Limited Supply
Limited Supply

S17 E1: Founder Mode: Building a Second Brain to Supercharge Execution (Live from Ecom AI Summit)

7/8/202638:277,526 words

The Podafi briefing · AI analysis

Founder builds local AI agents to automate daily e-commerce operations

5 min brief · 4 takeaways · Based on this episode’s transcript

The 30-second brief

TL;DR
  • 01Local AI agents on personal hardware offer more data control than closed cloud chat interfaces.
  • 02Connecting agents to business APIs enables automated daily briefs and real-time performance monitoring.
  • 03Pairing junior and senior AI agents helps catch formatting and strategic errors in creative.

The big picture

E-commerce operators face a flood of AI tools. Nick Sharma argues against closed-loop cloud chats, advocating for local autonomous agents connected to business APIs. This creates a real-time chief of staff for daily briefs and creative reviews, though it demands significant technical setup and ongoing maintenance.

Useful for: This helps e-commerce operators looking to automate daily management tasks and centralize business data using local AI agents.

Synthesis of podcast conversations. Speaker claims are not independently verified.

01AI infrastructure

Local AI agents provide deeper business context

Nick Sharma observes most founders use closed-loop chat interfaces for basic research. He argues for running autonomous agents on local hardware to maintain full control over data and avoid cloud model lock-in.

Why it matters. Local instances prevent vendor lock-in and allow brands to retain proprietary data while building customized workflows that evolve with the business.

Your next move · Podafi’s suggestion

Evaluate your current AI usage. If you rely solely on basic chat windows, research local setups to test a single automated workflow.

The catch. Local setups require technical knowledge to maintain. Do not attempt this if your team lacks engineering resources or if data security protocols prohibit local hosting.

Transcript evidence

From the transcript

“because you're doing a second-brain on your own Mach Mini, one, you can yank that power cord. If the world starts to be taken over by AI, which I think is a good safety tail book, too, is that you basically have all this knowledge stored locally.”
Read the source transcript ↗

Evidence summary · paraphrased

Sharma explains that running agents locally allows users to build separate workflows with real-time knowledge, avoiding the closed-loop limitations of standard chat interfaces.

02Operational efficiency

AI agents automate daily operational briefs

Sharma states his AI agent polls Slack, email, and Shopify every 15 minutes. It generates morning reports, drafts follow-up emails in his specific voice, and flags conversion rate dips without human prompting.

Why it matters. Automating routine operational checks and communications frees up founder time for high-leverage strategic work, reducing the cognitive load of daily management.

Your next move · Podafi’s suggestion

Identify repetitive daily tasks, such as checking ad spend or drafting vendor emails. Map the data sources required and test if an AI tool can draft these updates.

The catch. Automated emails and alerts can cause confusion if the AI misinterprets context. Always require human approval before the agent sends external communications or makes account changes.

Transcript evidence

From the transcript

“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, anything that came up as a red flag.”
Read the source transcript ↗

From the transcript

“every 15 minutes, my Mac is paying Slack, Fireflies, email, Meta, Motion, Google, Shopify. Calling or pushing out, requesting an update, and then getting that information, it comes back in.”
Read the source transcript ↗

Evidence summary · paraphrased

Sharma describes his agent waking him up with a morning report of updates, noting it drafts responses to vendors in his exact voice and flags anomalies.

03Creative workflows

Pair junior and senior AI agents for creative

Sharma suggests structuring AI agents like a team, pairing a junior agent for execution with a senior manager agent for review. This multi-step review process helps catch specific errors before finalizing copy.

Why it matters. Reduces the risk of publishing flawed creative or missing brand guidelines, saving time on manual quality assurance and preventing costly public mistakes.

Your next move · Podafi’s suggestion

Assign your primary AI tool to write initial ad copy, then prompt a second instance to review it against your specific brand style guide.

The catch. Running multiple agent reviews increases API costs and latency. Do not use this for low-stakes internal communications where speed matters more than perfection.

Transcript evidence

From the transcript

“the junior coper writer, basically a coper writer writes it. And then a senior manager goes and reviews it.”
Read the source transcript ↗

Evidence summary · paraphrased

The speaker describes using a junior copywriter agent for execution and a senior manager agent to review the work, noting this catches details like specific punctuation preferences.

