14:05Manthan | Lead Gen Man · Tutorial
The NEW Way to Use Claude Code in 2026 (Loops)
A 14-minute walkthrough of how polling loops and dynamic runs let Claude Code handle entire feature pipelines without a single typed prompt.
June 15thThe argument in one line.
Businesses will pay $3,000–$20,000 for a voice AI agent that calls new leads within ten seconds, because the average business currently takes 47 hours to respond and that gap is where most revenue gets lost.
Read if. Skip if.
READ IF YOU ARE…
- You want to sell AI automation services to local businesses and need a concrete, repeatable system to build and offer.
- You have a basic comfort with API keys and web hosting and want to see Claude Code scaffold a multi-integration production app from a single prompt.
- You are evaluating Hermes Agent as a self-hosted alternative to cloud-based automation platforms like Make or n8n.
- You already have clients or leads for an AI agency and need a proven done-for-you offer with real proof of sales.
SKIP IF…
- You need thorough explanation of Docker, Node.js, or server fundamentals — this moves fast and skips the basics.
- You need a free stack — this requires paid Twilio, ElevenLabs, and Hostinger accounts from day one.
- You are looking for an enterprise-grade production setup; the .env-in-Docker approach here is a beginner-friendly shortcut, not a hardened deployment.
The full version, fast.
The average business takes 47 hours to respond to a new lead; responding in five minutes instead makes you 100x more likely to reach that prospect. This video shows how to close that gap with an automated voice AI agent — built live in one afternoon — that calls, qualifies, and books appointments within ten seconds of a form submission. The full stack is Claude Code for scaffolding, Hermes Agent (self-hosted via Docker on Hostinger VPS) as the runtime, Twilio for telephony, ElevenLabs for voice, Cal.com for scheduling, and Google Sheets for logging. The closing segment covers how to package and price this as a $3k–$20k done-for-you offer for local service businesses.
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 – 00:47
01 · Intro: The $10,000 AI Agent Strategy
Hook — $10k claim, overview of the build, promise to show selling at the end.
00:48 – 01:07
02 · Proof of Concept & Real Client Contract
Shows a blurred signed contract for an $18,000 deal across four months as social proof.
01:08 – 02:06
03 · The System Blueprint: Speed-to-Lead Breakdown
Diagram showing Lead Form → Agent Call (10s) → Qualify / Book Inspection / Log to DB. The 47-hour vs. 5-minute response stat is explained.
02:07 – 03:45
04 · Setting Up the Hosting: VS Code & Hostinger VPS
Installs Claude Code VS Code extension, creates project folder, sets up Hostinger VPS with Docker Manager and deploys Hermes Agent via one-click catalog.
03:46 – 04:01
05 · Inside the Hermes Agent Dashboard
First look at the Hermes Agent UI — shows chat interface, tool list, error when no AI model is connected yet.
04:02 – 06:55
06 · Gathering API Keys: Anthropic, Twilio, ElevenLabs, Cal.com
Walks through getting API keys and credentials from four platforms. Sets up phone number on Twilio ($2.15/mo), picks voice on ElevenLabs ($6/mo plan), creates Cal.com event and API key.
06:56 – 07:57
07 · Setting Up Lead Capture: Typeform & GitHub Repo
Creates roofing inquiry Typeform with 7 fields, generates personal token for API access, creates GitHub repo to bridge code between Claude Code and Hermes.
07:58 – 09:33
08 · Google Sheets API Credentials (GCP Deep Dive)
Enables Google Sheets API in Google Cloud Console, creates service account, downloads JSON key, shares the Google Sheet with the service account email.
09:34 – 10:55
09 · Prompting Claude Code to Build the App Scaffold
Pastes a system prompt into Claude Code describing the full Node.js/Express app structure. Claude Code creates all folders and files. Second prompt creates the .env template with all 17 variables.
10:56 – 11:59
10 · Configuring the .env File
Fills in all API keys, Twilio credentials, ElevenLabs voice ID, Cal.com event type, Google service account JSON, Typeform secret, server URL, timezone. Explains each variable.
12:00 – 12:59
11 · Pushing Code to GitHub Safely
Prompts Claude Code to push to GitHub. Claude Code flags .env as a security risk and asks permission before pushing. User tells it to exclude .env — Claude Code complies and pushes everything else.
