09:45Jacques Hopkins · Essay
ChatGPT Dots: The Warning Before You Install It
OpenAI's new always-on agent looks incredible. Here's the question the launch demo never answers.
September 30thThe argument in one line.
A persistent AI agent becomes a genuine business partner only when it runs on its own always-on computer, is fed a deliberate written brief about you and your business, and is reachable from a messaging app instead of a desktop window.
Read if. Skip if.
READ IF YOU ARE…
- A solo founder or freelancer who wants a hands-off way to triage email, calendar, and daily priorities without hiring a human assistant.
- Someone already comfortable paying $20-200/month for ChatGPT and willing to buy a dedicated always-on computer for an AI tool.
- A creator curious about persistent, memory-holding AI agents beyond a single chat window.
- Anyone who wants a concrete, step-by-step reference for macOS setup, always-on power settings, and Telegram bot wiring.
SKIP IF…
- You want a free or low-cost solution — this setup assumes a spare/used computer plus a paid ChatGPT plan on top of the Hermes software.
- You're not comfortable giving an AI agent standing access to your email, calendar, and chat history exports.
- You wanted a comparison of agent platforms — this video only covers Hermes Agent, presented uncritically by someone who appears to sell adjacent coaching.
The full version, fast.
The video walks through building a persistent AI business partner using Hermes Agent, a desktop app from Noose Research. The core mechanism: buy a dedicated always-on computer (he uses a used Mac mini), configure macOS so it never sleeps and auto-logs back in after a power blip, install the free Hermes software, and connect it to a paid ChatGPT/Codex subscription for its underlying model. The agent gets its personality and context from a hand-written 'operator brief' plus imported ChatGPT and Claude export files, then gets wired to Telegram so it can be messaged from any device. The actionable conclusion: start with one narrow task, review what the agent does, give feedback, and let its responsibilities grow from there rather than trying to hand over everything on day one.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:45
01 · Cold open: meet Rocky
Creator introduces Hermes Agent as his AI business partner 'Rocky,' claiming it's worth a $30,000/month employee for $100-200/month, running his email, calendar, finances, SEO, and ads.
00:46 – 02:10
02 · What you need: a computer and a ChatGPT plan
States the two prerequisites: a dedicated computer (recommends a used/refurbished Mac mini or a ~$12/month cloud computer) and a ChatGPT subscription starting at $20/month.
02:11 – 04:16
03 · Unboxing and physically wiring the Mac mini
Unboxes a used Mac mini from eBay, converts USB-A keyboard/mouse to USB-C, plugs in HDMI to a monitor, and powers it on.
04:17 – 06:05
04 · Initial macOS setup
Walks through region, Wi-Fi, creating a Mac account named after the agent (Adrian), skipping Apple ID sign-in, and choosing an age range, recorded via phone since screen-record software isn't installed yet.
06:06 – 08:10
05 · Locking in always-on behavior
Declines FileVault encryption on purpose so auto-login survives a restart, sets the Mac to never sleep, to start up automatically after a power failure, and to auto-log-in as the agent's account.
08:11 – 09:15
06 · Software update, energy settings, lock screen
Runs the macOS software update, restarts, configures Energy settings to prevent sleep, and turns off the password requirement after screensaver.
09:16 – 10:03
07 · Network and sharing lockdown
Confirms the firewall is off for the initial setup and turns off sharing settings for security before moving on.
10:04 – 10:56
08 · Installing Hermes desktop
Downloads and installs the Hermes desktop application, noting the install used to require terminal commands but is now a simple drag-to-Applications install.
10:57 – 14:23
09 · Connecting the model: ChatGPT/Codex
Skips the Noose Portal option, goes to Settings > Providers > Other Providers > ChatGPT or Codex subscription, signs in, selects GPT-5.6 Sol, sets reasoning to high, and enables fast mode.
14:24 – 16:38
10 · Giving the agent identity: the operator brief
Confirms the LLM is responding, then explains the 'operator brief' document (a description of the user, their business, goals, and relationships) plus importing ChatGPT and Claude chat-history exports as long-term memory, all pasted into the agent alongside a prepared setup prompt.
16:39 – 18:10
11 · Setting up Telegram messaging
Explains why Telegram (cross-device, simple) is his preferred interface, downloads the app, and signs into an existing account.
