Greg Eisenberg joins Sam to walk through a six-step system for building a business using AI tools — from finding an idea with IdeaBrowser, to scoping an MVP with Manus, to vibe coding a prototype with Bolt.new, to automating customer acquisition with Lindy AI. The episode doubles as a live demo: Greg builds a fictional LLM SEO business called LLM Boost in real time. Sam’s reaction throughout is that this is genuinely mind-blowing and changes how he thinks about AI in his own companies.
Speakers: Sam Parr (host), Shaan Puri (host, referenced but not present), Greg Eisenberg (guest, founder of Startup Ideas podcast, LCA design agency, IdeaBrowser)
Introduction: Replacing a Full Team with AI [00:00:00]
Greg: By the end of this episode, Sam, we’re going to see me replace a developer, a salesperson, a designer, a marketer, a researcher, a product manager — with AI agents.
Sam: Okay, I’m in. Are you going to tailor this to someone like me who’s like a Neanderthal?
Greg: I’m going to tailor this to anyone who is an idea person. Anyone who listens to your podcast, My First Million, or my podcast, the Startup Ideas podcast. Anyone who considers themselves an idea person, a solopreneur, someone who wants a side hustle, someone who wants a business to make money and is interested in trends and ideas.
Sam: Okay, I’m into this, dude.
Greg: So we’re going to go through six steps. The first is how to find the right idea and trend. The second is sketching out the idea. I’m going to talk about the tools I use and I’m also going to give away all the workflows so people can just copy them. We’re going to scope out the MVP using a tool called Manus. We’re going to vibe code a prototype using Bolt.new. We’re going to vibe market the business and automate it using Lindy AI. And then we’re going to use an AI agent product manager using IdeaBrowser.
Sam: Because I’m above the age of 30, I feel a little uncomfortable using the word “vibe.” Do we get a pass? Can I say the V word?
Greg: You can say the V word.
Sam: I don’t know. Let’s — yeah, for now we can both say the V word even though it’s cringe as hell.
Greg: All right. Let’s get into it.
Step 1: Finding the Right Idea with IdeaBrowser [00:01:30]
Sam: All right, so what’s the first step?
Greg: The first step: I’m assuming you don’t have an idea, so let’s go find one. Every single day, IdeaBrowser.com — this is actually something I created for myself. Basically, it uses AI and every single day a new idea comes with a trend. So today’s idea is to create an AI SEO agency. It gives you a name — LLM Boost — and it basically says there are 400 million people querying ChatGPT and similar tools. Someone should start an agency specializing in LLM large language model SEO.
Sam: Did your methodology give you this idea for IdeaBrowser.com?
Greg: 100%. I run a holding company as my day job and we’re constantly incubating and investing in ideas. We basically said: how can we have an unfair advantage using AI to find the latest trends and ideas?
Sam: All right, I’m into this. But there’s a twist.
Greg: So it says we’re going to offer a free AI-powered audit quiz that instantly shows businesses where they rank in AI searches. Most will be shocked they don’t exist. And then we’re going to sell premium optimization services to fix it. So for example, it’s going to ask — like Hampton — I don’t know how much of your traffic is organic SEO. Let’s say a thousand people a day from search. What people are noticing is that LLM search as a part of organic search is maybe now 5 or 10%.
Sam: Dude, we’ve gotten a bunch of people who found us via ChatGPT. So it would be cool if Hampton showed up more often.
Greg: Exactly. And it gives you an opportunity score, a problem score, and all of this uses AI agents. You can go in depth — it tells you exactly what business model you should use, what your pricing should be, who your competing customers are, your go-to-market strategies, your target audience. It scrapes and goes through Facebook groups, YouTube channels, Reddit. It’s almost like your AI co-founder.
One of the cool features is: let’s say there’s a lot of ideas that are really good, but Sam, you might not be the best person to go after a particular idea. So you can go through a founder fit score and say, “I’m Sam. I’m the founder of Hampton. I specialize in community. Is this a good idea for me?” And then it uses AI to generate an assessment of whether you should actually pursue it.
