The argument in one line.
Compressing time-consuming tasks from 30 minutes to 2 minutes through AI agents and structured prompts is how solo operators build million-dollar revenue-per-employee businesses before competitors catch up.
Read if. Skip if.
READ IF YOU AREβ¦
- A solo founder or small business owner running a SaaS, agency, or service business who handles multiple operational roles and wants to understand which AI agents can automate support, sales, or admin work.
- An entrepreneur with an existing audience or customer base who's heard AI hype but doesn't know how to actually deploy agents into their tech stack and needs concrete examples of ROI.
- A content creator or business owner who's been using AI as a search engine replacement and wants to learn the framework difference between prompt-and-pray versus building systems that work while you sleep.
SKIP IFβ¦
- You're a large enterprise with established AI ops and procurement processes β this focuses on solo founder implementation and doesn't address compliance, governance, or team handoff at scale.
- You're a non-technical founder who needs step-by-step technical setup instructions β this teaches strategy and framework but assumes comfort with API integrations, CRM connections, and prompt design.
The full version, fast.
Treating AI like a search engine is the losing move; treating it like a stack of context, constraints, and tools is how a one-person business now competes with companies of thousands. The mechanism is a four-step staircase οΏ½ prompts, projects with persistent context, tool and MCP connections, then proactive scheduled agents οΏ½ combined with the TCCA prompt framework (task, context, constraints, ask clarifying questions) and a rule to decompose roles into discrete workflows before automating. The actionable conclusion is to audit your day, find any task that takes thirty minutes and compress it to two with AI, connect your real systems like Gmail and support tools, and bet on yourself by stacking these wins until revenue per employee reaches the millions AI-native operators now hit.
Free for members
Chat with this breakdown β free.
Sign in and you get 23 free chat messages on us β 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 βWhere the time goes.
00:00 β 01:18
01 Β· Intro: AI Stack vs AI Toy
Hook with credibility proof. AI as toy vs. AI as a stack that replaces 10 employees.
01:18 β 03:26
02 Β· Takeaway 1: The Time Is Now
Stick figure diagram of person running up the hill vs. waiting. Sabrina 0-to-2M solo as proof.
03:26 β 04:09
03 Β· Try This: Audit How AI Threatens Your Income
Prompt: list 3 ways AI puts your income at risk plus 3 things to learn now.
04:09 β 06:54
04 Β· Takeaway 2: Build an AI-Native Business
AI-native means structuring roles and workflows around AI from day one. Speed is the only real startup advantage.
06:54 β 08:30
05 Β· Inside Anthropic Solo Growth Marketing
One person ran all of Anthropic growth marketing with AI agents: ad creative, AB testing, deployment loop.
08:30 β 08:34
06 Β· AI-Native Startups Hitting 100M ARR
Cursor, Lovable, Higgsfield: 5+ years compressed to under 18 months to 100M ARR.
08:34 β 11:13
07 Β· Takeaway 3: The Staircase of AI Leverage
4-step staircase: Prompts > Context/Projects > Tools/MCP > Proactive agents. Most users at step 1-2.
11:13 β 14:58
08 Β· The Revenue Per Employee Shift
Buffer ~250k/employee is old gold standard. AI-native companies do millions per employee.
14:58 β 18:13
09 Β· How My Customer Support Agent Works
3-agent system: Primary (70% ticket resolution at 96% confidence), Cleanup, Adversarial double-checker.
18:13 β 19:23
10 Β· Try This: Connect Gmail to Claude
Claude > Connectors > Gmail > summarize unread emails > schedule daily brief. That is level 3-4.
19:23 β 23:54
11 Β· Takeaway 4: Think Workflows, Not Roles
Break a role into discrete tasks (read, search, draft, reply, escalate), ask AI which it can handle today.
23:54 β 27:18
12 Β· Takeaway 5: The TCCA Prompt Stack
Task, Context, Constraint, Ask clarifying questions. The A lets you skip the rest.
27:18 β 29:48
13 Β· Takeaway 6: Bet on Yourself
Learn AI, build income streams, repeat. Gains at big companies do not trickle down.
29:48 β 33:22
14 Β· Takeaway 7: Run a Daily Task Audit
Write down daily tasks. Find what takes 30 min. Use AI to compress to 2 min. Nano Banana example.
