This video explains how to use personal AI agents such as Meta Muse to find and connect customers and providers in Facebook groups, Marketplace, and Instagram, then develop those connections into a highly profitable matching business. The key is to go beyond simply finding information: accumulate curated lists and trust-related data as assets you own, then expand into communities and software businesses. The presenter strongly believes this is an early opportunity that could create "thousands of millionaires." 🚀
1. A New Money-Making Opportunity Created by Personal AI Agents
The presenter opens by saying that anyone who wants to make money online should be exploring personal agents such as Muse right now. Few people are yet discussing exactly how to build multimillion-dollar businesses using personal AI agents, but he sees enormous opportunity in this market's information asymmetry and automation of execution.
"If you're looking for ways to make money on the internet and you're not playing around with personal agents like Muse, honestly, I don't really get it."
The opportunity he describes is straightforward. Buyers who need help and sellers who can provide services already exist all over the internet. They are simply scattered, and finding, vetting, and connecting them is cumbersome. Muse can search and organize on their behalf, while the business owner earns a matching fee on top.
He calls this "an easy time to make your first internet dollar, or your second." In particular, Facebook groups, Marketplace, and Instagram already contain vast numbers of users and posts. Instead of creating new demand, he emphasizes connecting existing supply and demand more efficiently.
The video also introduces its sponsor, Brex. He says its high-limit cards, banking capabilities, and AI features for back-office work such as expense reporting make it well suited to an agent-centered work environment.
2. A Business Model Discovered While Looking for a Night Nurse
The idea began when the presenter recently had a baby and was looking for a night nurse to care for a newborn overnight. He tried finding someone through acquaintances but struggled to find a suitable person, eventually consulting an agency. The agency openly explained that it found candidates by looking through Facebook groups in Miami, South Florida, New York, and other places.
"The agency told me they were finding night nurses in Facebook groups in Miami, South Florida, or New York. I thought, why should I pay that fee when Muse can do that work now?"
This led him to an important realization. Traditional agencies have people manually search groups, read posts, organize candidates, contact them, and coordinate schedules. If Muse can do much of that work, someone could build a lean agency offering a similar service at lower cost.
He does not limit the model to this particular example. He explains that defining a location + niche + vetting rules could turn each of the following into its own business:
- Night nurses in Tampa, Florida
- Used furniture in Austin, Texas
- Plumbers in Phoenix, Arizona
- Caregivers charging around CAD 30 per hour in Montreal, Canada
The business owner's first job is to decide which market to target: which niche, which city, whom to approve, and how to judge quality. Muse can then help with searching, organizing information, matching, drafting messages, texting and calling, and booking.
"Muse handles the structuring, matching, drafting DMs, calls and texts, and even bookings."
3. Own Curated Data Assets, Not Just Information
Some might see this business as simple "information arbitrage." But the presenter argues that the real value is not merely finding posts; it is turning that information into a clean, trustworthy database.
Facebook groups, Marketplace, and Instagram contain information of widely varying quality. Someone might be highly skilled but have a poor introduction, while another person might be inexpensive but show warning signs in their experience or reviews. Muse can be used to collect this public data and organize it in Airtable, Google Sheets, or similar tools.
What accumulates is more than a list of names. The data assets he describes include:
- Providers' and customers' names and contact details
- Services needed and services available
- Desired prices, hourly rates, and locations
- Trust signals such as testimonials, experience, and recommendations from parents
- Information that improves matching quality, such as response times, whether someone is active, and expertise
"You get a compounding piece of data: names, needs, prices, and trust signals. And ultimately, that asset is yours."
He mentions that the data itself could eventually be sold and says he will cover data brokering in a separate episode. The video's main point, however, is not selling data; it is that a curated list becomes more valuable over time. As vetted candidates and customer information accumulate, you can match new customers faster and improve service quality.
He does draw a boundary: there is no single correct way to organize the data. A customer seeking a night nurse cares about different trust factors than someone seeking a plumber. You must therefore first understand what your ICP, or ideal customer profile, values most.
"There's no one way of cleaning data that works for every niche. Ultimately, you have to start with, 'What does my ICP want?'"