18:11 – 19:20
12 · First test message and memory check
Setup completes after roughly 10 minutes; the agent introduces itself as Adrian and correctly recalls a personal detail (a 4-month RV trip) pulled from the imported ChatGPT export.
19:21 – 20:45
13 · Wiring the Telegram bot token and user ID
Uses BotFather to create a bot and token, uses Raw Data Bot to get a Telegram user ID, pastes both into Hermes, and restarts the messaging gateway.
20:46 – 22:02
14 · First message through Telegram
Messages the new bot from Telegram and gets a live response introducing the agent (Adrian) as executive assistant and business operator for the creator's media company and programs.
22:03 – 23:05
15 · Next steps: groups, morning brief, and scaling responsibility
Recommends setting up a Telegram group with topics so the agent can work multiple threads, building a morning-brief automation, and asking the agent directly what else it could do based on everything it now knows.
22:53 – 23:06
16 · Wrap-up and opt-in pitch
Recaps that Rocky's responsibilities grew gradually over eight months of use, then pitches a free downloadable setup guide (email opt-in / waitlist) for a deeper program.
Lines worth screenshotting.
- An AI agent needs its own dedicated computer, the same way a human employee gets their own machine instead of sharing yours.
- Disabling FileVault and enabling auto-login after a power failure keeps an unattended agent computer from getting stuck at a lock screen it can't unlock itself.
- The agent's personality and business context come from a written 'operator brief' plus imported chat history exports, not from the underlying LLM alone.
- Reasoning effort and 'fast mode' trade speed for usage: fast mode runs about 1.5x faster but consumes roughly 2.5x the usage quota.
- Routing the agent through Telegram instead of a desktop app is what makes it reachable from every device with the same ongoing conversation.
- A Telegram group with multiple topics lets one agent work on several threads in parallel instead of being pinned to a single chat.
- Agent responsibilities should expand gradually: start with one task, review the result, give feedback, then hand over more.
- The creator prices ChatGPT usage in three tiers ($20/$100/$200 per month) and says the model choice, not the tier, drives the agent's output quality.
Treat the agent like a new hire, not a chatbot tab
WHAT TO LEARN
The setup that makes a persistent AI agent actually useful is less about the model and more about giving it its own machine, a written identity, and a growing scope of trust.
- Give any persistent AI agent its own dedicated, always-on device instead of running it in a browser tab you close every night — that's what turns it from a chat session into a standing employee.
- Configuring a machine to survive power blips (auto-login, disabled sleep, restart-after-power-failure) matters more for an unattended agent than it does for a normal computer, because there's no one there to log back in.
- An agent's usefulness comes from context you deliberately hand it — a written brief about you and your business plus imported chat history — not from the underlying model alone.
- A messaging-app interface (Telegram, or similar) rather than a desktop app window is what makes an agent feel like a colleague you can reach from anywhere instead of a tool you have to go sit down at.
- Speed settings and usage costs trade against each other directly: a 1.5x speed increase can cost 2.5x the usage, so match the setting to how much you actually need the extra speed.
- Expand an agent's responsibilities incrementally — one task, reviewed, then more — rather than delegating your whole business to it on day one.
- Be alert to promotional framing in tutorials like this one: the presenter sells adjacent coaching and collects email addresses through the 'free guide,' so treat enthusiastic claims (the $30,000/month-employee comparison) as marketing, not an independent benchmark.
Terms worth knowing.
- Hermes Agent
- A desktop application from Noose Research that runs a persistent AI agent on a dedicated computer, connecting to a chosen LLM provider and to messaging platforms like Telegram.
- Operator brief
- A written document describing the user, their business, goals, and relevant history, loaded into the agent so it has consistent context about who it's working for.
- Noose Portal
- A subscription product from Noose Research (the maker of Hermes Agent) that bundles access to multiple AI models, offered as an alternative to bringing your own ChatGPT subscription.
- FileVault
- Apple's built-in full-disk encryption for Mac. Leaving it off on a dedicated agent computer avoids a password prompt blocking auto-login after a restart.
- Fast mode
- A Hermes Agent setting that speeds up model responses by roughly 1.5x at the cost of consuming usage quota about 2.5x faster.
Things they pointed at.
