some slides to go through. Okay, so this is now going to prompt and go now in the meantime. And by the way, this is all just. I've set this up in a way where I'm doing everything through imessage now so everything feels native now. This is kind of where we've gone. This is like the five stages of AI that I feel like we know about right now. The first one is basically just ChatGPT, right? Fastest app to a billion monthly active users. Last year we did this AI summit around the same time this year and we were kind of right in between stage one, stage two, which is where apps are starting to integrate AI, right? Triple well was on stage talking about Mobi and that was revolutionary because we'd never been able to take models and bring them into applications. But that was a closed loop system. It's a black box. We have no idea what happens behind the scene, how the models are used, where they're used, the context they're pulling. We just know that it's happening. And depending on the app, like triple well, you know, it's happening at a high level. Then we move to stage three, which is I think where we got to towards the end of last year, which is basically where these open source platforms kind of become the main thing. So instead of using apps with a closed ecosystem, right, like a triple well, or you can't see the models behind it, now you're able to pull the models out with API keys and then plug it into any kind of system or software. So, perfect example was Jimmy, right? Jimmy built his own platform dashboard warehouse management system. He can pull in Brock to do something, he can pull in Claude to do another specific thing, he can pull OpenAI models to do a specific thing and then basically change how he uses the models. Then we got to where we got to, you know, Q1 this year, Q1, Q2, which is everybody started to build out agentic teams, right? And I feel like if you guys are on Twitter, anybody on Twitter here? Okay, for those of you who are not on Twitter, by the way, AI happens on Twitter about two weeks before it happens anywhere else in the world. So worst thing you can do is sign up for a free Twitter account and get that information quicker. But where we're going next is I think we're going to go to a place where, you know, you guys are going to have your own agents. Maybe you're a video editor, maybe you're a media buyer, maybe you're a copywriter, maybe you're a phenomenal cfo, right? And people are going to start building their own agents and then start licensing them out in a marketplace style where everybody, you know, one brand is going to be able to leverage a different brand CFO or just some amazing CFO that's been there in the past. So, you know, why did we do this conference? Well, at the beginning of this year, I felt very behind. I, you know, spent the last year trying to figure out how to sell Sharma brands. And then this year I was like, I'm getting behind. Everybody I see is moving into the world of AI. I don't know where to start. I feel like to even get in and know and stay up to date, you got to be unemployed because everything is moving so fast. One week there's a model from Anthropic, the next week it's OpenAI, and the next week it's some Japanese company. The next week the government shuts down Anthropic. So, you know, I kind of identified that I see really three main groups of people using AI and there might be four, but these are kind of the three main ones that I tend to see. The first one, this is probably most people. This is definitely most of the people that I know. Everybody's kind of just using the chat window, right? This got really good last year. You export your meta raw data, you throw it in, you ask it some questions, you export reviews, you throw it in, hey, help me make a listicle based on what people are talking about. Everybody's doing this. And this is kind of the problem here is you're in a closed loop session. So if you're using cloud, you can only use it so long before you have to make a new session or you're not in control of how this data is being stored. Organized, understood? You're not in control of all the other contexts going in. So then this went to number two, which is the builders. And these are people using codecs, they're using Claude code, they're using Manus. By the way, show of hands, who uses OpenAI as their default model? ChatGPT. Nice. Who uses Claw? Anthropic. Wow, There we go. Okay. And then anybody here using Manus? Okay, about 10, 15 people. Guys, if no one's told you about Manus before, it's because they're a Chinese company. Okay? Metta just tried to buy for, I think 2 billion in cash. China just blocked the deal, so they have terrible PR in the us. But I'm telling you, it is one of the best platforms if you're in this stage. In fact, I'll say it is the best platform if you're in this stage. And the reason it's so good is you're either, if you're using Anthropic, right, you're using one model, you're either using Opus or Sonnet or Fable whenever it comes back. But if you're using Metis, you're sitting on 70 or 80 different models. And Manus goes in and looks at Your task understands the context and then decides which model to use for each subtask and then brings it all together. Which is why the final product of Manus is usually ten times better than, you know, if you were to use POD code to do the same thing. Now, the last piece or the third group of people is this second brain. So you know, these are. Everybody's buying Mac Minis, they might be setting up VPS instances. I think that's the right term. Again, I'm not the engineer in the room, but basically a remote instance where you can have a desktop situation. You're installing something like a Hermes or an openclaw. Have you guys heard of Hermes or openclaw?