33:22 β 34:36
15 Β· Wrap Up + CTA
Claude CoWork tutorial plug. Subscribe CTA. Final board shows all 7 takeaways.
Lines worth screenshotting.
- 70% of customer support tickets handled automatically by three AI agents β triage, responder, escalator β is the number that converts 'AI for customer support' from a vague idea into a measurable operational outcome.
- The TCCA prompt framework β Task, Context, Constraints, Action β produces structurally complete prompts that give AI agents everything they need to execute without follow-up clarification.
- Revenue per employee is the benchmark that reveals when AI is actually changing the economics of a business: a solo operator generating $2M+ with zero full-time employees is a data point that traditional headcount planning cannot explain.
- The 30-to-2-minute compression audit β listing every task that takes 30 minutes or more and asking which ones AI could do in 2 minutes β is the prioritization method that identifies the highest-leverage automation targets in any business.
- Using AI like a search engine (ask, read, close) is categorically different from using AI like a stack (context, constraints, tools) β the first produces information, the second produces shipped work.
- AI agents connecting to email, calendar, CRM, support systems, and social accounts are not productivity features β they are business infrastructure that runs while the operator sleeps, which is the property that makes solo businesses economically competitive with teams.
- The window to learn this before competitors catch up is real: the advantage of early adoption in an exponential technology is not permanent, but the compounding knowledge lead from starting now persists even after the technology becomes mainstream.
- Three support agents with distinct roles β triage, responder, escalator β mirror a human support team's org structure, which is the design pattern that makes multi-agent systems legible to business owners who understand team dynamics.
- A solo SaaS founder with thousands of paying users and no team is the business model that AI infrastructure makes viable β the economics require AI doing the work of 10 employees, not just assisting one.
- Hormozi's AI playbook validated by a solo operator's own revenue and support data converts abstract advice into proof of concept: the frameworks work at the scale where they are being taught.
- Context given to an AI agent is what determines the quality of the output: more relevant context produces better work, which is why the TCCA framework starts with context as the second element rather than an afterthought.
- The compression audit is not about finding tasks to eliminate β it is about finding tasks where AI can do the same work in a fraction of the time, which frees the human for the tasks where time compression is not possible.
- An AI stack that does the work of 10 full-time employees is not a metaphor β it is an operational claim that can be verified by comparing output volume, quality, and cost against what 10 employees would have produced.
- Customer support automation at 70% auto-resolution means the human support capacity goes entirely to the 30% of tickets that require judgment, empathy, or technical depth β which is the distribution of work that maximizes both human contribution and customer experience.
- AI agents that are real β connecting to actual systems, taking actual actions, producing actual output β are the 2026 baseline that separates builders who understand what AI can do from users who are still treating it as a smarter search engine.
Steal the TCCA stack and the 3-agent model.
Joe's operator playbook
The TCCA prompt framework and the adversarial-agent support stack are both ready to teach and ready to ship inside JoeFlow or MCN+ today.
- Teach TCCA in a short-form: Task, Context, Constraint, Ask clarifying questions. Four words, one result.
- The A bullet alone is a hook: tell people they can skip the other three and just ask AI to interview them.
- The 3-agent support stack (Primary / Cleanup / Adversarial) is a concrete JoeFlow or MCN+ feature demo.
- The 4-step staircase is a positioning ladder: show where most people are stuck (step 1-2) and what unlocks at step 3-4.
- The 30-to-2 compression audit is a weekly newsletter section or recurring short-form series.
- The revenue-per-employee benchmark (250k Buffer vs. millions AI-native) is the single best argument for building solo-first.
- Sabrina's reaction-commentary format: take a viral video, add your own receipts, assign an action per point. Replicable content engine.
Terms worth knowing.
- AI stack
- A curated combination of AI tools, agents, and automation workflows configured to work together as a system that handles business tasks end-to-end.
- TCCA framework
- Sabrina Ramonov's prompt engineering method: Task, Context, Constraints, and Action β a four-part structure for writing AI prompts that produce reliable, work-ready outputs.
- AI agent
- An AI system that can take actions autonomously β browsing the web, sending emails, updating a CRM, or running code β without requiring a human to approve each step.
- SaaS
- Software as a Service β a business model where customers pay a recurring subscription to access software hosted in the cloud rather than installing it locally.
- context window
- The maximum amount of text an AI model can process in a single session, which limits how much background information, instructions, and conversation history it can use at once.