00:46toolMac mini
01:28toolChatGPT
10:26toolHermes Agent (Noose Research)
10:57productNoose Portal
19:36toolTelegram
19:38toolBotFather (Telegram)
19:58toolRaw Data Bot (Telegram)
Lines you could clip.
00:00
“Hermes Agent is the greatest tool that I've ever used.”
00:08
“I now have the equivalent of what I'd say is the value of a $30,000 a month employee.”
19:13
“In spring 2025, you and your family took a four month RV trip around the country and homeschooled your daughters for one semester.”
22:19
“Rocky's responsibilities grew as I gave him more work, as he earned more trust with me.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
Hermes Agent is the greatest tool that I've ever used. It's allowed me to not just work with AI or even build simply an AI agent, but it's allowed me to build Rocky, an actual AI business partner that works for me 24 -7, understands me and my business, and is just a ridiculously smart business partner I've been working with for most of this year so far.
I now have the equivalent of what I'd say is the value of a $30 ,000 a month employee. for a cost of only one to $200 a month. Rocky manages my email and calendar, my finances, my SEO and ads.
He runs my productivity and priority systems. So I always know exactly what I should be working on at any given time and nothing falls through the cracks. And he proactively takes work off my plate when he knows that he can.
Since working with Rocky, I get more done, do better work, make more money, and I'm just having the most fun in my work that I ever have. And so in this video, I'm gonna show you exactly how to set up your own version of Rocky so you can have an AI business partner too and experience these same things that I have. All right, look, you basically need two things to get started.
First of all, you need a computer for your new business partner. When you hire a human employee, You wouldn't just let them share your computer.
You would get them their own computer. So I highly recommend getting a Mac mini for this. You can absolutely go used or refurbished, or you could even go with a cloud computer option and pay around $12 a month for that.
I'll include a link below if you want to go that route, but I prefer having my own physical computer in my possession where I can control it for my AI business partner. And that's the way I'm going to do it in this video. And the second thing that you.
you need is a ChatGPT subscription. Hermes agent works unbelievably well with the latest models from ChatGPT. And yes, you can tie just about any model in the world into your Hermes agent.
But at the time of this recording, the ones that work the very best with Hermes comes from OpenAI and ChatGPT. So I do recommend going with that. You can start with the $20 a month plan.
once you start seeing value from this you may quickly upgrade to the 100 a month plan and because i use this all day every day i'm on the 200 a month plan and for some of you maybe those monthly fees sound like a lot but i promise you within just the first month you should get a return on your investment if you do this right if you do this well and so those are the two things you need a computer somewhere for your hermes agent to live on and use in a chat gvt or similar subscription the hermes software itself is free and i'll show you how to set all that up here in just a minute so with all of that out of the way let's go ahead and get started all right here is the mac mini that we are going to set up i just got it in the mail it's actually used from ebay but i know that it does work let's go ahead and start setting it up so i've got the mac mini and i've got the power cord so i'll go ahead and plug in power cord to the back and give it power and then to make things easier on yourself you will want a hardwired
keyboard and mouse. I have an old keyboard and mouse lying around that I'll use for this purpose, but I have an M4 Mac mini, which only has USB -C ports and these old keyboards and mice are USB -A. So I got these nice little converters on Amazon, really cheap.
I'll put a link down below for these. So I'm going to convert. this keyboard and mouse to USB -C and plug it in right to the front of the Mac Mini.
And I'll do the same for the mouse and the keyboard. And then I'll get an HDMI cable to plug into the back of the Mac Mini and attach it to a monitor so I can actually see what I'm doing. All right, so now we have a monitor plugged into the HDMI port.
We have our power cable plugged in. We got our mouse and our keyboard plugged in. The only other thing you might want to plug in at this point is an ethernet cable for hardwired internet if you have access to that.
Otherwise, we'll just do... which is what I will do here in this video. Now we go ahead and power on the Mac mini.
There's a button right on the bottom. And then once you do the initial setup, feel free to switch to a wireless and keyboard and mouse if you would like. This hardwired keyboard and mouse is just for this initial setup.