Sam: Okay, this is awesome. And this is a product, too? I just signed up for it. Do you have a lot of customers?
Greg: I haven’t publicly posted about it.
Sam: Wow. All right. So it gave me a six and a half out of 10. It says my skill alignment is four out of 10.
Greg: Kind of savage.
Sam: I actually don’t disagree with that. So it goes through this and gives you some immediate actions — maybe partner with an AI and SEO expert, maybe launch a community-driven platform for client engagement. The point is: this is awesome.
Greg: Yeah, and there’s way more you can do. It tells you exactly what your offer should be. It gives you different frameworks like Alex Hormozi’s value equation. And it has an AI assistant — you can ask it what are the key risks of the business and have a full conversation. It’s like ChatGPT for ideas.
There is a feature — we can go to it at the end if we have time — where you upload your own idea and it generates a report based on all the data from YouTube and Reddit and AI. But step one is: use this website to get an idea based on a trend. You want to build an idea based on a trend because it’s easier. So step one: use something like IdeaBrowser.com to get an idea based on a trend.
Why Ideas Actually Matter [00:09:00]
Sam: Okay. So I think ideas are important, but I also think that for a lot of people just starting out, ideas are unimportant — just get into something and you’ll figure it out. But I think if you have a proven track record of execution, if moving forward is not a problem for you, then ideas are actually incredibly important.
I talked to Kevin Ryan — I believe he’s a billionaire. He founded MongoDB, which is a $35 or $50 billion company, Business Insider, Gilt Group, Zola. He started all these amazing companies through his incubator. And he told me that ideas are incredibly important. He said, “I get one good idea a year and I want to make sure it’s great, because I go hard on that idea and I have to make sure I’m going in the right direction.” I’ve grown to believe that ideas really matter — it’s not just execution. If you have a past of moving forward, I think you’re right.
Greg: So that’s step one.
Step 2: Sketching the Idea with TLDraw [00:11:00]
Greg: Step two — I kind of cheated in the sense that you do need a human being involved in this process. It’s not 100% AI. So step two is you’ve got to sketch out the idea.
I use a tool called TLDraw. Have you seen this?
Sam: No. What is this? I’m going to it as you’re talking.
Greg: I wanted to give away the stack that I use. It’s kind of like a FigJam competitor. I think it’s free to sign up. I just wanted to sketch out what this quiz would look like. Because, if you remember, the idea is: people land on this website. IdeaBrowser gave us the name LLM Boost. We need to learn about the business — like Hampton, what’s the URL, what type of customers do you want. Then we need to do research with agents and check: is Hampton coming up in ChatGPT? Is Hampton coming up in Claude? Is Hampton coming up in Grok? And then we need to give it a score.
So I just drew this out. And the reason I drew it out is because I’ve noticed — this is a tip for everyone listening — when you go to an LLM and give it an image like that, you’re going to get better results.
Sam: That’s crazy. How long did it take you to draw that out?
Greg: Like seven minutes.
Sam: I hate drawing on computers.
Greg: The beauty about TLDraw is it makes it easy for bad drawers like you and me to just make these boxes. It’s for us.
Sam: FigJam is also really good.
Greg: It’s tldraw.com. Exactly.
Step 3: Scoping the MVP with Manus [00:13:30]
Greg: The next step is to scope out our minimal viable product. We need to figure out how we’re actually going to build this thing. There are so many question marks, and instead of going to a product manager or trying to figure it out yourself — you and I are lazy — we’re just going to get AI to do the whole thing for us.
Sam, have you ever heard of Manus?
Sam: No. I’m on their website right now. It says, “Manus is a general AI agent that bridges minds and actions. It doesn’t think, it delivers results.”
Greg: All right, that sounds great. So it’s almost like a ChatGPT supercharged. It literally goes and surfs the internet for you, learns stuff based on that, and then executes on the task. It’s like having 100 agents working for you.
Sam: Doesn’t OpenAI Research do this too?
Greg: Yeah. But I can watch it happen in real time and give it feedback. Sorry — OpenAI has Operator too.