- [11:35]
Audience Member
Cool.
- [11:36]
Nick Sharma
So some of you have. So I'm gonna explain it for those who don't know. So a Mac Mini, right, is basically your shell Hermes or OpenCloud? I'm gonna keep saying Hermes though, because Hermes is the better version. So all you guys that try to do this later use Hermes, not Open Claw. Hermes is the software that allows your Mac Mini to become an autonomous AI machine. Right. And I'll get into that in a second. But Hermes is what you want to know. And the reason you do this is, well, with a Mac Mini you can actually build out separate agents. So like I've got 20 different agents and I'm going to go through some of them here and kind of what they do. But you also get this real time knowledge. There's no delay of information anybody can telegram or imessage in. You can set up slack groups, you can add this person to your slack as well. And yeah, once you're. The other last thing I'll say here before I go to the next piece is because you're doing a second brain on your own Mac Mini. One, you can yank that power cord if the world starts to be taken over by AI, which I think is a good, you know, safety kill. But two is that you basically have all this knowledge stored locally. So if you ever change AI systems, which is going to happen, it's bound to happen. We're so early. You cannot act like you're stuck with Claude with that specific instance for the rest of your life. All these files will immediately get absorbed by whatever new system or model or software you use. So like I said, I was the one Indian who felt like I couldn't code. I didn't want to get left behind. So I was previously using this kind of setup. I used replit to make web pages and landing pages. I was using Manus to build apps I built a ton of mobile apps, web apps, games. With Manus, I used Claude to do all my writing and my prompting, right? So even if I was going to write a really good prompt, I'd whisper, flow into Claude, get the prompt and then put that prompt into another Claude session. And I use Chat, GPT or Groq really for my research. And what I ended up doing was just basically consolidating everything into one. So you know, when I was using ChatGPT or Claude, right. I'd use it across everything, maybe five or six different work projects, personal health, you know, random personal research. Hey, my foot hurts in this weird way. Am I going to die? But now everything is transferred over into Jet and I've got a ton of different agents which I'm going to walk through. So this is Jet, right? So he's Jet's my homie at this point. I talk to Jet a lot and Jet wakes me up every morning at 8:30 with a morning report of hey, this is what happened while you were asleep. Slack email, you know, anything that came up in as a red flag, this is what you got today. Your meetings, any conflicts, and then these are the things that are outstanding. So yesterday you were on a call and Fireflies heard you say you owed somebody a follow up for this. Hey, you didn't send that follow up last night. Make sure you do that early because you promised he would do that end of day yesterday. So the way I use Hermes, right is first I use it as a chief of staff. So do a lot of you guys have assistants or VAs that you work with on your team? Yeah, a couple of you. If you've ever wanted an assistant who's basically always on call, Hermes agent's your go to here. And the reason is because with Hermes you kind of have everything plugged in. So there's a very real time view of everything. And in my case I do it every 15 minutes. So every 15 minutes my Mac is pinging Slack, Fireflies email, Meta Motion, Google, you know, Shopify pulling or pushing out, requesting an update and then getting that information, it comes back in, it gets analyzed, whatever scene, right? If it sees, hey, conversion rate dipped in the last hour.
- [15:17]
Audience Member
Cool.