- automation audit
- A systematic review of business workflows to identify tasks that are repetitive and rule-based enough to be delegated to AI tools or agents.
- revenue per employee
- A business efficiency metric calculated by dividing total revenue by headcount, used to benchmark how much output each person (or AI) generates relative to cost.
- customer support automation
- Using AI agents to automatically handle, triage, and resolve customer support tickets without human involvement, typically deflecting a percentage of total ticket volume.
- solo operator
- A business owner who runs a profitable company without a team by using AI tools and automation to handle work that would traditionally require employees.
Things they pointed at.
00:26channelAlex Hormozi - How to Win With AI in 2026
11:49productHiggsfield
31:03productNano Banana
33:34linkClaude CoWork - 5 insane use cases tutorial
Lines you could clip.
00:00
βIf you're a business owner still using AI like a search engine, you're already losing.β
00:56
βThe difference between ChatGPT yapping uselessly versus a coworker that's actually shipping work while you sleep.β
12:37
βI have never seen a SaaS business that hit 10k per month that couldn't be scaled to 100k per month.β
13:58
βYou are just an item in a spreadsheet.β
27:07
βThe only metric I actually look at is: is AI helping me scale my business?β
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken β crank it to 2Γ and still catch all of it. The same dual-channel trick behind Amazon's KindleΒ +Β Audible.
metaphoranalogystory
00:00If you're a business owner still using AI like a search engine, you're already losing. I went from zero to 2,000,000 followers solo. I built a highly profitable SaaS app with thousands of paying users solo.
00:12No team, no agency, just me, plus an AI stack that does the work of 10 full time employees. You do not need a tech degree, and you do not need anyone's permission to learn this stuff and make money for yourself.
00:25Alex Hormozi dropped his recent video, how to win with AI in 2026. And today, I'm gonna walk through every single key point he makes with real examples from my own work plus insights on why these approaches work.
00:382026 is the year that AI agents got real. They can now connect to your email, your calendar, your CRM, your support system, your social media accounts, your customers.
00:49The window is right now to learn this stuff before competitors catch up. Now most people teach AI like a toy.
00:55I teach it like a stack, context, constraints, and tools. Understanding that is the difference between versus .
01:08So here are the seven key takeaways from Hormozi's video, how to win with AI. But first, hit like, hit subscribe, and hit the notification bell so you don't miss my next training. So the number one point that Hermozy starts with is that the time is now.
01:23Now I know if you've been in AI for the past couple years, you've heard this. Like, time is now. You gotta really learn this stuff.
01:29You gotta really learn this stuff. You may have heard AI will never be worse than it is right now, which is true. Okay?
01:35How to think about this? Imagine there's a really, really big hill to climb. Okay?
01:40And then you have two stick figures. One is running up this hill and the other one is staying at the bottom thinking, I am going to learn this stuff later. Well, the person who runs up this hill is going to be literally drowning in millions of dollars in revenue while this person who keeps waiting and keeps saying I'm going to figure this out later is not.
02:03They're just falling behind. They don't even realize it honestly. Like, most people I talk to just don't even realize all of the things that AI agents can do and help automate and help you be more productive.
02:14So in my particular example, right, so I've gone from zero to now 2,000,000 followers on social media solo. I'm now the number one AI educator for entrepreneurs in the world.
02:23And then on top of that, while I was doing that honestly as a part time hobby, I was also building a SaaS business, which is what takes up most of my time that now has thousands of paying customers. Okay. And this honestly takes up most of my time.
02:36It's a ton of work. I don't ever wanna say that it's easy. But because of AI, I've been able to build leverage for myself in ways where I would have traditionally had to hire a really big team just to even get started or scale to this level.
02:50So the first thing I want you to do, especially if you're new to AI, is open up ChatGPT or Claude and type this prompt. Like, even if you're really well versed in AI, the prompt is list three ways AI puts my income at risk and three things I should learn right now to stay ahead.
03:07And you should do this even if you're a beginner in AI, even if you are pretty inter if you're intermediate or advanced in AI, just see what it says. Just list three ways that AI puts your current income source at risk and the three things you should learn right now to stay ahead. Okay?
03:23Go ahead and try that. So Hormozi's second point is there has been there's never been a better time in the history of all startups in the world to build an AI first business. Now I typically use the term AI native.