4. Operating Muse Like a Real Agency
The presenter sees Muse not simply as a search tool, but as operating infrastructure for a small agency. Ask it, for example, to "find night nurses in Montreal at around CAD 30 an hour," and it can gather candidates by exploring public Facebook groups, Instagram, Marketplace, and other sources.
There are limits to public access, however. Muse can search public data, but to view information inside a private Facebook group, the user must first join that group. This also means taking platform rules and access permissions into account.
He explains that after receiving candidates, a person must check their quality, either directly or with AI assistance. Even if someone appears on the list, further review is needed to determine whether they have sufficient real experience, reviews and recommendations, and the qualifications the customer requires.
Connecting a dedicated phone number to Muse takes operational automation a step further. The presenter gives Bland AI's Agent Line as an example. It provides a US phone number capable of calls and texts, allowing Muse to contact people, receive responses, and move the matching process forward. He mentions a possible price of around $15 but adds that the figure is based on his recollection.
"You can prompt your Muse agent, 'Create a phone number and call me.' Now it can reach out to people and even close sales."
The model is attractive because it can substantially reduce a traditional agency's cost structure. Previously, several employees had to find, contact, vet, and book candidates. If AI handles much of that work, the business may be able to offer parents or other customers lower prices while retaining high margins.
"This might have cost thousands of dollars before because it took a team of five. But now the cost structure is much lower, so you can pass savings on to customers and still make enormous margins."
5. Facebook Groups and Niche Pages He Considers Undervalued
The presenter identifies Facebook groups as particularly undervalued assets in the Muse era. He estimates that Facebook groups have around 1.8 billion monthly active users and expects huge numbers of Muse agents to explore their public information in the future.
He therefore argues that rather than only mining other people's groups, you should start building your own now.
"If you believe millions of Muse agents will be roaming Facebook groups, why aren't you creating a Facebook group right now?"
For example, you could create a Facebook business page called "Miami Night Nurse Agency" or run a niche Instagram page specializing in a particular city and service. This makes you a source of public information that agents can discover, while also creating a branded channel that builds trust with people.
He mentions that buying Facebook groups might also be an arbitrage opportunity. However, he warns that this may violate Facebook's terms of service, so you must check the platform's terms first. He is not suggesting that you ignore rules simply because something can make money.
"I'm not sure whether buying groups is okay under the terms. So make sure you check Facebook's terms."
He also sees Instagram pages as undervalued in a new sense. That may sound odd given Instagram's already enormous size, but his reasoning is that trusted accounts in specific niches could become more strategic assets as AI agents explore and use posts, comments, bios, and stories far faster.
6. Prompts and Platform Strategies for Better Search Results
The presenter emphasizes that Muse will not automatically take every optimal action. You get better results by specifying where to look, what to find, and what criteria to use.
In the initial demonstration, Muse explored public listing sites and Instagram, but the presenter explicitly asked it to search Facebook groups to find higher-quality candidates.
"Find 20 night nurses in Facebook groups. That's where the highest-quality data is."
Muse then found posts in relevant groups such as "Moms Looking for Babysitters in Montreal," "Montreal Babysitter Jobs," "West Island Community," and "Filipino Caregivers," prioritizing candidates recommended by parents. The presenter considered the results high-quality data suitable for a real matching business.
He does not expect Meta to configure its agent to aggressively cannibalize its own platforms by default. To have it actively explore Marketplace or Instagram, users must give deliberate instructions and design a strategy.
On Marketplace, he particularly highlights opportunities in used goods. Because all the information is often public, you can look for conditions such as:
- Mispriced items or items with poor descriptions
- Items discounted after remaining unsold for a long time
- Relisted items or those with more room for negotiation
- Items with fewer potential buyers because they are "pickup only"
- Opportunities to bundle geographically scattered items for resale
He corrects himself by noting that a heavy piano is not a good example, but says it could work to find a cheap, pickup-only item such as a Mac mini and resell it. You could have Muse track price reductions or relisting dates and even negotiate prices on your behalf.
On Instagram, comments, bios, and even stories can be sources of information. However, Muse will not always independently decide, "I should look here," so specific prompts are needed to identify which accounts, formats, keywords, and behavioral signals to seek.
"Muse won't come up with all these ideas on its own. You have to tell it where to look and what to look for."