Sam: Operator. Yeah.
Greg: This is like a supercharged version of Operator.
Sam: You think this is better than Operator?
Greg: I have to give a disclosure: it’s Chinese, so be careful. I mean, I’m not putting my financial data and uploading it to…
Sam: So do they like put a little PS — “capitalism is horrible, we’re going to dominate you eventually”?
Greg: There’s a subtle nuance to that in the vibe, for sure. But I will say it’s extremely good for this use case. I’ll go through the prompts and by the end of this section we’re going to have a good idea of what we’re building and all the specs.
Sam: I just signed up for it as we were talking. It has all these cool toggles — research, data analysis. Does this cost money or is it free?
Greg: Free to use initially. They give you a lot — they get you hooked. They’re like drug dealers. They get you hooked and then you have to buy credits.
Sam: All right. I have an account now.
Greg: And by the way, this was invite-only up until recently. So for people listening to this: get on this now before everyone finds out.
Sam: I looked at their traffic. It looks like they went live in March because they had zero traffic and then in March they had 23 million site visits.
Greg: That’s insane. Yeah.
Using Whisper Flow and Manus Together [00:17:00]
Greg: Have you heard of Whisper Flow?
Sam: I love Whisper Flow. It’s my favorite AI tool.
Greg: Same. You can use it on phone or desktop. It’s a button on my computer, and anywhere I would normally be typing, it transcribes what I’m saying. So I’m just talking all day instead of typing.
Sam: That’s what I do too.
Greg: So here’s what I did. I took the idea from IdeaBrowser, uploaded the image, and basically just explained what I’m doing — I’m starting an agency for LLM SEO. The key here is when you’re doing the initial prompt, don’t forget to say: “Ask me any questions before you get started, so we have the right strategy for this.” I’ve noticed that small one-sentence addition gets you significantly better output.
Sam: So you were talking to it, got the text, and also attached the image.
Greg: Yeah. You should attach any images or documents that you think are relevant to whatever you’re trying to do. And then the reply was: “Here are a bunch of questions I have before we get started.”
Sam: Yeah?
Greg: It asked the right questions — who is the target audience for the quiz? What is the main goal for the user? What are the key differentiators between traditional SEO and LLM SEO? I answered using Whisper Flow, so it’s really casual. I said things like, “The ultimate goal is to give them a benchmark for where they’re ranking in LLMs, and then we provide a service for helping them level up their LLM SEO.”
Sam: Would you also ask it to critique the idea?
Greg: Oh yeah, I do that all the time. Sometimes I’ll take an idea from IdeaBrowser, put it in, ask it to critique it, and through the conversation I’m like, “You know what, I don’t want to do this anymore.” And it’s better than ChatGPT for that, I think.
Sam: Because I do the same thing with OpenAI, but with my own company. I’ll upload my financials, upload a book I like — say a Warren Buffett book — and say, “What would Warren say about this? How would Warren solve these problems?”
Greg: That’s actually a really good hack for using any of these LLMs. Pretend you’re XYZ person you look up to. Would Warren Buffett start this business, why or why not? Would Shaan Puri start this business, why or why not?
Sam: Or I’ll say, “Pretend you’re a BCG or McKinsey consultant — cold-hearted, all about operations. Explain how that personality type would execute on this problem.”
Greg: That’s a good hack. People should definitely do that.
Manus Builds a Full Project Plan [00:21:00]
Greg: So Manus is awesome. But for what I’m doing right now, this is night and day better than ChatGPT for this. It’s actually night and day.
Once we clarify some stuff, it literally creates a to-do list for the project. This is what a project manager would do. Phase one: research and planning, clarify quiz objectives. Phase two: question and prompt development, create detailed quiz questions, draft specific, clear, actionable questions for each category. Phase three: validation and finalization, validate questions and prompts with users. Then delivery: report and send quiz materials to the user.
Sam: Wow. This is insane. I feel like all of my employees need to know how to do all this.