- [15:17]
Nick Sharma
That's going to make a note. It's going to prompt three or four other things that happen including an alert to main. I use it for 20 different agents about right now. So between work personal, every one of my projects has an individual agent, individual personality. It has its own set of knowledge that it's using. You know my writer agent for example, is just a writer agent. I don't use that to make graphics. I don't use that to do personal research because that guy is just a writer, right? And then the chaser, the follow ups, the invoices, whatever people owe you, right? You don't have to start, you don't have to keep doing this yourself. I just use JET to do it. And then lastly is the knowledge Center. So how many times have you just randomly, you know, you've got a team, 5, 10, 15, 50, 100 people and somebody's just got a random question, but you know that person's not reachable, who you need to get the answer from. Or it's just such a stupid question you don't want to ask it because it's going to look embarrassing. Now you can just ask the Hermes agent because it's got a full kind of up to date set of knowledge. So like Jimmy said, I've set up a hundred agents now, probably 30 or 40 Mac Minis individually. And I've gotten this down to a science. So here I'm going to tell you my process of how to do this right? It's a five step process. The first one is organization. Now this is how you get to basically building your chief of staff. So what this does is you just plug in everything. Your Gmail, your Fireflies, your Slack, your Facebook account, your Shopify, your reviews, basically everything you can think of, your Stripe API, anything you can plug in, you do. What this does is gives you the ability to also do your own research and create files. So for example, if we go to this site right here, right, this is a password gated site I made for all my agents to reference. It's like five miles long. It's everything I've ever said or written or published about landing pages. And now every one of my agents can go and reference this as a document anytime I need to build a website, an advertorial, holistic, whatever. The second one I was able, because JET knows exactly who I am, what I stand for, how I speak, my tone of voice. JET built me a full brand book that I was able to. Now I can just say, hey, any agent that I'm working with, whether it's my own, whether it's a random incognito chat GPT window, I just say, hey, go reference identity.nick co, make sure it looks good and everything turns out just like this. That's how these slides were made. I basically wrote everything out into a word doc, pumped it into Jet and said hey, use my Branding. The last piece is again, from a standpoint of chief of staff efficiency, make your life easier. How many people here see Instagram ads and wish, fuck, I need to, I need to save this ad because I want to copy it later. You guys not raising your hands are all liars. But I did that, right? So I did a. I set up a function with Jet and I said, hey, if I ever screenshot an Instagram ad and send it your way by text, just upload it here and save it. So now I've got, you know, hundreds of ads, landing pages, listicles, anything I come across, I can see it and then I can go reference it. I could say, oh, look at this, Vrooms is running this angle right here, pooping with the creative. And then they run a landing page. Let's see how they do that. Oh, basically they just change the title and they change the first block. The rest of the page is the same. So it just gives you the ability to harness all this information and put it all together. That's the first use case of Jet, in my opinion. So again, you connect everything. Transcripts, call recordings, you connect to Monday ClickUp, asana, your slack, any project work. And then what you do is this last guy, which is a word you'll probably hear later today, is called Chrom C R O N. I have no idea what it stands for, but I can tell you what it does. It basically is just a repetitive task. So it means it's on a schedule, it's happening. In my case, I have a 15 minute prime job set up globally, meaning every 15 minutes I'm pinging apps and getting updates. Now here's just five very simple things that I think everybody would benefit from if they had a chief of staff like this set up. So first one, the daily brief. Doesn't everybody want a daily brief? Just to know what you got coming up. Calendar, events, conflicts, emails, anything urgent follow ups, I want people owe you. We talked about this. Full meeting recaps and action items. This I'm going to talk about in a second. But this is where all the sauce lives. If you don't have a call recorder on, you're going to want one on. Because