04Prompt engineering

Export AI context for agent onboarding

To migrate to a new AI model, Sharma prompts his current AI to export everything it knows about his preferences, goals, and working style into a plain text file to establish immediate context.

Why it matters. Reduces the time required to train a new AI assistant, ensuring the new agent immediately understands brand voice and operational preferences without starting from scratch.

Your next move · Podafi’s suggestion

Ask your current AI assistant to summarize your working style, preferences, and goals into a single document, then use that document as the foundational context for new tools.

The catch. The exported context may contain outdated preferences or hallucinated details. Always review the generated document for accuracy before feeding it to a new agent.

Transcript evidence

From the transcript

“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”
Read the source transcript ↗

Evidence summary · paraphrased

The speaker advises asking the current AI to export all known information about the user into a plain text file, which is then given to the new agent to learn.

From listening to doing

Take one idea into the week

Suggested experiments, not proven results. Choose what fits your brand.

Test senior agent copy review

  1. 01Generate a batch of new ad copy variations using a junior AI agent without any review process.
  2. 02Generate an equal batch of variations using the junior agent, then prompt a senior agent to review and refine them against your brand guidelines.
  3. 03Launch both batches as controlled ads with identical budgets and measure click-through rates.

Measure: Compare the average click-through rate and cost per click of the reviewed variations against the unreviewed variations.

Guardrail: Stop the test if the reviewed variations show a decrease in click-through rate or if the additional API costs for the senior agent exceed the media spend.

Context & limitations
  • The speaker's setup requires specific local hardware and software configurations that may not be accessible or cost-effective for all operators.
  • AI agents can hallucinate metric anomalies or misinterpret data context, requiring human verification before making significant operational changes.
  • The described workflows rely heavily on the founder's ability to write effective prompts and maintain the agent's underlying knowledge base.

Listen to the conversation

0:000:00
Original episode description

Most founders are still stuck living in a chat window…and it's quietly costing them hours every day. In this episode, Nik shares his recent keynote from the 2026 Ecom AI Summit in NYC, breaking down exactly how he built "Jet Damon," his personal AI chief of staff, and the second brain system now running nearly every part of his business. Listen as Nik walks through the three stages most operator...

Transcript

EN

Welcome back to Limited Supply, the podcast where we get deep into the tactic...

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 in my weekly newsletter at nip.co/email, I'm here to share everything I've learned. The wins, the losses, the experiments, the tactics, and the insights, also you can unlock your next $100,000 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 gas 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. Now, to kick it off though, everybody's 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 name them jet Damon, which is why I call them my jet agent sometimes. But I basically put together a whole 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 set this up now, and just a few hours, you've got your own cockpit of your business, and performance metrics, and all that good stuff. But if you've got any questions, please email me directly, [email protected] that's [email protected], or DM me on Twitter @MrSharma, 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. All right, so without further ado, I 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.

All right, guys. How are we feeling?

“Yeah, I know Jimmy just did that, but you guys are a little quiet, all right?”

And I got the jitters, because I'm the first one here, and this room is packed. So let's give me a little another one. Who's ready for A, I saw it, 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 turns 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 five times a minute to Q&A, and then I'll be here all day too, okay, somebody's got anything else.

So I'm going to bring this up, so as Jimmy mentioned, you know, I talked to Jet, probably more than I talked to my parents, my family, anybody else, and how did I come up with a new job? Well, I just, I was just chat GPTA, give me a strong douchebag looking finance guy name. That's my guy.

All right. When I introduce Jet, then do an email, comes off strong.

That's always my recommendation, making a strong man.

So, you know, today we're going to go through these six things, so I'm going to prompt to lie that editorial right now. So let's actually start with that. So give me a brand name. I want somebody raise their hand and you can get a brand name right here.

Markos popcorn, one more time. Burkos popcorn, BGRCO, all right, cool. And then let's talk about a reason that somebody would buy a popcorn. Is it healthier? Is it tasteier?

Is it better price? Most expensive popcorn in the world.

“What do people say when they try your popcorn for the first time?”

Okay. And who is buying this popcorn? Is it moms? Is it dads? Is it kids?