The presenter himself is more interested in service matching than used-goods trading. He sees Marketplace as stronger for goods, and Facebook groups as the primary place where service demand and providers show up.
7. Four Stages for Growing a Small Matching Side Hustle into a Large Company
The video's central structure is the presenter's four-stage framework for personal-agent arbitrage. He explains that even if you begin as a small side hustle, you can own the data, establish direct customer touchpoints, and eventually develop into a technology company.
Stage 1: Mine Other People's Platforms
The first stage is finding opportunities in Facebook groups run by others, Marketplace, and Instagram. Here, you use AI to identify supply and demand, connect people who need one another, and learn how the market works.
"The first stage is mining. You're finding gold in other people's groups, Marketplace, and Instagram."
The purpose is not just to make money quickly. It is also a process of gaining market knowledge: which requests recur, what customers worry about, and what problems providers face.
Stage 2: Own a Curated List
The second stage is taking data off-platform and organizing it. Once you have a list that incorporates prices, quality, reviews, and trustworthiness, you no longer need to search public posts from scratch every time.
"When you take it off-platform, clean it up, price it, and build a trustworthy list, that's where the business gains value."
This is the stage of assets you own that he repeatedly emphasizes. The goal should be to accumulate your own data and customer relationships rather than depend entirely on a platform's algorithms or policies.
Stage 3: Own the Space and Community
The third stage is having your own "room" where customers and providers gather. This is not a physical space, but a channel where people congregate and trust develops, such as a Facebook group, niche Instagram account, or Marketplace storefront.
"Third, own the room. Own the Facebook group, own the niche Instagram, and have a Marketplace storefront."
At this stage, you shift from someone finding opportunities on outside platforms to a trusted operator whom people in a particular market seek out first.
Stage 4: Build Vertical Technology
The final stage is building technology based on the data, customers, and operational knowledge acquired along the way. This could be SaaS, an industry-specific AI agent, or a vertical operating system that automates an industry's work from start to finish.
"Once you have clean data and customers and understand the workflow, you start thinking, 'What's the startup that could automate this end to end?'"
The presenter stresses that this is not merely a story about a side hustle earning $1,000–$2,000 a month. He argues that deeply understanding a particular market, standardizing recurring tasks, and building technology to automate them can create much greater enterprise value.
8. Content Operations and the Personal-Agent Competitive Landscape
Muse is useful beyond matching and organizing data: it can also research which content and formats work on Instagram and generate content ideas. The presenter says Muse can serve as your content hub.
If you run a niche account, for example, you can ask Muse to research related accounts, popular post formats, comment reactions, and potential customers' interests. You can then create content, grow a community, and ultimately earn customers' trust and attention.
He summarizes the value of a service business this way:
"Muse listens to the need; Instagram finds the people and proves they're good. It watches life's moments and lets you build your own audience. Then you own trust and attention."
He also notes that Muse is not the only option. There are several personal-agent platforms, including Dots and Instinct, each with strengths and weaknesses. He says Dots feels intelligent at the model level, but currently sees Muse as particularly strong at mining Facebook groups, Instagram, and Marketplace.
"Where Muse really stands out is its ability to mine Facebook groups, Instagram, and Marketplace."
He recommends starting with Muse's free plan and upgrading once you have real customers and revenue. He uses the $100-per-month plan himself and understands it to offer a better model and more tokens. However, exact plan details may change, so you should check the actual service page.
9. Conclusion: Start Small Now, Then Expand Deeply
The presenter says he cannot be certain how long this opportunity will last. But he sees this as a very early moment when AI agents can quickly find supply and demand across the public internet, while people build vetting, trust, relationships, and brands on top.
The core path he proposes is clear: first connect supply and demand within a narrow niche, accumulate curated data in the process, build your own community and customer touchpoints, and finally create technology that automates the work.
"This is the best time in history to build something."
Ultimately, the video goes beyond recommending Muse itself. It is closer to a business strategy: use AI to solve people's real problems faster and more cheaply, then turn the resulting data and trust into assets you own. The presenter's conclusion is that even if you start in a small local niche, a well-built foundation can grow into a matching service, a community, and eventually a vertical software company.