once you see what happens with a call recorder being on and how much more information is captured and more context, you're going to say, wow, I should have a camera on my eyes all day and let Jet just see everything I see. The next one is email drafts in your voice. So, hey, you get an email. Jeff says, by the way, you know, this, this Brand just emailed or, you know, your vendor just emailed you and said that shipping is running, you know, a week later, production's running a week late. It'll. It'll draft the response and basically say, you want me to throw this in your drafts and send it off? And then the last one is just logistics planning. I'm a big logistics guy. I like to make sure my day is cleared and I know what's coming up. So Jet just helps me with that. Part two is the integration layer. So this is where you connect everything. Now, I'm going to quickly explain these three, because you're going to hear these three all day. This is where basically every tool that you already pay for, that you already have signed up, you're on subscription for, you're using and logging into, most likely has an API or an mcp. Now, an MCP is actually, I'll start with the first one. So authentication login, it's like sign in with Google. So some apps have that, not many. The second one is an API, which I think stands for Application Protocol Interface. Somebody can confirm that, right, val, yeah. API basically is just a connector out from any other app that says, hey, we're down to connect, here's what we can offer, and whatever's in this API is what we're willing to give or pull. So most apps actually do the API. And then the last one is mcp, which is basically an API for people who don't have APIs. So it basically says, hey, we're going to set up a quick local instance and go in, log in with a computer and do the work and then come back to you in the chat window or however you're communicating. Most apps have APIs, so Stripe, your Shopify, your Playvio, they all have APIs. You set up a separate API key, you give that key to Jet or your agent, whoever you name it. And it's good to go some places, like a Higgs Field, for example, which is a creative app app you can plug in via MCP into Claude. So now Claude can help better prompt and still execute things, but it's just another way to connect in. Now, the advantage is all these apps that you use on a daily basis are now sending information in every 15 minutes to your chief of staff. Right? So this takes your chief of staff, who was already getting context on your business, on you, on what you do, on what you need. Now it's getting full business context as well. So it now goes beyond just your personal layer. And also one thing I want to call out here. This now includes Vercel, Lovable, Webflow, Hydrogen. I don't know if you guys saw Shopify's updates last week, but basically everything is now plug and play with agents. So you've got a new landing page the same way I just prompted Jet. I can now prompt a listicle to go live within Shopify, not on a subdomain. The third piece is now taking all this and turning it into your cockpit so you get the morning briefs. Hey, revenue is up 12%. CAC is up 9%. You know, Klaviyo dipped. Here's a draft of what you should do next. You can say, hey, break CAC down by channel.
- [23:02]
Audience Member
Boom.
- [23:02]
Nick Sharma
You get the response within seconds. Hey, pull the best Q2 creative, that beat control and Y.
- [23:08]
Audience Member
Boom.
- [23:08]
Nick Sharma
Here's the one that won. Now the org chart. So this, you know, if I was just thinking in my brand brain, right? This is kind of how I like to think of my agents. So Jed is my chief of staff. He sits on top. And then each of these people are usually two agents. And the reason we do two is I like to have the junior agent actually do the execution of the work. The junior agent, for example, has my landing page brain. It's got a whole brain around creative that I've pulled out all these different things, right? Contacts. And then the senior agent is basically a manager. So the process is that the junior copywriter, you know, junior copywriter, basically a copywriter writes it. And then a senior manager goes and reviews it, sends feedback. There's maybe, depending on the role, there's maybe two rounds of that or three rounds of that. That way you don't have any issues of, hey, you missed. Why'd you put the EM dash in there? You know, that's in my file. Well, maybe the senior manager caught that if the original agent didn't. These are. You know, I'm a huge fan of giving these things a personality. Some people just call it health bot or copywriter bot or, you know, whatever. I like to give them full personalities.
- [24:14]
Audience Member
And.