03:37And what we mean by that is, like, you think about how to structure a team, how to structure a role and build out workflows, thinking about AI first. Right? So it's like you kinda like map things out.
03:48You how do you architect to the business and where are you going to put AI? You're thoughtful about that instead of kind of everybody else where you just kind of do things manually first and then you slot in, hopefully, AI afterwards. Um, so an AI native business is really, like, structured so that AI can play a productive role in every single department.
04:07And there's never been a better time to start this company. So just think about it this way, like a really big company with thousands of employees. These are supposed to be people.
04:15And then there's you, like tiny little startup. It's it's well known in the startup literature. Like, the biggest advantage a tiny company has is speed.
04:23K? Like, it's not resources, obviously. It's not people.
04:26It's not brand. It's you'd really don't have anything except speed. If you have ever worked at a big company by the way, my first startup was acquired by a company with over 5,000 employees, over 1,000,000,000 in revenue.
04:37K. So big company. I was a director there.
04:40Like, it moves slowly. It's cool when something happens because it can, like, hit the ground running. Like, it can go from zero to like, new product can go from zero to 10,000,000 very, very quickly.
04:51Right? But to even get it out the door this is a door. To even do this, to get it out the door, there's a ton of work involved.
04:59And it's a lot of bureaucracy. It's a lot of red tape. It's also just collaboration across many different teams, many different stakeholders, communicating updates, uh, deciding on trade offs, negotiating trade offs.
05:10Then there's politics on top. Like, there's just a lot of stuff going on when you have a big company, and so they naturally move a lot slowly. So if you combine your little company plus AI, then you get, like, lightning speed.
05:23Um, and that's really what it boils down to when you hear everybody talk about, like, leverage and AI can help you be more productive. It's really about moving faster so that you get the data you need to make better decisions, and then you go make those decisions faster with the help of AI. Okay.
05:40So an example of this that's really gone that's recently gone viral is Anthropix growth marketing departments. It was one person. And it's kinda cool.
05:49He, like, released his playbook in different ways he uses AI agents. So for example, for ads, like, he'll have an AI agent analyze the performance of his ads, look at other ads as well, come up with new copy, come up with new creative, and then deploy those ads and then wait again. Right?
06:05Keep AB testing ads in a continuous cycle. And this is what it's about. Like, whenever I think of AI or AI agents, I always think about, like, this feedback loop with your AI in the center.
06:14It's really about, like, can we create 10 different variations, test them very quickly, get the data we need, double down on the winners, test new creative, etcetera. It's just how quickly can you actually do this loop.
06:24And the old way of doing things, like every single step I just mentioned would be, like, manual and take time. If you're at a big company, approvals as well. Right?
06:33But in a tiny company with AI, right, you're augmented by AI, you can do each of the steps I listed very, very quickly again and again, getting the data you need, doubling down on the winners, trying new creatives, etcetera. And that is what leads you to better outcomes faster. Um, but let me give other examples.
06:50So AI native startups, we're seeing handfuls of them, okay, absolutely shatter the timeline to go from zero to 100,000,000 in annual recurring revenue.
07:00So AIR just stands for annual recurring revenue. It means like people will pay you each year. Um, so for example, companies like Cursor, Lovable, Higgs Field.
07:10Now all of these companies have raised traditional venture capital. K? So they're not bootstrapped, uh, $100,000,000 AR companies.
07:17But even so, like, in the traditional Silicon Valley playbook, you would still raise capital, but it it would still take you more than five years to hit a 100,000,000 in annual recurring revenue. I mean, this is an insane number. Uh, most of you here, like, we just want 1,000,000 AR.
07:31We don't need, a 100,000,000. These startups, and there's a handful more examples that I'm just not listing here, but these startups have compressed the timeline.
07:39So that's the other word I want to you to always think about when you think about AI and agents. I literally just think this is the image I think of. It's all about like speed, doing things quickly, and compressing timelines and compressing costs.
07:52This is what makes AI so exciting to a one person business because traditionally, to scale to even 1,000,000 in ARR, you would typically have to hire a team. Um, now you can do it with the help of AI in core functions, which we'll get to in a bit.
08:08Okay? But, yeah, these startups have compressed the timeline from, five plus years to, in some cases, like, under eighteen months to get to a 100,000,000 in annual recurring revenue.