Greg: Send this to everyone. It’s just going to make everyone a lot more productive. So we got the project plan, and then it goes and gets to work. It’s your AI product manager. And it actually creates the two things we need — because remember, we’re trying to create essentially a SaaS product to score how a company like Hampton comes up in LLMs. We need two things: the questions we need to ask the business, and how do we test on the different LLMs. I don’t know how to do that on my own. So we asked Manus to figure it out for us, and it did.
Sam: Holy sh—. So what you’re looking at, Sam, is the quiz detailed questions that we can use as almost like a Typeform on our product. What’s the name of our product?
Greg: It’s called LLM Boost. So you’d go to llmboost.com and right on the first page, you see a quiz, and you start taking it. And I’m going to actually show you how I vibe coded it in like four minutes.
Sam: How’s your heart rate right now?
Greg: I feel like I’m taking notes and I’m like, “I need everyone at my company to know exactly how to do this.” Do I hire a Greg? Do I just have to get good at this and teach everyone? Is my job as the boss just teaching people how to use AI? Is that it?
Greg: I don’t think this is something you can outsource.
Sam: So I need to get good at this and then teach people.
Greg: Unfortunately, yeah. The unfair advantage and the returns you’re going to get by understanding these tools is just worth it. Why would you want to outsource it?
Sam: Well, because I’m not an expert in it, but I guess I have to become one. How do you stay on top of all this?
Greg: It’s my job, man. It’s my job to stay on top of this stuff.
Sam: But tell me how. Like, how did you know Manus is awesome?
Greg: Because I’m a nerd. I’m a nerd.
Sam: That’s such a cop-out. Are you on Twitter all day? What are you doing?
Greg: I do have tweet notifications for some creators I follow. I love playing with the tools, not just reading about them. But honestly, you’re hearing this through word of mouth — the trades. You’re reading the trades. It’s not PC Weekly anymore. It’s a guy on Twitter.
And I would say what I like about podcasting is that it’s an opportunity to learn in public. I’m just learning in public on the podcast.
Refining the Spec: Shortening the Quiz [00:26:00]
Greg: So we got these two things — basically the quiz for the business and the prompt testing for the LLMs. But when I went through it, I felt the quiz was too long. No one’s going to answer a thousand questions. So I basically said, “Is there any way to make the quiz shorter?”
And I also downloaded an LLM SEO mini-course from vibemarketer.com. It’s a course — a co-founder of mine started it, actually. I paid for it — well, I got it for free as the co-founder — but the point is, you can paste anything into Manus for context: PDFs, documents, anything. And Manus said, “My plan is to analyze the transcript you provided to extract key insights relevant to LLM SEO and how businesses are found in LLMs. Then I’ll work on strategies to shorten and consolidate the quiz.”
It’s so pleasant dealing with an employee like this.
Sam: Yeah. Compare that to “Hey, my grandma died. I’ve got to go to the funeral in Tampa. I’ll be back in 2 weeks.”
Greg: Manus takes no vacations.
Sam: Okay, cool. So it is a great employee.
The Biggest Hack: Ask It for the Prompt [00:28:30]
Greg: So it goes and does it for us. And I want to give away one more tip on Manus — actually, it works on any LLM. We’re lazy. So I know what I want to do next: go vibe code. Build the product. But I need to know what the best prompt is. To prompt Bolt.new.
And the best thing you can do is ask Manus or ChatGPT: “What is the best prompt?” So I said: “Can you create a prompt that I can give my AI developer to generate this landing page with a multi-step funnel that asks the LLM questions, includes all the necessary fields I need, and has a clean, modern design?”
By the way, this is another huge hack: ask it for the prompt you should use. You will never be able to out-prompt the software that has seen all the prompts.
Sam: I use Kubr — do you know Kubr?
Greg: I love Kubr. I know the founder. It’s a net worth tracker — like Mint, you can track all your finances. They have an AI chat integration. And I was like, “What questions do you think I should ask you on happiness, life strategy, whatever?” And it gave me a list of all the questions I should ask it that I never even thought of. Asking the LLM what you should ask it is shockingly useful.