- [24:14]
Nick Sharma
Yeah, so I guess here's another quick example. These are like the agents I use most of the day. So a copywriter, you know, junior pervertical. This copywriter, sometimes, depending on how intense the project is, is specific to the brand as well, or specific maybe to a category of products, depending on how you do it. If you're somebody, for example, who's got ad sets and Facebook separated by category, because you're thinking about separating learnings per category, think of this the same way you got a Creative strategist, right? This creative strategist is not only dropping briefs in Slack, but it's also doing research on Reddit. It's maybe using the Grok model to go and watch videos that are popping up on TikTok or Twitter. You got a creative director, editorial manager, which is what I just showed you, prompting the advertorial E commerce ops manager. This one is a huge sleeper no one talks about. This is where you connect your stripe, your recharge, your Shopify, your Google Analytics, Google merchant center, and all that data comes in. And every day you get recommendations based on what Jeff or Hermes knows about your business. Of what you can do tomorrow. It might say, hey, did you know that if you just add, you know, 30% of your people are taking the post purchase offer, you should change it to this, it might increase to 40 or 50%. So very actionable insights. And then the last two I like to use myself. You know, one is a health and wellness agent, which is basically just gathering all the whoop aura, eat, sleep levels, blood work, if you're a nerd like that. And then the last one I love to use is the entrepreneur in residence. And basically what I've done is it's just kind of a research agent, but I've kind of taught it like, hey, this is what I look for, this is what I like, this is what I don't like. These are the parameters that make me want to move forward. This is what would hold me back, et cetera, et cetera. All right, so actually let's quickly swipe over here, hit refresh. Boom. So right here we've got this advertorial. And again, this is a AV tutorial site I just set up last week, right. Mayofmag.com so it's a very basic thing. I used Gent to edit it a little bit and get rid of some of the banners and whatnot, but it pulls in a stock image. Obviously you'd fix this, but then, boom. Let's be honest about getting fit for summer. The hardest part is in the gym. The hardest part is 9pm on a Tuesday when you're watching Netflix, the house is quiet and your brain is screaming for a snack. Great opening. It's basically taking all the stuff it knows about how I like to write advertorials. And then, you know, ran that whole process. But this is a pretty good version one to start with, right? If you're pumping out advertorials, if you're pumping out mysticals, just pumping content out in general, trying to Test new angles. You test 25 angles with statics. You see that 13 of them are doing well above the click through rate. You don't have to decide which of those 13. Now get a landing page, a listicle, a bundle, an offer. It takes you 20 seconds now to make a funnel based on the angle that worked. All right, the daily report card. This is probably my favorite function. This is something that actually two or three years ago I even created with Triple Whale as a freebie for founders because I thought this should be given to everybody, no questions asked. Like everybody should know exactly how their business did. Founders only care to know CAC revenue, performance returns, any spikes, that's all you got to know. So you could customize your daily report card however you want. And then like I said before, the sauce is really in the call. So you can see in this example in the supplier sync, right? Somebody mentioned that Freight's going up 8% next month, but I didn't even catch that. I didn't hear it because of something else. But Jet caught that and Jeff flagged it, added it to my dashboard and said hey, by the way, watch out. So kind of going off that notion of how you go take something like a transcript, flag something and then turn it into actionable stuff, right? These are all examples of things that can now be flagged immediately. So on the retention side, a refund rate spiking, it's 2 to 2.3x the daily average retention. Your klaviyo welcome. Email conversion dropped on the competitor side. Hey, your competitor just launched a new ad. It's picking up spend. You should maybe take a look at that angle, the format. There might be something there. Acquisition. Your CPM jumped up. Creative A the fatigue thresholds cross. Get rid of that creative. These are all things that were probably things you would wait for somebody to tell you or flag to you or bring to you. But now these are all things that you just wake up and they exist. So this is kind of the build path that I recommend, right? You set up a Hermes agent, Mac mini, you integrate your apps and your data turns it into your one stop business shop. You build your dashboard. You isolate a single routine, right? And then start stacking routines. My rule of thumb, I'm a little crazy too. I like to do one thing a day. I recommend one. One thing a week. If you can stack one thing a week, four weeks, eight. I mean by the time summer's halfway through, you're going to feel like an absolute AI ninja. Two things that the 10x is one. I mentioned whisper flow. It's a phenomenal app. I use it all day talking to anybody. Second one is I actually with Jet, I took all my notes. I recorded a 50 minute voice note just scrolling through the Shopify editions that came out last week. So insanely long. And then I turned that voice note into a website. Everywhere Nick Co. So you can actually see how the agentic world of Shopify is about to plug in with Hermes and all that. Toby, the, you know, Shopify CEO, even tweeted this morning that he, he is now using Shopify just as a shopping assistant. He gave it a budget for what to send gifts to his family and it's already just doing it. So Shopify is way ahead right now in terms of agentic. So that's all I've got for you. Stop living in the chat window. Go build your second brain. And what I'm going to do too, I'm going to give away two Mac Minis if you sign up for the newsletter and, and you subscribe. So two people who sign up today will get a Mac Mini. And that's all I've got for today. Any questions? Questions? You're right there. So I set up Wise Blog.