13:00 – 15:15
12 · Terminal Execution: Docker Container & Going Live
Inside Hostinger VPS terminal via browser. Creates docker-compose.yml with all env vars pasted in, runs docker compose up -d, confirms container is running, clones GitHub repo into the container.
15:16 – 15:56
13 · Connecting Typeform Webhook
Goes to Typeform Connect → Webhooks, adds endpoint URL (Hermes Agent domain + /webhook/typeform), enables it.
15:57 – 17:20
14 · Live Test: Speed-to-Lead Voice Demo
Fills out the Roofing Inquiry form as 'James Michael'. Within seconds receives a live AI call visible on screen. AI asks about the issue, offers next available slot (Tuesday June 9, 3:30 PM), books it. Calendar shown with appointment confirmed.
17:21 – 18:27
15 · Packaging, Pricing & Selling to Businesses
Framework for pricing based on lead value. Example offer script. Two pricing models: per-booked-appointment or $3k-$10k setup + $1k-$2k/month retainer. Plug for upcoming live workshop on client acquisition.
18:28 – 19:14
16 · Resources & CTA
Points to free School community templates vault for the Claude Code prompts and Docker Compose file. CTA to next video on 20 AI agency offers.
Lines worth screenshotting.
- The average business takes 47 hours to respond to a new lead; responding in five minutes makes you 100x more likely to reach them and 21x more likely to qualify them.
- Every lead a business fails to follow up immediately is lost revenue — which is exactly why a speed-to-lead agent commands a $10,000+ price tag.
- Claude Code can scaffold a complete multi-API Node.js app from a single natural language prompt, without the builder writing a single line of code manually.
- Hosting Hermes Agent on a Hostinger VPS via Docker Compose gives you a persistent, always-on autonomous agent for roughly $8-12/month.
- The .env file is your system's credit card — never push it to GitHub, even when Claude Code flags the risk and asks permission first.
- Pricing an AI agent based on the business's lead value, not your build time, is the key to charging $3k–$20k for a system that took an afternoon to build.
- A done-for-you offer positioning ('no extra staff, no ad changes') closes faster than a technical explanation of how the agent works.
- GitHub acts as the bridge between Claude Code (where you build) and Hermes Agent (where you deploy) — keeping them decoupled is the architecture that makes the system portable.
- Cal.com's API lets the voice agent check real-time availability and book appointments mid-call without any human handoff.
- Per-booked-appointment pricing aligns your fee directly with client ROI — they only pay when the system delivers a result.
Speed kills — and that gap is the product.
WHAT TO LEARN
The 47-hour average business response time is not a minor inconvenience — it's a conversion killer, and closing that gap with automation is worth thousands to the right client.
- Responding to a new lead in five minutes instead of thirty makes you 100x more likely to reach them — so the agent's speed, not its intelligence, is the core value delivered.
- Claude Code can scaffold a complete multi-API Node.js application from a single detailed natural language prompt, meaning the bottleneck shifts from coding to prompt quality.
- A GitHub repository acts as a clean handoff layer between AI-generated code and a self-hosted runtime — keeping the two environments decoupled means you can update the agent without touching the server config.
- The .env file is the system's single point of failure; an AI coding assistant that flags credential exposure before pushing to a public repo is actively protecting you from a costly mistake.
- Pricing an automation deliverable based on the client's lead value — not on your build time — is the mechanism that makes a one-afternoon project worth a $10,000 contract.
- A done-for-you positioning frame ('no extra staff, no ad changes') closes faster than a technical explanation of how the agent works, because the client is buying the outcome, not the architecture.
Terms worth knowing.
- Hermes Agent
- An open-source self-hosted agentic AI platform that can browse the web, execute code, manage files, and run long-horizon tasks autonomously. In this video it serves as the always-on server runtime that receives webhooks and orchestrates the voice call flow.
- Speed to Lead
- The practice of contacting a potential customer as quickly as possible after they express interest. Research cited in the video shows a response within five minutes (vs. thirty minutes) increases contact likelihood by 100x.
- Docker Compose
- A tool for defining and running multi-container applications using a YAML file. Used here to deploy the Hermes Agent and the Node.js lead-qualification app as a single persistent service on the VPS.
- Webhook
- An HTTP callback that one platform sends to another when a specific event occurs. Here, Typeform sends a webhook to the Express.js app the moment a lead submits the form, triggering the outbound AI call.