Greg: Amen. So we asked it, and it does it beautifully. Here’s the prompt it generated. It goes through the project goal: “Develop a high-converting landing page and integrate a multi-step quiz funnel.” And this is the prompt you’re going to give to Bolt.new, which is the thing that will make the website.
Sam: Wow. So it’s a really long prompt.
Greg: Like a term paper. Obviously you and I would have never written this on our own. That would have taken a week.
Sam: Look at this.
Greg: Thank you, Manus. And thank you, People’s Republic of China. Now we move on to a Silicon Valley startup called Bolt.new.
Step 4: Vibe Coding the Website with Bolt.new [00:33:00]
Greg: So Bolt.new is the same category as Cursor, Lovable, and all the others. There’s Bolt, Lovable, Cursor, and Windsurf — which I think just got acquired for $3 billion.
Sam: Yeah, I saw that.
Greg: Cursor and Windsurf are for more technical people. Lovable and Bolt are for non-technical people who want to ship software.
Sam: Why do you prefer Bolt over Lovable?
Greg: I started using it first. I find the output to be really good. But use whatever works for you.
Sam: Okay. So you copied that huge prompt in there. And then?
Greg: I get this web page. And by the way, I’m not creating copy here — I literally one-shotted it. It says: “Your customers are searching for you on LLMs, but you’re not there.” Sad face. “Our free quiz helps you understand your visibility in ChatGPT AI models and what to do about it. Take the free LLM SEO quiz now.”
Sam: Oh my god. Okay. That’s not bad.
Greg: More than not bad.
Sam: “Your customers.” Yeah, I mean, it’s the best. You’re the words guy. But it’s a close second.
Greg: And there’s even a testimonial from a “CEO of a major internet company.” And you want to know something? I would leave that in there.
Sam: Why?
Greg: I would just leave it there as: “We’re seeing a much higher conversion rate from prospective users coming from organic LLM traffic versus organic search — from the CEO of a major internet company.” I think a person has probably said that before. So it’s not wrong to keep it in there.
Sam: It’s like — define “major.”
Greg: A person has said that. Totally. I don’t know if they’ve said it about llmboost.com, but like a person has said that for sure.
Running the Live Demo: The Quiz in Action [00:37:00]
Greg: So it creates the quiz. While we were talking, I just said on Bolt.new, “Make a personality quiz website,” and I got a web page. I just clicked “start now” and it works.
Sam: The website works. Yeah. It’s absolutely crazy.
Greg: This is the thing where you can literally say English to it and software comes out. Which is bonkers. So it says: “What type of business do you want to operate?” Let’s say like a service business. “Do you have a website?” Yes. “How often do you publish new content?” Weekly or more often. “What topics does your content typically cover?” How-to guides — that’s relevant. Case studies.
Sam: Let’s click case studies.
Greg: “Before today, how familiar were you with LLM SEO?” Somewhat familiar. “Do you or your team use AI tools like ChatGPT for business?” All the time. “Are your competitors visible in AI search?” I am not sure. “How important is SEO for your business currently?” Somewhat important. “Company URL?” Hampton.com. “Name?” Sam Parr. “Company name?” Hampton.
Look at this. They put a little privacy notice: “We respect your privacy. We’ll only use your information to send you your quiz results and related LLM SEO information. We will never share your information with third parties.”
This is the kind of small detail that if you don’t use Manus or an LLM to write the prompt for you, you will miss. So now: “In 24 hours, Sam, you’re going to get a personalized LLM SEO score, an analysis of your current AI search visibility, specific recommendations for improvement, and additional resources to help you optimize your business for LLM visibility.”
Sam: We just built a SaaS in like 30 minutes.
Greg: Well — it’s not doing the work automatically yet.
Sam: Right, it’s not going to actually do a personalized LLM SEO score. Who’s going to do that?
Greg: The software. All I have to do — Bolt has integrations — is get a ChatGPT API key and hook it up. And I can give you your score.
Sam: How hard will that be?
Greg: Under an hour. You’d probably want to integrate with Supabase. Do you know Supabase?