All right, for this part, I'm gonna record my screen with my phone because I can't install screen record software on it just yet, but that will change shortly. So here's my screen. First thing you need to do is select your country or region so for me i'm going united states then this is going to be your agent's computer so there's nothing to transfer we're going to select setup as new i'm not doing accessibility now i can connect to my wi -fi all right now we do create a mac account this is where i'm going to start naming it based on the agent name that i picked out so my agent is adrian this is rocky's mate and the account name will also be adrian set up a password You can put a hint there if you want to.
And we're not going to sign in with our Apple account, so I'm going to uncheck this box and hit continue. And we don't need to sign into an Apple account, so we'll come down here and select sign in later in settings. And click skip.
Agree to the terms and conditions if you so choose. Agree. And you select your age range.
So I am personally an adult over 18, so that's what I will select. We're gonna enable location services on this Mac so that time zone sets itself automatically. And I'm going to not share any data with Apple.
Screen time, I'll set up later. I'm gonna skip Apple intelligence. And we do not want to turn on FileVault.
So Apple Silicon already encrypts stored data. And we're gonna need this computer to automatically log in in case you get a power blip and it restarts. We want our agent to come back online and FileVault turned on would prevent this from actually happening.
So we're gonna do not now on FileVault and continue. You can choose your look between dark mode and light mode. I like to do auto.
All right, and we do not want it to update Mac automatically because we don't want it to reboot for us and kill our agent or anything like that. So we're gonna click only download automatically. All right, so Mac OS is starting.
I'm going to click get started. All right, so the very first thing I'm going to do is install some screen record software. So I have to hold my phone like this up the rest of this video.
So let me go ahead and do that. All right, fresh Mac OS installation. The very first thing I'm going to do is run any Mac OS update.
So I'll come up here to the Apple menu, click system settings, software update. And we're going to update now to the very latest and greatest. It may prompt you for whatever password you set up.
So go ahead and enter that in and let it do its thing. While we wait for the update to complete, we can go ahead and set up some of our energy settings so that we can be confident that our agent is always online. Even if a power blip happens or power goes out for several hours at a time, there's a way you can set this up to where your Mac mini will automatically come back online once.
power is restored. So to do that, we're going to go to energy here inside of settings. We're going to turn on prevent automatic sleeping when the display is off.
So we don't want our computer to go to sleep or shut off at any point. We want this to be an always on 24 seven machine and then start up when power is connected. We're going to select after power failure.
Then in our settings, we're going to scroll down to users and groups and under automatically log in as we will select our agent name that we set up here. Once again, if he needs a password, go ahead and type that in and we'll type it in again here. And that makes sure that anytime it reboots, everything automatically comes back online and we're signed into the right macOS user and any interruptions overall are mitigated as quickly as possible.
And then we'll come to lock screen. And it's going to turn the display off when active for 10 minutes by default. This does not turn off the computer in any way.
So that's totally fine. It's just turning off the display. You can change this if you'd like to.
And then require password after screensaver begins or display is turned off after five minutes is the default. I personally like to go never here. But if for security reasons, you would like to leave this on, your agent still will be running.
But personally, I like to put this on never. All right, now it just got a notification about a restart about to happen, and I'm sure that's because of the software update. So I'm going to go ahead and let it restart and complete that macOS update.
And here in the software update section, it's telling me it needs a restart, so I'll go ahead and do that restart now. All right, my Mac has restarted. I see this little notification at the top right saying my Mac has been updated to macOS.
Well, the version as of this video. And if I go to general and software update, it's saying that I am now up to date, which is great. Now in settings, we'll go to network and we'll make sure the firewall is off for now.
This may be something we wanna turn on later, but for this initial setup and connection, it's gonna make our lives easier if we in fact leave that off. Next, you wanna go down to privacy and security, scroll down, and let's just confirm that firewall vault is in fact off we should have done that in a very very initial setup when we were first booting up but it's just good to confirm that then we'll go back up to general and then we'll scroll down to sharing and let's just make sure all of this is off for right now just for security purposes And at this point, our initial Mac OS setup is complete.
At this point, you could go ahead and do a reboot. And if you wanna take it a step further, go ahead and pull the power, pull the power cord on your Mac mini and make sure when you give it power back, everything comes back up and is working like a dream. Once all that is good, then we can go ahead and start our Hermes setup.