- Service Account (Google Cloud)
- A non-human identity in Google Cloud that allows an application to authenticate and access APIs on its own behalf. Used here to give the Node.js app permission to write rows into Google Sheets.
- ElevenLabs
- A voice synthesis platform that converts text to realistic speech. Used as the voice engine for the AI phone agent so the caller hears a natural-sounding voice rather than a robotic one.
- KVM2 (Hostinger)
- A specific VPS tier on Hostinger offering dedicated kernel-level virtualization with more RAM and CPU than entry plans. Recommended in the video because Hermes Agent requires sufficient memory to run reliably.
Things they pointed at.
02:56toolHermes Agent (Docker one-click)
Lines you could clip.
00:00
“Businesses are paying $10,000 for AI agents that realistically now only take an afternoon to build.”
01:19
“Every single lead that they don't follow up with is lost revenue for the business. Hence why you are able to charge $10,000 plus for this exact system.”
17:09
“Boom. Done. Finished. Let's go.”
17:21
“The build is the easy part. The hard part is not so much can we build this, but it's can we sell this and get clients.”
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.
analogy
00:00Businesses are paying $10,000 for AI agents that realistically now only take an afternoon to build.
00:06Thanks to Hermes, which is by far one of the most powerful agentic platforms in the world right now. And And it becomes even stronger when you pair it up with Cloud Code, which is why in this video, I'm gonna be breaking down this exact two tool stack and build one of these agents live right in front of you, start to finish on this screen.
00:25And the best part that I'm not gonna use any code. I'm not gonna have any team behind me, and I'm not gonna skip through any of the boring but important parts of the actual build. And at the end of the video, I'm also gonna go through the part which most people struggle with, which is how you can actually sell these systems for thousands of dollars to businesses.
00:44So if it sounds like something that you wanna learn, let's dive in. Alright. So before we get to the actual build, um, I wanna show you that we've done this, meaning that we've sold AI agents to dozens of different companies, one of which was a service company, and we closed the deal for $18,000 across four different months.
01:00On the screen right now, you should see a signed contract as proof that we've done this, and, hopefully, you can do it too. So with that said, let's get into the build. Okay.
01:08So we're gonna use ClawCode and Hermes agent to build a $10,000 AI agent, which in this case is speed to lead. Now there was a study that came out that reported that the average business owner takes forty seven hours to respond to a lead and responding in five minutes instead of thirty makes you about 100 times more likely to reach them and 21 times more likely to qualify them.
01:28Meaning that every single lead that they don't follow-up with is lost revenue for the business. Hence, why you are able to charge $10,000 plus for this exact system right here. So the way that it works is that we have a form.
01:38In this case, it can be a Facebook ads, Google ads, or just a simple form on the website. The lead will fill out the form, and then we have an AI agent that calls them, uh, within ten seconds of them opting in. It will ask them some questions.
01:51It will book a property inspection because it's a roofing company. So my roof is broken. It's leaking x y and z.
01:57So it will ask questions about that, and it will book the the actual inspection the free inspection for the for the lead itself. And then we add details to a database, which will be a Google Sheet.
02:07Now the first step is actually setting up the hosting platforms for both different softwares. So we have Versus Code for Clockcode, and we have Hostinger for the Hermes agent. To download Versus Code, you can go to visualstudio.com/download, download it for your desktop, and then you'll be able to get on a page that looks like this.
02:25You can go to extensions, and then you can download the ClawCode extension right here. Press install, and you'll see that you have it right here. And then we wanna add or open a new folder in our desktop in this case, um, and we can name this Hermes agent.
02:40There you go. Oh, yes. Open this.
02:43And we're all set up for the claw code that we're using as an extension inside of Versus Code. Now in terms of Hostinger, you can go to hostinger.com applications Hermes agent, and you'll get here.
02:52I recommend that you get the k m v two plan option just because it gives you the most flexibility with the power, but also the memory itself. And once you log in, you get the plan, you will get on a page that looks like this where you have your Docker, um, installed. And you can go to manage, and then you can go to Docker manager.
03:08You can install this. It will take about a minute. And once you have this here, you can go to compose.
03:12You can go to one click deploy. Look for Hermes agent instead of the catalog here. So you can do Hermes agent select.
03:21So make sure to copy this and save this somewhere safe. You can press deploy, and now it will take about thirty seconds to one minute to actually deploy in the server itself. Alright.
03:29Now the Hermes agent is opened. It's finished. We can go here to open, and it will ask us for a username and password, which is the one that we had before.