Sam: No, dude.
Greg: Supabase is just a database. So this is just the front end right now. In simpler English: it’s just the thing you’re seeing, but you need to hook it up to a database — a place to store the data. And if you want to take payments, you’ll want to add Stripe as well.
Sam: What?
Greg: I know. And the way I’d create your report is: back in Manus, it actually tells you exactly what to prompt ChatGPT and Perplexity — we just need API keys because it costs money to use the intelligence of those LLMs.
For example, section one: brand and company info, prompts to check if an LLM can accurately retrieve basic information about the company and its offerings. So it would take your quiz information — “company name: Hampton” — and just put it into all these questions. Agents would then ask ChatGPT, “What can you tell me about Hampton?” The output would go to Supabase. Then ChatGPT would say, “Based on that, Hampton performed well / not well — here’s a score of 70 out of 100.”
Sam: So Supabase is what stores the data. OpenAI is what crunches the numbers. Manus gave us the questions. And ChatGPT analyzes how well you rank and what you need to do to rank better.
Greg: That’s the business model — you send people the report and say, “You scored really well here, not well here. We can help you improve your LLM SEO. For $2,000 a month, here’s a package. For $5,000 a month, here’s another.”
Sam: But if I bought the $2,000 package, what would you actually be doing?
Greg: Here’s the thing no one says: 80 to 90% of good LLM SEO is just good regular SEO. Getting backlinks. And another thing worth doing: in a world where you have tools like Replit, Bolt, you know — creating calculators and software. That’s a high-quality signal for a lot of these LLMs too.
Step 5: Vibe Marketing with Lindy AI [00:44:00]
Greg: All right. The next piece is going to blow your mind even more. You might be thinking: “Okay, cool, Greg. You built something and it’s a prototype. But the hardest part about getting a business to a million dollars a year in revenue is getting customers. How the hell do you get customers?”
Well, let’s do what I call vibe marketing. I’m using a tool called Lindy. lindy.ai. We’re going to go through two to three Lindy workflows that you can copy, that can help you get customers on autopilot.
I want to be clear: I’m not showing this so you have to use Manus, Bolt, IdeaBrowser, or Lindy. I’m showing this so people can think: “How can I use a tool like this in my own business?” That’s the mindset.
This is a flow I use for our design agency LCA — which does design for AI interfaces and AI agents — that has literally made us millions of dollars. It could be used for LLM Boost too.
The way it works: I take my tweets, I create content on Twitter, I post it to LinkedIn — literally just copy and paste. Then if I have a post on LinkedIn, it looks at who comments and likes it. It puts all that data into a database and does data enrichment. So it says: “Sam Parr, he lives in New York, Hampton, 5,000 followers.” It decides if the lead is qualified. It LLM-scores the prospect zero to five based on our criteria. If the lead is qualified, we get their email and phone number from Prospeo. It updates a record, and then we get a Slack notification that says “Sam Parr, potential customer, just commented.”
For our business, our average deal size is about a million dollars. So we’re only looking for, like — look at our list of customers. It’s the GM of Nike. It’s the president of Dropbox.
Sam: And you make content that they engage with, and then you get notified.
Greg: Oh my god. I almost didn’t want to share this.
Sam: What video titles do you have that are all about marketing?
Greg: Things like this. All my vibe marketing stuff. I just did a whole presentation on vibe marketing that goes through a bunch of workflows. Watch that. That’s a good primer.
Sam: You’re making me weak at the knees, Greg. And I know how much money I have to spend doing this normally. Like, before this, I literally had a person combing through my LinkedIn likes.
Lindy Workflow 2: The AI Email Negotiator [00:48:30]
Greg: Okay. Check this out. So we notify the potential prospects, and then what happens is: we have a salesperson. If the salesperson “hearts” the message in Slack, it automatically sends that person a text message or email.
Sam: Wait — “hearts”? What does heart mean?
Greg: Like a heart on Slack. The love symbol.