08:18Like, I can't even fathom that. Like, it's it's very hard to fathom. Um, and you may have heard of the other viral example.
08:24Um, I think it's like MedVee, right, that was selling basically Ozempic. It's like a marketplace for you to hit up Ozempic providers.
08:32They're obviously in a lot of hot water for unethical business practices. Okay. Number three, they talks about, it's all about, like, skills and leverage.
08:42So I'm gonna talk about this in detail. I think it's really important. Um, so Hormozi's point is that, like, every skill that you give AI gives you disproportionate leverage over your competitors who just haven't figured it yet.
08:55And I view it as like kind of like a staircase of learning. Like, you you kinda just start with prompts and stuff, I guess. This would be the first one.
09:02And that's good. Like, a lot of my short form content are honestly just like really basic prompts. It's cool.
09:09I like to create prompts from different perspectives so that people kind of understand, oh, like AI will give you a totally different answer if you change a few words in the prompt. Okay. The next is like being more structured and methodical about the context that you're feeding into AI.
09:23Right? So instead of just a single prompt, you might create a project that contains a bunch of your context. And then the third level after this, would say is tool slash MCP.
09:32So for example, maybe you created a project in Claude and it knows how you like to write email replies. Okay. But the next step here would then be connecting it to your actual email system, not only to learn from past emails that you've written, but also to help you draft replies.
09:48Um, and then you can even schedule, like, a recurring task for your AI agent to send you an email brief every single morning or to escalate urgent emails to the top of your attention. Right? So this is generally the learning path I see most people go through.
10:03So we'll call this number 1. Call this number 2, we'll call this number 3, and we'll call this number 4 if you have proactive AI agents kind of, like, scheduled running in the background or with the webhook set up to trigger uncertain events. But I would bet that most people are still here.
10:19Like, you have to pause to think about this. Like, we're here on a Friday afternoon watching AI on TikTok.
10:26Like, first, that's just weird. Like, that's not a thing that cool people do. And then on top of that, like, we're in a tiny bubble of the world already or over here.
10:35And then within this tiny bubble, most people are still at number one and number two. That's why I say the opportunity for AI education alone is massive. We're a little dot inside a little dot.
10:45But I'm very excited when people kind of progressively learn and proceed through these steps. Um, I know the amount of education out there is very confusing, um, but, like, this is generally the sequence that you get.
11:00Um, and, like, as you go here to proactive stuff, this is where it gets, like, a bit more technical. Um, but Claude is releasing a lot of, like, utilities where it doesn't have to be so technical. Like, now in Cowork, for example, you can schedule recurring tasks, which is really, really nice.
11:15So in the video, Hormozi gives this example that he has now launched AI native companies that have achieved multiple millions in revenue per employee. Now for those of you who've never run a business before, like maybe you don't have context for how groundbreaking this is, so I can give you a real example.
11:33In my space, there's a company I love. They're a competitor, so don't sign up for them. But I still love them because they publicly release all of their metrics.
11:43So the company is called Buffer. Okay? And if you Google or ask chat Buffer open metrics, you will actually see all of the revenue and metrics over time, the number of employees over time, etcetera.
11:54And for the longest time, their revenue per employee was around like 200, let's say $2.50 k. Recently, it's reached above 300 k per employee. Okay.
12:04But this is pretty typical actually for a SaaS startup for the longest time. In fact, I'd say this is on the higher end. When you look at AI native companies and Hormozi has proven this because he's launched his own AI native companies.
12:16And they're according to him, they're currently doing millions in revenue per employee because they were structured to be AI native from day one. So the change from this being the gold standard to this being, like, expected, quite frankly, is pretty massive, and it occurred in, like, such a short period of time.
12:35Like, I haven't been an entrepreneur that long. I graduated college in 2013, had no idea what I was doing.
12:42Don't know why people gave me money, to be completely honest with you. But at that point in time and for years after that point in time, again, this was the gold standard. Like, even if you're venture backed, even if you're bootstrapped, whatever, like, this is great.
12:55Like, you're you're playing great. Um, but now with an AI native company, this is just, like, kinda normal. Like, why don't you have millions in revenue per employee?
13:05Um, but this this just happened overnight. Like, this was not a thing five years ago. Like, not even five years ago.
13:12Okay? So if you're watching this and you have business background, you're doing the math in your head and honestly, you're probably not at $2.50 k per employees. Most businesses aren't.