03:37So we can paste Hermes here, and we can also paste the password that we had as well. And now we're inside the actual system.
03:46As you can see from the beautiful interface, this is Hermes dashboard. It's where you chat to Hermes. You can say hello, And it says error because we have not connected the AI model yet, which is fine.
03:55But this is what you should see. If you don't see this, something went wrong, go back into the video. Alright.
03:59So that's it in terms of the hosting for Hermes agent and Cloud Code. Now we get to the actual softwares that we need to use. So the first one here is Anthropic for the API key.
04:07So you can go to platform.cloud.com, setting workspaces default keys. You can press create key, and you can name this Hermes agent or whatever it is that you wanna name it.
04:17You can press copy key, and you're fine there. Make sure you have enough credits here. I currently have $17 because you're gonna need this to run the agent itself.
04:25Then we have Twilio. So Twilio is a platform that we use to buy the number that we're gonna use. I recommend that you're on the plan of pay as you go because numbers obviously are not completely free to use.
04:35So you wanna go to the search bar, look for phone numbers, and then you can go to phone numbers here, and then you can go to set up a new phone number, Choose your country. Again, not all the countries are gonna be in Twilio, so heads up.
04:47But the ones that we wanna use in this case, US is fine. You can choose voice as SMS is not what we need right now. Toll free is fine, and then you can press search, and then you can go through the whole set up of buying the number itself.
04:58Put your credit card there, buy it, and and you're good to go. It's only $2.15 per month, so it's all good. It's not gonna be anything crazy.
05:04Now once we have this, so in this case, I already have a number here, copy your number and paste it somewhere safe as well because we're gonna add all of these into Clockcode itself. We also want the developer. We wanna go to API keys and auth tokens, auth token, and we wanna copy the account SID, and then we also wanna copy the primary auth token.
05:21So these are the passwords and the way that, you know, Hermes agent is able to access the number and is able for the number to call the lead whenever, you know, a a form is submitted. Alright. So now that we're done with the actual Twilio, which is the number that we're gonna use, we go to 11 labs.
05:34So 11 labs, you can get the $6 a month plan. You can go down to developers. You can go to API keys, and then you can press create key.
05:41Name the service. I'm not gonna put any restrictions. Create a key, and then you have the key right here.
05:46We also need a voice. In this case, you can go to voices. You can choose your voice.
05:50I can go to English. I can choose American accent, and then I can choose Hi, friends.
05:58I'm Lydia. Lydia is fine. Lydia is fine.
06:01We can go here, copy voice ID, and then you can paste it somewhere safe as well. By the in case you're wondering where I'm looking, I'm looking at a doc that I'm gonna paste everything in. And we're done with Elevenlabs.
06:08We can go to cal.com to set up the event for the voice agent to actually schedule. So you make an account on cal.com. It's completely free.
06:15Make an event. In this case, I called it free property inspection. You can go here.
06:19Make it thirty minutes. All the settings are pretty standard, you know, availability and so on. And you wanna copy the event ID.
06:26So it's the number that you see on the URL on the top. Paste the number somewhere, then go to settings, go to profile, copy your username right here because we're gonna need this.
06:34And then we also want an API key, which you can find down here. API keys, new Hermes.
06:41As you can see, already have four Hermes agents. Um, press create, and you have the API key here. So we're essentially laying the foundations for the whole system and making sure that the system has access to all the softwares that we Typically, do do this in the first step just because it doesn't bite me in the ass later because it's much easier.
06:56The next step is Typeform. So this right here is a software that we're gonna use to be able for the lead to fill out the form and then get called in ten seconds, uh, by the actual system. Go to Typeform, make an account, and then you wanna create a form.
07:07In this case, the form that I made, uh, contains the full name, phone number, email, services you need, describe your issue, where is your property located, and also when do you need help by. Um, that's fine. So make sure you have this.
07:18Alright. Once you made the actual form, we can go to forms here and we can go to profile. We can go to account settings.
07:24We can go to personal tokens, and we can generate a new token. Name this Hermes. All scopes is fine.
07:31You can copy this, and that will be your API key. Alright. The last step is making a GitHub repository where we go to make an account.
07:37We press new, name this Hermes, and then you can press create a repo, and then you can copy the link right here, and we can keep it.