Is it anybody? Okay. So, we've got Burkos popcorn, expensive product, very tasty. Does it get shipped to your house? Is it a right-fresh?

How, yeah, all that? Cool. Okay. Still aren't Chicago. And is it a 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 going to do, we're going to do this live promise on my phone. I'm going to use an app called Whisper Flow, which is basically replaced by keyboard. You guys have not heard of Whisper Flow to W-I-S-P-R, F-L-O-W-2 words.

Whisper Flow is amazing because if you've ever talked to chat, GPT, and done the mic feature,

you know that you can say, "Hey, I got to go to the grocery store, I got to get eggs,

Bananas, bacon, bagels," right, and it turns it into a bulleted list versus j...

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 the live publisher site that I set up last weekend. It's a WordPress site, you know, some of you guys may have that for app tutorials, for listicles, the pre-cell pages on your site, and from this prompt, it's going to go live on the site.

So, I'm going to start reporting. Yo, so, I need an advertiser for Berko's popcorn. It's called B-E-R-C-O-A-Poster-V-S.

“So, go to 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 some of these sweet at the end of the day. So, Berko's 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 popped warm? Oh, yeah. Let's put a customer test in the morning.

What's something somebody would say? Nice, so a big test for my nihilism, 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 with. 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 our message now, so everything feels natives. Now, you know, this is kind of where we've gone. This is like a five stage of AI that I feel like we know about right now. The first one is basically just chat GPT, 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 will was on stage talking about Lodi, and that was revolutionary because we've never

been able to take models and bring them into applications. But that was a closed loop system, right? That is 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 will, you know, it's happening at a high level.

Then we move to stage three, which is I think, where we got 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 will, 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 does his own platform dashboard warehouse management system. You can pull in, brok, to view something, you can pull in, clawed to do another specific thing. He can pull open AI 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 the build-out, pigenetic teams, right? And I feel like if you guys are on Twitter, or anybody on Twitter here, okay. So 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 when 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 do we do this conference? Well, at the beginning of the year, I felt very behind. I, you know, spent the last year trying to figure out how to sell Sharmor 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 the day, you gotta be unemployed, because everything is moving so fast. One week there's a model from anthropic, the next week that's open AI, the next week that'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 done it just using the chat window, right? This got really good last year. You export your meta, you know, raw data, you throw it in, you ask it some questions. You export reviews, you throw it in. Hey, help me make a list of gold based on what people are talking about.

Everybody's doing this. And this is kind of the problem here is that you're in a closed loop session. So, you know, if you're using cloud, you know, you can only use it so long for you to make a new session. Or you're not in control of how this data is being stored or 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.

These are people using codecs.

They're using cloud code. They're using madness. By the way, show of hands.

“Who's using, who uses OpenAI as their default model?”

Chat and GPT. Nice. We'll use this claw and drop it. Wow, there we go. OK, and then anybody here using Manus?

OK, about 10, 15 people. Guys, if no one's told you about Manus before, it's because they're a Chinese company. OK, Meta, just tried to buy him.

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 it

to drop it, or you're using one model, you're out of using Opus or Sonnet or Fable whenever it comes back. But if you're using Meta, 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 sub-task. And then brings it all together, which is why the final product of Manus is usually 10 times better than, you know, if you're going to use Cloud Code to do the same thing. Now, the last piece or the third group of people

is the second brain.

“So, you know, these are everybody's buying”

Machminis. They might be setting up VPS instances. I think that's the right turn. 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 open cloud. You guys heard of Hermes or OpenClaw? Cool. So some of you have.

So I'm going to explain it for those who don't know. So a Mach Mini, right, is basically your shell. Hermes or OpenClaw, I'm going to keep saying Hermes, though, because Hermes is the better version. So all you guys that try to do this later,

use Hermes, not OpenClaw. Hermes is the software that allows your Mach 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 Mach Mini, you can actually build out separate agents. So 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 IMES agent. You can set up Slack greens. You can add this person to your Slack as well.

And yet, 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 Mach Mini, one, you can yank that power cord. If the world starts to be taken over by AI,

“which I think is a good safety tail book, too,”

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

quad 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 replete to make wet pages and landing pages. I was using Manis to build apps. I built a ton of mobile apps, web apps, games with Manis.