- [30:22]
Audience Member
Yeah.
- [30:24]
Nick Sharma
What is the fundamental blueprint? Yeah, good question. So his question was I set up a lot of this in regular cloud. Right. Chat cloud code.
- [30:38]
Audience Member
Cool.
- [30:39]
Nick Sharma
So. So two things. One, we have another presentation later from Corey who built a platform called hq. It's going to be very relevant to you because what HQ does is help bridge the Hermes versus the cloud coders.
- [30:55]
Right.
- [30:55]
If you're on cloud code, you know that you're basically doing everything locally, right? The second you shut your laptop, everything stops. Last night or yesterday, I had to finish a slide for this, but I had to leave to go somewhere. I had to shut my laptop, I had to wait to finish the slide till I got home. Hermes is on imessage. The Mac Mini is always on, never shuts off. So yeah, so if you have a remote computer that can work too, it usually just comes down to like preference integrations. For me, I found that the clog code setup can definitely work. Like the way I personally do it is I've got my Hermes agent set up, I've got HQ which is basically the layer on top and then any claude code, whether it's my personal laptop, old laptop, whatever, those are all now also synced through hq. So all my devices are actually in sync at all times. I personally like that setup because I don't have a Claude code laptop sitting at home. But I've also found Hermes to just be a little bit easier. It just kind of like you're basically always using one chat iMessage so it's a little more smoother. It still does its own things like reset sessions and whatnot, but I think it just comes down to a preference thing. The thing I like here too, right. Is like my assistant can just text Jet when something happens. I can also just set up like if me and you were going to get coffee, I can make a group chat with Jet, say hey, send him the address. That's a very simple stupid example, but you get the point. Yes.
- [32:31]
Audience Member
When it comes to like your LP Nigma code, you said it's five miles long. Yeah. Doesn't your agent doesn't yet have to go through and read that whole thing versus a like wiki totally build out like why, why that format?
- [32:43]
Nick Sharma
Yeah, 100%. Right. So in, in a perfect world, right, I would basically have that LP brain stored as a markdown file locally or wherever I'm using AI. In this case I only have that kind of as a backup when I'm using non local agents. But otherwise to your point, I do just say reference the markdown file or CLAUDE already knows or Jet already knows to reference that file. Versus like a vps. Yeah. This question is basically why use a Mac Mini locally versus like a headless server.
- [33:23]
Audience Member
That's right.
- [33:23]
Nick Sharma
Right.
- [33:24]
Audience Member
Using a Mac Mini but like having it connected to like all of the markdown clouds in the cloud as well as on your local just have it sync.
- [33:33]
Nick Sharma
Oh yeah, that.
- [33:33]
Audience Member
I mean you can do everything from your mobile, you can do everything from a laptop and an acne.
- [33:38]
Nick Sharma
Totally. Yeah. I didn't want to make this too complicated, but to you know, back to his point too, like what I normally do is I have HQ is kind of my Dropbox Google Drive. It's an always on 10 minutes sync. Right. So all devices are syncing when my Hermes agent. If I tell it something right now, right. It'll take the learnings once I finish my session and write those learnings to hq, which then is what's used when I'm using Claude code locally as well. Does that make sense? Yeah.