07:47Alright. With that said, we're done with getting the API keys and passwords from all the softwares. I wanted to show you that part just because it is actually important to cover because you might be stuck actually finding these yourself.
07:57So I hope that was helpful. Now let's go to building the app on Cloud Code and start there. Hey, guys.
08:02Quick one here. If you are working nine to five and you do want to start and scale your own AI agency, then check out the first thing down below, which walks you through a full video on how you can do that step by step by working with me one to one. Now let's get back to the video.
08:15Alright. So the next step is getting the Google Sheets credentials to be able for the system to access Google Sheets to add the details of the lead that called in. So go to console.cloud.google.com.
08:24And by the way, all the links are below. You can go to select a product. You can make a new product.
08:29In this case, I already made one. Once you make the name, you should see the name right here. It will take about a few seconds to to download.
08:34You're gonna go to API and services. You wanna go to the enable API and services and go to Google Sheet. When you're in the Google Sheets, you can press enable, and then you are enabling the connection of Google Sheets.
08:45The next step is making the credentials, you can go to API and services credentials, and then you will be able to create credentials, service account. You can name the service account Hermes one 23.
08:56Create and continue. You can press done. Once you made this, you can see it here.
09:00I already made two before. So but once you go here, you can press this. You can copy the email here that we have, and then we need to go to keys.
09:08We need to go to add a key. Create a key here. JSON.
09:12Create it. And now it might get a bit bit technical, but don't worry um, because all we have to do is open the file. So it should open the actual JSON thing, uh, in front of you.
09:22By the way, this makes no sense to me whatsoever, so don't worry. I've just done it so many times that I know what to do here. Uh, you can copy everything from here.
09:30Let me zoom in. Everything from the start until the end and paste it somewhere safe.
09:36And essentially, now that this is active, the system is able to access our Google Sheet, which is actually the next step. So we're gonna go to sheets.new, create a Google Sheet, and then name this lead agent, and we can copy the ID of the sheet right here, which is anything from the d until the edit and paste it somewhere.
09:54So we made the folder in ClawCode. We're gonna go here, and we're gonna paste this prompt right here, which tells exactly ClawCode the structure of the project, how it's supposed to, uh, look, the different files that it's meant to have here, and how everything goes together. Alright.
10:06Now if you're wondering where you can get the prompt, uh, it's in the second link down below in my free school community. You can go to the classroom section, templates vault, and you'll find everything there. We have this here.
10:16We can press go, and we can do bypass permissions. And by way, you can also screenshot this, and you can give it to chat GBT, and it will give you the prompt itself. As you can see now, it's starting to make the actual folders and files, which is great.
10:26Alright. It just finished making the different folders. This is what we call the file directory where all the different folders are and files.
10:32I can now paste the next prompt, which again you can find, um, in the same document. This right here is gonna tell it to create a dot EMV file, not example, just dot EMV file so that we're able to paste all the different credentials and API keys that we just had. And in here, it's telling it, hey.
10:46When this happens, do this. When this goes wrong, do this to make sure that we have those guardrails when we are actually doing this. I'm gonna press go, and now it's gonna implement the next set of rules.
10:55Alright. So Clocker just finished making all the different files. One important thing is that we have to replace the keys here.
11:01So this is the dot EMV. This is the way that the system takes all our credentials, all our passwords, and it uses it to access the softwares when the system is running. So now we wanna paste all the API keys that we had before onto here.
11:13So everything from here on out, you should probably have. Right? Because we went through it before.
11:17The one thing that does change is the port, server URL, and time zone. So time zone, you can use your time zone. In this case, I have Asia slash Dubai, uh, because I'm here right now.
11:27Server URL is something that you can get by going to the Hermes agent, and you can actually just copy everything from before chat, antupuntil.cloud, you can paste this here.
11:38And then the port will be 3,000, which is fine. And all of these others, we can, uh, simply just copy this. You can either copy this here or you can just paste it right here.
11:47It's blurred right now, but this should be the Anthropic API key, 11 labs, Twilio, cal.com, Google, Typeform, um, the server URL, which I already have, and my phone number as well. Nope.
11:57I'm not showing it, so don't try. I'm gonna go here, file, save, and now we're good to go.
12:03The next step here is to push everything to GitHub. So we wanna take all the code and push it here because that is gonna be the middleman between, uh, Cloud Code and Hermes agent. Right?
12:11So copy this and say, hey.
12:14I want you to take all this code and push it directly to GitHub. Here is the URL of my repository. Um, let me know when it's done.