I used Cloud to do all my writing and my prompting. So even if I was going to write a really good prompt, I'd list per flow in the cloud, get the prompt, and then put that prompt into another cloud session. And I used Chat, GPT, or Vrock, really,

to my research. And what I ended up doing was just basically consolidating everything in the one. So when I was using Chat, GPT, or Cloud, I'd use it across everything.

Maybe five or six different work projects, personal health, random personal research, A, 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 jets from my home 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, anything that came up as a red flag. This is what you got today. You're 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 you would do that end-to-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 assistance or VAs

that you work with on your team, yeah, I completely have.

If you've ever wanted an assistant, it was basically always

on call, Hermes agents, 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 paying Slack, Fireflies, email, Meta, Motion, Google, Shopify.

Calling 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”

and dipped in the last hour, cool. That's going to make a note. It's going to prompt three or four other things that happen, including in the alert to main. I use it for 20 different agents

about right now. So in between one of my projects has an individual agent, individual personality. It has its own set of knowledge that it's using. 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-up, the info, I says, 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, he's the knowledge center. So how many times have you just randomly-- you've got a team, five, 10, 15, 1500 people,

and somebody's just got a random question. But that person's not reachable. You need to get the answer for him. 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 Herbie's agent, because it's got a full kind of up-to-date set of knowledge. So like Jimmy said, I've set up a 100 agents now. Probably 30 or 40 macminies 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, your basically-- everything you can think of.

Your Stripe API, anything you can plug in, you do. And 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, this is a password get excited and 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, right? And now, every one of my agents can go and reference this.

As a document, anytime I need to build a website and a editorial, a listicle, whatever. The second one, I was able, because Jet knows exactly who I am, what I stand for, how I speak, my total voice. Jet built me a full brand book that I was able to now.

I can just say, hey, any agent I'm working with, whether it's my own, whether it's a random incognito charge in ET 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 by 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 which flow? I need to say to this ad, because I want to copy it later.

Yeah, you guys are not raising your hands or all wires. But I did that, right? So I did a sit, I sap a function with Jet, and I said, hey, if I ever screen shot an Instagram ad and send a year away by text, just upload it here and save it.

So now I've got hundreds of ads, landing pages, listicles, anything I come across, I can see it. And then I can go reference it, I can 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 say it's the title, and they change the first block, but the rest of the page is the same, right? So it just gives you the ability to harness all the

“information and put it all together. That's what, that's the first”

use case of Jet in my opinion. So again, you connect everything transcripts, call recordings. You connect to Monday, clicked up a song on your Slack. Any project work. And then what you do is this last die, which is a word you'll

probably hear later today is Plum Crum, CROM, CROM. And I'm not going to do what it stands for, right? 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 prom job set up globally, meaning for every 15 minutes, I'm picking apps and getting updates. Now, you're just five very simple things that I think everybody wouldn't benefit from if they had a chief of staff

like this set up. So first of all, did daily grief? I mean, doesn't everybody want a daily grief? Just to know what you got coming up, calendar, events, conflicts, emails, anything urgent.

Follow up. So what people will you be talking about this? Full meeting, recaps, and action items. This is where all the sauce looks. 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 Jeff just see everything I see. The next one is email drafts in your voice.

So hey, you got an email. Jeff says, by the way, this friend just emailed or you know,

Your vendor just emailed you and said that shipping

is running a week later, production is running a week later. 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 Jeff 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 are ready to pay for, 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 based-- or actually, I'll start with the first one.

“So off authentication, login, it's like signed in with Google, right?”

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, well, 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 the 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 in basis says, hey, we're going to set up

a quick local instance and go in, login 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 strike your Shopify, your PlayView, they'll 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 like a Creative app.

You can plug in via MCP into Cloud.

“So now, Cloud 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. 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 now goes beyond just your personal layer. And also, one thing I want to call out here,

this now includes Versel, Loveable, Webflow, Hydrogen. I don't know if you guys saw Shopify's updates last week,

but basically, everything is now plug-in play with agents.

So you've got a new landing cage in the same way I just prompted Jet. I cannot prompt a listicle to go live within Shopify, not on a sum of it. The third piece is now taking all this

and turning it into your cockpit. So you get the morning briefs. Hey, revenues up 12%. Kack is up 9%. You know, a clear view of the dipped.