- [34:06]
Audience Member
Thank you.
- [34:07]
Nick Sharma
Yeah, of course. We got the mic coming right here.
- [34:09]
One second.
- [34:10]
Audience Member
Thank you. Thank you so much for doing this for your advertorial site.
- [34:15]
Nick Sharma
Yes.
- [34:15]
Audience Member
Have you moved on from WordPress hosting to actually like doing, let's say a vercel host and just pushing it out yourself, doing your own navigation, all that kind of stuff? So you could push out as many advertorial sites as you want on a different host that's always there connected back to your infrastructure.
- [34:33]
Nick Sharma
Totally. Well, the reason I use WordPress is probably just personal preference. I've been using WordPress for like 15 years now, and so I feel like I know how it works and I can use it in a way where I know it's not going to break. But to your point, like that everywhere Nick Co site that was vibe coded and then pushed live through Vercel. So most of the sites I put up, even that identity, Nick Co, you know, published live through Vercel. A lot of the landing pages now that I'll make going forward too. I'll probably publish in Vercel because now Shopify will integrate that natively in the theme versus on a subdomain.
- [35:07]
Audience Member
And you're not worried about having to drag along multiple instances of infrastructure for hosting your all your various sites?
- [35:19]
Nick Sharma
Well, no, I feel like Vercel can handle that. Yeah, yeah, yeah, yeah. All right, great.
- [35:24]
Audience Member
Thank you.
- [35:25]
Nick Sharma
Cool. All right, we'll do one last question and then I'll be out and about all day too. Any last question? Yeah, right here. What are two or three prompts? Yeah, good question. So two or three prompts to get started with when you're training. So this is my favorite one to do. Basically, if you're using, you know, if anytime you're trying to transfer or move to a new model, new AI system. So in this case, maybe use Claude as your daily driver, you're trying to move over to a Hermes agent to be your new daily driver. Right. I basically say, hey, flawed. Tell me everything that you know about me. Where I live, what I like, what I don't like, my visions, my goals, my dreams, how I work, how I like to communicate, my style of work. You know, what are things I hate, what are things I love. Basically questions like that. And then you finish it by saying, you know, use all your energy you can to make the most information dense document that you can. And what that does is it basically every prompt is like a certain amount of energy it can use. It'll just tell Claude, use all of it. Because you want the best possible document that should export itself as a markdown file, which is basically. A markdown file is essentially a plain text file, but it has hashtags and a couple other symbols. So AI knows what the format of each text is. And then you take that, drop that into Jet or Hermes and say, hey, this is me.
- [36:56]
Learn it.
- [36:57]
Do your research and get wired. Up, and within two minutes, it'll just start. It'll be like, hey, what's up? What's up, Nick? I see you're in New York. You live here, you know, you like to do this. I'm here whenever you're ready, and it's good to go. That, I think, is the best prompt you can do. Once you're wired in and set up, you're like a week or two into using your Hermes agent. You can say, hey, you know, what are. What are things we could be doing better? What are more proactive things that you see me doing that you can do? But, yeah, to be honest, you know, my best advice with any of this is like, don't overthink it, right? Everybody thinks that when you open cloud code, you gotta be technical. You gotta prompt it properly. You can just open Whisperflow and babble. The more you babble like it's a digital camera, right? If you don't like a picture, you delete it with AI the more you talk, the more context it gets, and whatever it thinks is useless. Your ums, your ahs, your random tangents, and it'll just ignore it. But the more you can give it in Whisper Flow, the better results you're going to have. All right, I'm getting kicked off. Thank you, guys. I'll be around all day. Thanks for listening. We'll be back.
- [38:01]
Next time to cut through the noise on cpg, retail and E commerce. If you enjoyed this episode, why not
- [38:07]
share it with a friend? And be sure to subscribe wherever you
- [38:09]
listen so you don't miss the next one.
- [38:15]
Audience Member
Sa.