12:21Let me paste the link. There we go. So now I should take all of this, push it to GitHub so that Hermes agent is able to read it as well.
12:29Alright. So just as expected, it's gonna ask me if I wanna actually push the dot EMV. I'm gonna say, no.
12:34Don't push the dot EMV, but you can push everything else. So the reason why we don't wanna push the dot EMV is because our dot EMV contains all our service secrets, API keys, passwords.
12:44It's like your credit card number. You're not going put it in the web. Right?
12:47So we don't want that going through, so very good for asking. And everything else, yes. Alright.
12:52So everything has been pushed to GitHub. So if I go here and I refresh, then I should see different files.
12:58Perfect. Alright. So the next part is actually using the terminal.
13:01Now for those of you who are not technical at all, this might scare you, but stay with me. It's actually very, very easy, and I'm gonna give you the full thing to add as well, so don't worry because I'm also not technical so we're on the same page here. Okay?
13:12We wanna go to terminal here so make sure you're in Hostinger, go to terminal and then and then you can press and you can type exit right here and you have the actual thing itself. Now it's listening for any commands or any prompts that we give it.
13:23Alright. So this right here is gonna be the code that we have to paste inside the terminal right here for it to actually download everything. So we need to replace all the keys that we have here.
13:31So the first one is the GitHub repo URL. So if you remember before, I mentioned to copy the repo URL and paste it here, which I can do right now. I can paste it here.
13:42There we go. And now we have the API keys. So Anthropic, Eleven Labs, Voice ID, all that stuff.
13:47You can go and copy each one. Alright. I just pasted all my credentials here.
13:50Make sure you go through each one step by step, and you can also just copy this whole thing, paste it into chat gbt, and ask it, hey. What are some things that I need to replace? And you can go back and forth, uh, which is fine.
14:00And we have this at the end as well. K? Now in the doc that I'm gonna give you, it has all the instructions as to what the network is.
14:06It'll tell you all the different naming conventions of these different variables that we have to put here. K? Once you're done, you can copy this whole thing.
14:14We can go to the terminal again. You can put exit here. And the first thing we wanna do is we want to make a folder and open the file editor.
14:21So we do this by pasting this command right here. Just copy everything step by step. M k dir p docker lead app docker lead app and nano docker compose dot y m l for us to be able to do this.
14:33And then we can press run. And in here, we can include all the different EMVs and all the different passwords and keys and all that stuff. Okay?
14:39So paste this whole thing from the Google Doc that we made, and then once you have this here, you can press control o, enter control x, and we're back here.
14:48And then all we have to do is paste this right here, docker compose up SD. And now as you can see, it's pulling up the document. It says plus 12 out of 12 container lead up.
14:58K. Started. Pulled.
14:59Okay. Cool. That's all good.
15:00There's one more thing that we have to do in terms of the terminal. We have to paste this, uh, thing right here, which is gonna allow us to take the code, clone it, and actually use it inside the agent. I can press run.
15:10Alright. Cool. So if you see Hermes agent listening on 00003000, that means that the agent is now live.
15:16So Hermes agent, Cloud Code, GitHub, all the softwares are connected, and we're good to go. The last step here is the Typeform. So we got the scopes, but all we have to do now is actually create the webhook so that when we fill out the form, it sends the data directly into the system itself.
15:30So you go to the form, you go to connect, you go to webhook here, you press add a webhook, and then here, you wanna go back to the actual thing here.
15:40You want to copy everything up until cloud, paste it here, and put slash webhook slash type form and press save form or save webhook.
15:53Turn this on and now we can test it. Moment of truth, we can test it. If it doesn't work, we'll fix it.
16:00If it does, amazing. Alright. Finally, we wanna paste this command right here to watch this live, and we can now run the actual platform.
16:06So we go here, start, name this James Michael. Okay.
16:11Put my phone number here. I'll put my email here. Press okay.
16:15I need roofing services, that's fine and then my roof is leaking. I can press okay.
16:22Dallas is fine and then this week submit. Alright.
16:27So as we can see, I'm getting a call from the agent right now on the screen. Hopefully, you guys can hear it. Hi, James Michael.
16:33This is Hermes from Home Inspection Pros. Thank you for reaching out to us. Can you tell me a bit more about the issue or the inspection you need help with?
16:40Yeah. My roof is leaking. It's broken.