And by the way, if you wanna step -by -step checklist guide that you can download, print out, work through on your own, I've got a link to that down in the description below. absolutely for free. That's going to be a valuable resource for you to follow as you go through this process for yourself.
Next, we're going to open up a browser on your Mac mini, and we're going to download Hermes desktop. We'll find the Hermes desktop application download link here. I'll include that in the description down below as well.
Here, I'm going to click download for Mac OS. And this process is so much simpler than it used to be. We used to have to do all the installation through terminal commands and it looked really geeky and techy.
And now they've made it a lot simpler for us, which is great. So find that in your downloads folder. I'm going to click this Hermes setup icon.
It's going to open this window. I'll drag my Hermes icon into the applications folder. Then I'll open my applications folder, find that Hermes icon and open it up and click the install Hermes button.
And it's going to run through this process of starting to build out and install our Hermes agent on our Mac Mini. All right, prerequisites installed successfully and Hermes installed successfully. So then we'll click launch Hermes.
So the first thing it wants you to do is connect up to what they call the Noose Portal. So to be clear, Hermes agent is a product made by a company in Austin, Texas called Noose Research. And they do have a product called Noose Portal, which is a subscription you can pay for and access to a lot of AI models.
I personally do not think this is going to be the best way for most people to tie in a model to their Hermes agent. What I'm going to do, is I'm gonna show you my recommended way, and that is simply to tie it into one of the very best models in the world at the time of this recording and works so amazingly with a Hermes agent, and that is Chad GPT's GPT 5 .6 Sol.
So I'm gonna click, I'll choose a provider later. But that is exactly the first thing that we really need to do is we need to give our agent a brain, an LLM, an AI model to be powered on. So we're gonna come to settings.
and then providers, and then other providers. And then we'll go ChatGPT or Codex subscription. Now, at this point, you can tie in whatever AI model you would like to tie into.
I'm just telling you, I've been using my Hermes agent for months and months and months. This is literally how I work now. All day, I'm just prompting my Hermes agent to do the work for me.
So I've got a lot of experience here and I can tell you the Hermes agent combined with ChatGPT 5 .6 Sol at the time of this recording is the best way to go. And that's the way I'm gonna show you how to set it up in this video. So for me, I'm gonna select this option, ChatGPT or Codex.
Codex subscription, and I've already got a ChatGPT plan. ChatGPT plans come in three levels. Well, really four if you count the free plan.
I do not recommend the free plan here. Either go with the $20 a month plan, the $100 a month plan, or the $200 a month plan. I'm on the $200 a month plan because I'm using this all day, every day, but you can certainly start at the $20 a month plan.
The difference between the three is simply usage. There's very little feature difference between the three. It's just how much you're going to actually use.
the model. So you can start low and you can always increase later, but please make sure you have at least that $20 a month plan. And then we're going to log into that account and tie that plan in together with our Hermes agent.
And now we enter the code that the Hermes app gave us to connect the two together. And this says signed into Codex. If you're not familiar with that word, it is a product by OpenAI by ChatGPT.
If it means nothing to you, just think of it as basically ChatGPT. All right, so we're connecting up our Hermes agent to our Chachapiti subscription. It actually says now that we're connected, I'm going to click begin.
Now that we've linked those together, we'll go out to model and make sure GBT 5 .6 Sol, or whatever the latest and greatest model at the time that you're watching this, maybe there's something newer, maybe GBT 6 has come out. That's the one I do want to be using, and that is selected. Reasoning medium is fine.
I toggle between kind of medium and high. I like to air on high as my default mode, and you can even enable fast mode. I use fast mode.
Fast mode is amazing. It makes it about 1 .5 times faster, but it uses your usage about 2 .5 times faster. So that's completely up to you, but I'm going to go ahead and enable fast mode and click apply.
Okay, now it connects out of settings. All right, now that our agent somewhat has a brain, we'll come here and just say, hey, are you there? And we'll see if it actually responds.
We haven't set up anything about its identity or any information about you yet. We're just checking to make sure that the LLM is responding. Yep, I'm here.
What do you need? Beautiful, that's a great start. Now comes the big important stuff.
So here's what we're gonna do. We're gonna give it a few files and a long prompt that I have the prompt prepared for you. We're gonna give it, its name, we're gonna give it its identity and a lot of information about ourself.