Sam: Oh, okay. So I post something on LinkedIn. The GM of Salesforce who I want to sell to clicks it and says, “You rock, Sam.” I get notified in Slack. Someone on my team clicks the heart icon. And then what?
Greg: Then that person gets a personalized email or text message.
Sam: Oh my god. Yeah. What’s the email say?
Greg: It’s personalized. Let’s use the LCA example: we just designed Dropbox Dash — it’s an AI version of Dropbox. And the VP of product at Shopify likes our post. So a personalized email goes out and says, “Hey Shopify, how can we transition Shopify from a cloud company to an AI company? I’d love to jam with you on it. Saw that you liked my post.”
Sam: That’s crazy.
Greg: But what’s even crazier is the next thing. What if you could have an email negotiator as an AI agent? Basically, it negotiates on behalf of you automatically using AI.
Let’s use LLM Boost as the example. We create a pricing page: three packages — $3,000 a month, $5,000 a month, and “contact us.” Someone clicks “contact us” and says, “Hey, I’ll pay upfront, but I want a discount.” The agent gets the email, then checks a knowledge base — could be Notion, Google Drive, Dropbox, a website. You do some upfront human work: “If someone contacts us and wants a discount, we won’t go lower than 10 or 15%.” If it finds a response scenario, it automatically responds to the inbound lead.
The prompt here, which you can see on the right-hand side, is: “Your job is to negotiate with the emailer and respond to their questions until a decision is made regarding the partnership opportunity.”
Lindy actually has a template section — you can copy many of these workflows and prompts. The template has things like: “Hey, appreciate the interest. We only discount 10% for fall deals — it is our busiest time for collabs. Let me know what you decide.” Keep responses to one to two sentences, never make the first offering.
Sam: This is insane. Do all of your employees know how to do this incredibly well?
Greg: We’re incubating AI products and have agencies for this stuff, so yes.
Sam: How many employees do you have?
Greg: Probably 55 or 60.
Sam: And if you didn’t have all this, how much bigger would your team need to be? How much are you saving?
Greg: We’re doing things at a speed and scale we wouldn’t be able to with human beings, realistically. There’s money we were leaving on the table — millions of dollars a year, especially on the LCA side with big partnership deals. It’s so important to reach out to someone 5, 10, 25 minutes after they engage, or else they might forget about you. I’d say we’re probably saving $5 million a year, plus.
Sam: God damn it. I wish I had a full-time staff member who just audited everything we do and said, “Let’s automate it.”
Greg: I don’t mean to plug my stuff, but you can go to boringmarketing.com — that’s what we do. Small team. A couple dozen clients. We spend most of our time building software and technology that automates this.
Sam: I just want you to come on every week and show me how to do everything.
Bonus: Claude Voice Calling and the MFM Hotline Idea [00:54:00]
Greg: Professor Greg. So, you can actually use — I call it “Clude.” People here call it “Claude.” In French you call it “Clo,” I think. But you can use Claude to call a phone number and have a full-on voice conversation.
Sam: I’ve been the recipient of those. I think those are horrible.
Greg: They’re 85% there. But it’s worth playing with because Lindy was at 60% six months ago and now it’s close. I wouldn’t use it to reach out to customers, but here’s what you should do, Sam: create a 1-800 number — like 1-800-JOIN-HAMPTON. People call it and you ask for feedback. You get feedback, someone says “I had an amazing experience,” it’s anonymous, it gets stored in Airtable. We summarize the call and post it to Slack, so we know every day what people are saying about our products and services.
Sam: We should do that for MFM.
Greg: You should totally do it. I can build it for you.
Sam: Yeah, let’s do it. I want an MFM hotline where people can call and say whatever, and we make segments out of it.
Greg: Easy peasy.
So those are three workflows worth considering. There are other tools besides Lindy — there’s Gumloop and there’s n8n.
Sam: Dude, these names are funky. Gumloop. Supabase. These are interesting.
Greg: I like things that are really easy. N8N is almost like Cursor and Windsurf for vibe marketing — a bit more technical. I like Lindy and Gumloop because they’re simpler for guys like us. You can go through their templates and workflows and just cut and duplicate if you don’t want to create the flows yourself.