“Here's a draft of what you should do next.”

You can say, hey, break, pack down by channel. Boom, you get the response within seconds. Hey, pull the best Q2 thread of that B control and why. Boom, here's the one that one. Now, the org chart.

So this, if I was just thinking in my brand brain, right? This is kind of how I like to think of my agent. So Jet 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 coper writer, basically a coper writer writes it. And then a senior manager goes and reviews it.

Since 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 M-dash in there? You know that's in my file.

Well, maybe the senior manager caught that if the original agent did it. These are, I'm a huge fan of giving these things a personality. Some people just call it health bot or copywriter bot or whatever.

I like to give me full personalities. And yeah, so I guess here's another quick example. These are like the agents I use most of the day. So a copywriter, junior pervertical. This copywriter sometimes, depending on how intense

the project is, is specific to the brand as well, or specifically to a category of products, depending on how you would do it. If you're somebody for example who's got assets and Facebook is 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 a research on Reddit.

It's maybe using the raw 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 editorial, 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.

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.”

I might say, hey, did you know that if you just add a 30% of your people are taking the post-purchase offer, you should change it to this, it might increase the 40 or 50%. It's a very actionable insights.

And then the last two I like to use by self, one is a health and wellness agent,

which is basically just gathering all the loop

or eating sleep levels, blood work, if you're a nerd like that. And then the last one I like to love to use is the entrepreneur and residents. 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 know, 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, refresh, boom, so right here, we've got this avatarial. And again, this is an avatarial side,

I just set out last week, right? Maeve Mag.com. So it's a very basic thing. I used Jen to edit it a little bit, get rid of some of the banners and whatnot,

but it pulls in a stock image, obviously it's fixed this. But then boom, let's be honest, now getting through it, fifth or summer, the hardest part is in the gym. The hardest part is 90, I'm on a Tuesday

when you're watching Netflix, the house is quiet and you drain a screaming for a snack.

Great opening, it's basically taking all the stuff

that knows about how I like to write out tutorials, and then, you know, ran that old process.

“But this is a pretty good version one to start with, right?”

If you're pumping out out tutorials, if you're pumping out listicles, 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 cage, a listicle, a bundle, and 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 an attribute well as a free piece of founders. 'Cause 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,

cack, revenue, performance, returns, any spikes. So you gotta know. So you 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's sake, right? Somebody mentioned that freight's going up eight percent next month, but I didn't even catch that. I didn't hear it because of something else. But jet caught that.

And jet flagged it, added it to my dashboard and said, hey, by the way, a watchout. So kind of going off that notion of how you go, takes 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, hey, refund rates, spiking. It's two point three acts of daily average. Attention, you're playing a welcome email conversion drop.

On the competitor side, hey, your competitor just launched the U.S., 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, created, hey, 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 town of the build path that I recommend, right? You set up a remisage, a map 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 thing a week. If you can stack one thing a week, four weeks, I mean, by the time summer's halfway through, you're gonna feel like an absolute day I ninja. Two things that the 10X, this, one, I mentioned,

whisper flow. It's a phenomenal app. I use it all day, top into 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. I came out last week, some 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 her bees and all that. The Toby, the Shopify CEO, even tweeted this morning that 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 work. So that's all I've got for you.

Stop living in the chat window, go build your second brain. And what I'm gonna do too, I'm gonna give away two Macminis if you sign up for the newsletter and you subscribe. So two people who sign up today

will get a Macmini and it's all I've got for today. Any questions?

- Thank you.

(audience applauds) - Questions? - You're at there. - So I have set up wireless wall. We're gonna drive that.

- Yeah. - We're in 30. - Where is the button under the hood?

“pretty free. Yeah, good question. So, this question was, I set up a lot of this in regular club, right, chat.”