16:43There's a storm the other night. We need this done ASAP. Got it.
16:45I understand.
16:46We'd love to get an inspector out to you as soon as possible. What day and time work best for you? Tomorrow?
16:54The earliest available slot I have is Tuesday, June 9 at 03:30PM. Would you like me to book that for you? Just say yes or no.
17:01Yes. Yes. That's fine.
17:03You're all set. Your inspection is booked for Tuesday, June 9 at 03:30PM.
17:08Our inspector will be there on time. If you need to reschedule, just give us a call. Have a great day.
17:13Boom. Done. Finished.
17:14Let's go. We can see in my calendar right here, everything else is blurred, but we have the free property inspection at 03:30PM my time, James Michael with my email as well and my phone number too.
17:24Alright. So back to it. There's one more thing that I wanna show you which is the packaging, pricing, and selling it to businesses.
17:30So one important concept about the system is that it is kind of based on how much a lead is worth for the business. Because if we are converting more leads, we have to understand what is the number, what is the amount that the lead is worth for the business. And so you can charge anywhere from 3 k all the way to 20 k for the system because if a lead is worth a lot more for y business than x business, you can charge them a lot more.
17:52And that's exactly how it works with these kinds of clients. Now in terms of packaging, your offer can literally be as simple as we build a speedily voice system that calls and qualifies your ad leads in under five minutes and books them into your calendar without hiring extra stuff or changing your ad setup completely done for you.
18:07And like I mentioned, in terms of pricing, you can either charge per booked appointment or you can charge a setup fee plus a monthly retainer, which can be anywhere from a 3 k setup fee to a 10 k setup fee and a 1 to 2 k a month retainer as well depending on the business. Alright. So there where there is a full build.
18:22You've now taken two softwares, and you have built a working agent that businesses are actually willing to pay thousands of dollars for. Uh, that's the build.
18:31Right? The build is the easy part. The hard part is not so much can we build this, but it's can we sell this and get clients, which is exactly what I'm covering in my free live workshop later this month.
18:40If you wanna check it out, you can check it out in the second link down below. It'll be there. You can apply, and I'll see you there.
18:45Alright. And once again, if you guys want the resources and templates and prompts of this whole video, then check it out in the link down below in my free school community. Uh, you can go to the classroom section.
18:55You can go to the templates vault, and you'll find everything there. And now that you've actually built and packaged one specific offer for a business, check out this video on the screen where I show you 20 AI agency offers that actually make money in 2026. With that being said, I really hope you guys found value from this video.
19:11And as always, I'll see you in the next one.
The Hook
The bait, then the rug-pull.
Businesses are handing over five-figure checks for voice AI agents that call new leads within ten seconds of a form submission — and the presenter has the signed contracts to prove it. This video is a live, uncut build of exactly that system: Claude Code scaffolds the Node.js app, Hermes Agent runs it on a self-hosted VPS, and by the end a roofing lead named James Michael has a property inspection booked without a single human making a call.
Named ideas worth stealing.
01:19model
Speed-to-Lead Value Equation
- 47 hours = average business response time
- 5 minutes = AI agent response time
- 100x more likely to reach lead
- 21x more likely to qualify lead
The statistical case for why speed-to-lead automation is worth thousands to a business: the gap between 47 hours and 10 seconds is where revenue disappears.
Steal forAny B2B AI agency pitch deck or sales conversation with local service businesses
17:21model
Lead-Value-Based Pricing
Price the system based on what each booked lead is worth to the client business — not on your build time. A business with $10k average jobs warrants a higher setup fee than one with $500 average jobs.
Steal forPricing proposals for any AI automation deliverable where the output is booked appointments or revenue
17:55concept
Done-For-You Offer Script
- We build a speed-to-lead voice system
- Calls and qualifies your ad leads in under five minutes
- Books them into your calendar
- Without hiring extra staff or changing your ad setup
- Completely done for you
A single-sentence positioning statement that sells outcome, not technology.
Steal forAI agency service page, cold outreach, discovery call framing
How they asked for the click.
VERBAL ASK
18:27link
“Get the templates and resources — link down below in my free School community. Go to the classroom section, templates vault.”
Repeated twice — once at 7:58 (mid-roll with talking-head cut) and again at 18:38 (outro). Also promotes a 'free live workshop' on client acquisition. On-screen text card confirms the CTA visually.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
Visual structure at a glance.