Oh, and there are literally thousands of million-dollar-a-year business ideas just from taking Lindy’s and Gumloop’s workflows and selling them into the real world. Sell a medical scribe to, like, $299 a month.
Sam: Dude, I invested in a startup doing that.
Greg: I hope the customers don’t see that they can just use Lindy.
Sam: Dude, this is a good one — look at this template. “Elon Lindy calls your team to ask them what they got done this week.”
Greg: Oh my god. That’s insane.
Step 6: From Service to Software — Taking the Business Further [00:58:00]
Greg: Last thing, because I know we have to wrap up. I always get this comment on my podcast: “Okay, Greg, but that’s just a service business. You just showed us how to build service businesses.” Well, number one, it’s an AI-powered service business. And we also created SaaS. But okay, I hear you, commenter.
If we were actually building this business, how do we take it from being a service business to more of a tech / software business?
So one thing I always do: if you get to step five, you have a business doing hundreds of thousands of dollars a year, maybe millions, and you have ideas on where you take it. I’ll go back to IdeaBrowser and use the idea agent — upload an idea and see if it’s good.
So while going through Manus and building this, I was like: Ahrefs. You know Ahrefs?
Sam: Yeah, I love Ahrefs.
Greg: Do you know how much revenue they’re doing?
Sam: They’re big. Like $80 or $90 million — in the hundred million range? And they’re bootstrapped.
Greg: Incredible business. But there’s probably an opportunity to create an AI version of Ahrefs for LLM SEO. And there’s probably that niche. Semrush is publicly traded — they’re like a $450 million a year business. If you can get 1% of that, 5%, 10%…
So I went to IdeaBrowser, posted my idea — “Give me the breakdown.” It tells me exactly what my offer should be, my pricing, the value ladder. Alex Hormozi stuff, the value equation. Russell Brunson. If we were going to create an AI version of Ahrefs, what would that look like? We’d want to create an interactive SEO audit tool. We’ve already created that. We’d want a starter plan — $99 a month. It literally just tells you what to do.
Sam: But you’re skipping a big thing — you have to build the software. You’re telling me to create the monthly plan, but how do I create the thing that crawls the web and tells me how many backlinks I have?
Greg: That’s the old way of thinking.
Sam: What’s the new way?
Greg: You used to have to go and create it. If you think AI-Ahrefs is a good idea, you literally go back and repeat all the steps. Are you overselling this? Like, can Bolt.new actually make all the code that creates what Ahrefs does? There must be a reason Ahrefs has 300 developers on staff — or had, until recently.
The short answer: Bolt and Lovable are great places to build something simple and get your front end up. Once you’re ready to scale, you use tools like Cursor, Replit, Windsurf — more technical. Developers today are using those tools and they’re 10x developers. It’s just so much faster.
I’m not saying it’s easy. Building a startup is hard. It’s a roller coaster. There are going to be things you learn. But this is the framework for how to build it, and you can do it using tools like these.
Closing [01:02:00]
Sam: I’m hyped. I messaged Ari in the middle of this episode — I said, “Schedule Greg another time to come on, right now. My mind is blown.”
Did you message her, or did you do some Neuralink agent workflow you created in the past?
Sam: Brother, I am a Neanderthal. I’m not there yet. I literally messaged my team and said, “I have some mind-blowing stuff to show you. I’m calling you in 20 minutes.” And I took screenshots of the Lindy flows and told them we must do this immediately.
This was amazing.
Greg: Well, I’m happy. That was the goal — share some sauce, get you thinking, and hopefully not piss off your team too much. I think they’re going to come out the other side way more productive.
Sam: You’re the man, Greg. I call it just the Greg Eisenberg YouTube channel, but it’s the Startup Ideas channel, right?
Greg: The Startup Ideas show on the Greg Eisenberg YouTube channel. Exactly.
Sam: Thank you. You’re the best.
Greg: Thank you for having me. This was awesome.
Sam: That’s it. That’s the pod.