Cloud code, cool. So, so two things. One, we have another presentation later, St. Corey, who built a platform called HQ. It's going to be very relevant to you because what, what HQ kind of does is help bridge

the Hermes versus the Cloud code is, right. 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, you know, or yesterday, I had to finish a slide through this, but I had to leave to go somewhere. I had to shut my laptop, I had to wait to finish the slide to, like, at home. Hermes is on I message. The Macmin 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

Cloud code setup can definitely work. Like the way I personally do it is I've got my Hermes age at setup. I've got HQ, which is basically the layer on top. And then any Cloud code, whether it's, you know, my personal laptop, whole laptop, whatever, those are all now also St. through HQ. So, all my devices are actually in St. at all times. I personally like that setup because I don't have a Cloud code laptop sitting at home, but I've also found Hermes to just be a little

bit, like, easier. It just kind of, like, you know, you're basically always using one chat,

I'm massive. So, it's a little more smoother. It still doesn't sound things like reset sessions

“and whatnot, but I think it just comes down to a preference state. The thing I like here too,”

right, is like my, my assistant can just text Jeff when something happens. I can also just set up, like, if being you were not going to get coffee, I could make a group chat with Jeff, say, hey, sending me address. That's a very simple, stupid example, but you get the point. Yes. When it comes to like your LP. Nick Bach, co, you said it's five miles long. Yeah, doesn't your agent doesn't chat, have to go through and read that whole thing,

versus a, like, wiki. Totally built out. Like, why, why that format? Yeah, a lot of the same, right. So, in, in a perfect world, right, I would basically have that LP brain, stores 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 plot already knows or, or jet already knows to reference that that file.

“Versus like a VPS. Yes. This question is basically, why use a map many locally versus like a”

headless server. That's right. Right. You're using a map many, but like, how they get connected to, like, all of the make those Markdown files in the cloud, as well as on your local, just have it sink. Oh, yeah. That way, you can do everything from your mobile. You do everything from a laptop, and you're back to me. Totally. Yeah. I didn't want to make this too complicated, but to, you know, back to his point, too, like, well, I normally do as I've HQ is kind of my dropbox, we would drive.

It's always on 10 minutes sink, right. So all devices are sinking. When my armies 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 cloud code locally, as well. Does that make sense? Yeah. Thank you. You have, of course. We have to make kind of right here on one side. Thank you. Thank you so much. Thank you for doing this. For your adversarial site. Yeah. Have you moved on from WordPress hosting to actually like doing,

let's say, a for cell host and just pushing it, pushing it out yourself doing your own navigation, all that kind of stuff. So you could push out as many adversarial sites as you want on a different host that's always there, connected back to your infrastructure. 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 it's, I know it's not going

to break. But to your point, like that everywhere, dot nicked.co site, that was vibe coded and then push live through for cell. So most of the sites I put up, even that identity dot nicked.co, you know, publish live through for cell. A lot of the landing pages now that I'll make going forward to, I'll probably publish in Versel because now Shopify will integrate that natively in the theme. Versel's not a subdoment. And you're not worried about having to drag along multiple

Instances of infrastructure for posting your, all your various sites.

handle that. Yeah. Yeah. Yeah. All right. Great. Thank you. Cool. All right. We'll do one last question.

“And then I'll be out on about all day, too. Any last quote? Yeah, right here. One, uh, we're”

two. 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, uh, you know, if any time you're

trying to transfer or or move to a new model, new, new AI system. So in this case, maybe use cloth 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 to, to, to, 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, you know, every, every prompt is like a certain amount of energy it can use. It'll just tell Cloud News all of it because you want the best possible document. That should export itself as a Markdown file, which is basically, Markdown file is essentially a plain text file,

but it has cash tags and a couple other symbols. So AI knows what the format of each text is.

And, uh, and then you take that drop that into jet or Hermes and say, hey, this is me. Learn it. Do your research and get wired up. And within two minutes it'll just start going to be like, hey, what's up? Uh, what's up, Nick? I see you're in New York. You live here. You know, you like to do this.

“I'm here, whatever 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 in the 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? Um, but yeah, to be honest, you know, my best advice with any of this is, is like, don't overthink it, right? Everybody thinks that when you open cloud code, you got to be technical, you got to prompt it properly. You could just open a whisper flow and dabble.

The more you dabble, 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 um's, your eyes, your random tangents, 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 next time to cut through the noise on CPG,

retail and e-commerce. If you enjoyed this episode, why not share it with a friend and be sure to subscribe wherever you listen so you don't miss the next one.

Keep reading