So the first thing you need to put together is something I call an operator brief. And this is a document, like a text document, a Google Doc, something like that, which just describes you and your business and your goals and your life and your relationships and anything you want your operator to know about you and your history that will be relevant.
I've got a resource down below that will share with you the easiest way for you to put this together. I've already done this and I've got my operator. brief right here, ready to load into my Hermes agent.
The other thing that you may want to import at this point is any AI program history that you have. So you probably have been working with ChatGPT or Claude for a long time, and it's got a lot of great information about the way you work, who you are, some of the projects you've worked on in the past, and we can go ahead and export those from those programs and import that into the memories of our Hermes agent if you want to.
I'll put instructions below on exactly how to export that from Claude and from ChatGPT. I have those exports from my account here. I've got my Claude export.
I've got my ChatGPT. export and I'm going to include those into what I'm about to do as well. So the next step is to grab the prompt that I have available for you.
It's a long one and I'm going to copy the whole thing and paste it right here in this chat with my Hermes agent. So there's one thing you need to change about this prompt. Everything is good except there's a blank right at the beginning where you give it its name.
So mine is named Adrian so I'm just going to replace that. I'm going to attach all of these files to this message right here.
So I've got my attachments, I've got my prompt, and I'm going to submit that in. And it's going to take its time and start setting itself up for us. Now, while it's doing that, you can start preparing the messaging platform that you primarily want to interact with your agent with.
I imagine you don't want to go to this Mac mini and open up the Hermes desktop app. every time you want to communicate with your agent. That's just not very practical.
It's just not very easy. You can absolutely do that, but that's not how I work. I work with a messaging app called Telegram because it's easy, it's simple, and I can access it from all of my devices.
It's the same conversation no matter which device I am on. So let's go ahead and get you set up on Telegram if you're not already. what you'll want to do is go ahead and download the Telegram app on whatever computers you plan to use.
For the sake of this demonstration, I'm gonna go ahead and download it on the Mac mini itself. That's an optional step. I don't actually communicate with my Hermes agent through Telegram on its own computer, but because I'm sharing the screen right now, I'm gonna demonstrate that.
So do that on your main work computer as well as your phone too, and sign into Telegram or sign up for a Telegram account. It's completely free. So I'm going to download Telegram on this Mac device.
If you're on a Windows device, obviously you'll download Telegram for Windows. You can be installing it on your phone as well, if you would like. I've already got a Telegram account.
If you do not, then you'll set one up at this point. And I'm logged into my Telegram account. I've already got some bots I communicate with there.
If you have a fresh account, you won't see many. things to communicate to there over on the left just yet if at any point you see something like this it's asking you for your approval and at this point we're going to want to hit just hit run all right and for me that took about 10 minutes and it says setup is complete it gave us a little bit of a rundown on some of the things that it did so that was a lot your agent is now set up let's test it real quick hi jock I'm Adrian, your executive assistant and business operator.
Here's my job. All right, a little bit about me. Let me see.
Tell me something from the past that you wouldn't know otherwise. So I know the setup first. All right, so that's just gonna dig through and see if it found anything in like the exports or anything like that, just to confirm that this worked.
Yep, in spring 2025, you and your family took a four month RV trip around the country and homeschooled your daughters for one semester. Awesome, he said that came from your chat, GBT history export, awesome. So we are up and running, we are good.
The next step is to finish that Telegram setup. To do that, we'll come over here to messaging and then Telegram. So it basically just needs two things.
The first thing is a bot token. So to do that, we'll go to Telegram and we're going to search for bot father. And that's how we set up a new bot in Telegram.
So in the bot father, you're going to click slash new bot and hit enter. And then you give it a name. So this is the name of your agent.
So I'm typing in Adrian. And then you have to give it a bot name, which you will basically never see. So what I recommend you do is the name of your bot.
and then underscore your last name and then underscore bot. It just has to end in underscore bot. You'll almost never see this name anywhere.
All right, so that's now set up and we have our token. So we copy that token and we paste it here in the token field. So to get your Telegram user ID, that's your ID that it's allowed to communicate to, we go up here in the search and instead of Botfather, this time we're doing raw data bot.
And it's this one that just says raw data bot. So I'll click on that. and we'll do start.
And once I click, I get my client ID. That's what we need right here. So I'll copy that and paste it here.
Now we need to go down here where it says gateway ready and click on this open system panel. And it says messaging gateway stopped. We'll click restart gateway.
And now it says messaging gateway running, which is exactly what we want. All right, now you can go back to the conversation you were having with Botfather. And you see the name of your bot.
You can click on that. And now you can click start here. And let's say hi.
Who are you? And we can see our bot is typing. It says we need to set a home, but our agent is responding.
I'm Adrian, your executive assistant and business operator. I help you run Hopkins Media planning, research, execution systems, and accountability across the AI operator bootcamp, OCG, and piano in 21 days. And so we've now got Adrienne running on her own computer.
She has context about my business and my life, and I can give her work through Telegram no matter what device I'm on. Now we want to get in the habit of getting your agent to start doing the work for you. The next step, if I were you, would be to set up what's called a Telegram group with multiple topics so that you don't have to be tied to just that one conversation area.
Your agent can actually work on multiple things at the same time and creating a group with topics in Telegram allows you to easily do that. And instead of creating the group and the topics yourself, you can prompt your agent to start doing this for you. That prompt will be in the setup guide below in the description.
And once you've got a group set up, the first automation I would set up is a morning brief. Every morning, you can wake up with the weather for that day, any news around your niche, and any important reminders that came through from your email overnight or your calendar for the day. You can set up that morning brief any way you would like.
And the last thing that I want to advise you on before we get out of here is that your agent now knows so much about you. In terms of what it can do for you and what it can set up next, just ask it. Hey, based on everything you know about me, my history, my background, my goals, what can you help me with?
What can you set up for me? For me, Rocky is doing all kinds of things. And I mentioned some of these at the beginning.
He is triaging my inboxes, which is just a huge, huge time saver. He's helping me manage my priorities and productivity. He's doing SEO.
He's helping me manage ads. He creates graphics for me. He updates my website for me.
He helps me track my client progress. And the list goes on and on and on. He's even doing all kinds of personal things for me, like helping me with my workout plans and nutrition.
So start getting your agent to do the work for you and interact with your agent. That's what my work has transformed into. And this is so, so powerful.
I've been using this now for about eight months and it's completely changed the way that I work and what I'm able to accomplish as well. And like I said at the beginning, I'm just having the most fun I've ever had with my work and even my personal life as well. So for you, if you followed along, this is your starting point.
Rocky's responsibilities grew as I gave him more work, as he earned more trust with me, as we connected the tools that I use and explained how I wanted things done. And of course, give him feedback over and over again. You can take the same approach with yours.
Start with one useful task, review the result, give it feedback. and build from there you will need to give me your email address in order to get that guide but what that does for you as well is it gets you on the wait list teach you to take a deep dive with this ai agent stuff and let me hold your hand as we build out all the systems that i currently have set up and some custom systems that would make sense for you and your work as well so if you want to learn more about the programs that i have to offer around this to take this deeper then that is the path for that as well so go ahead and opt in for that guide and that wait list and good luck with your hermes agent setup
The Hook
The bait, then the rug-pull.
He calls it the equivalent of a $30,000-a-month employee for one to two hundred dollars a month: an always-on AI agent named Rocky that runs his email, calendar, SEO, and priorities. This is the full setup walkthrough he uses to get one running for anyone watching.
Named ideas worth stealing.
00:46list
Two prerequisites for an AI business partner
- A dedicated computer for the agent
- A ChatGPT (or similar) subscription
The creator frames the entire setup as needing only two things before any software installation: hardware the agent owns exclusively, and an LLM subscription to power it.
Steal forAny 'minimum viable setup' explainer for a new tool category
22:03concept
Grow responsibility gradually
Start the agent on one useful task, review the result, give feedback, and only then hand over more work — repeat as trust builds, rather than delegating everything on day one.
Steal forAny onboarding flow for a new AI tool or new hire
How they asked for the click.
VERBAL ASK
22:53newsletter
“you will need to give me your email address in order to get that guide... and let me hold your hand as we build out all the systems”
Soft-pitched throughout as a free downloadable 'step-by-step checklist guide,' then made explicit at the very end as an email opt-in that doubles as a waitlist for a paid deeper program.
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
AFFILIATECommission earned if you click.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