13:22But when you build an AI native company, you really do have the opportunity to massively scale this number. And by the way, that's why I'm personally so excited about teaching, like, solo preserve tiny teams. At the end of the day, like, if I were to teach big companies how to use AI, like, that's great.
13:39I'm sure that I could make them billions of dollars. But fundamentally, I know that those gains don't trickle down to employees.
13:46Like, I know I've worked at a big company, guys. Like, you are just an item in a spreadsheet. The gains like, if a if I were to teach a company how to make a billion more dollars, like, that billion dollars would not even go to you.
13:58Um, and so that's why I'm passionate about teaching, like, solopreneurs so that you can just build your own income stream and not be dependent on a company that honestly could lay you off, like, whenever they feel like it, whatever they want. And they don't care. Um, companies are not built that way.
14:13They are not built to care. You can truly build, like, meaningful amounts of revenue. Now I I'm I've given big numbers, like millions and millions.
14:20But, like, honestly, how many people's lives would be changed if you had an extra 20 k per month? Um, the beautiful part is, like, once you hit 20 k per month, you realize, oh, shit. I could hit 100 k per month.
14:33Like, I I have never seen, at least in SaaS, I have never seen a SaaS business that hit 10 k per month that couldn't be scaled to 100 k per month. And that's kind of the beautiful part. That's why I focus so much on, like, one person just get to a meaningful milestone for you and your family, and then you will realize, like, oh, wait.
14:52I can still use AI to scale this up a little bit further. And maybe I wanna hire my best friend. Right?
14:57Because it's fun. So an example of how I personally use this staircase. Right?
15:02So let's think about my customer support agents called Blue AI. Basically, whenever a new message comes in, Blue AI analyzes the support message. It reads the help docs.
15:12So it has access to tools. So help docs, it has access to it has access to logs. And then it tries to answer the person's question with some text.
15:22So here's some text. This is supposed to be a chat bubble. Okay.
15:27Um, but this is all happening, like, automatically. Like, it's happening right now. So if you actually go to potato.com, if you have an account, you could ask my support bot some questions, and it's gonna do its best.
15:37Now it's not perfect, but it has access to a lot of context. Right? So it has access to all that context through tools.
15:44And it can perform actions on your account, like canceling your subscription, restarting your subscription. It can even give you discounts. Um, like, can do all kinds of things because it has access to tools that allow it to change or inspect the status of your accounts.
15:59And it has access to my latest help documentation. So it can answer, like, product questions, what's the current promo, etcetera. And it does a pretty good job.
16:07And this is all proactive happening twenty four seven. Right? It's not 100%, but it's about 70% of customer tickets I get.
16:17Um, the remaining 30%, so I ask it to escalate. Right?
16:21If somebody sounds really, really angry, I want to make sure, like, I am seeing that.
16:27Um, if if, uh, the help documentation was not clear, for example, I pushed a new feature or I changed the name of a button and the help docs are inconsistent now. Okay. So it knows to, like, flag areas of the help docs that I should revisit and clean up to ensure they're consistent.
16:46And also just questions that don't have, uh, very clear answers in the help doc. I basically set a confidence threshold. I think it's like 96%.
16:56Okay. So it, uh, I have another agent that basically cleans up open tickets. And if it's 96% confident in the answer, it will automatically close the ticket.
17:06Okay. And then there's an adversarial agent that double checks its work.
17:13So this first agent will, like, close a bunch of support tickets. This adversarial one will be like, hey. Hey.
17:19You shouldn't have closed Bob over here. I think he could be angry. Okay.
17:24Um, so it's like, uh, three different agents working together. So this is kind of like the primary one that's running twenty four seven, answering support tickets. This is one that helps clean up any open tickets that can be closed.
17:36And then an adversarial one that's like, I don't know about closing this ticket. I think you might, you know, piss off this customer. I think we should leave it open and wait until Sabrina answers it.
17:45But overall, the this system, um, automatically handles about 70% of customer support tickets. Um, because honestly, a lot of questions people ask are in the documentation.
17:56It's simple things like, oh, how do I, uh, get cohort to use the pre pre signed how do I get cohort to upload my photo to like a question like that. Or like, how do how do I connect my accounts?
18:09I mean, you would be surprised how many times a day I get that question. How do I connect my social media accounts? Um, okay.
18:16So next, the task I want you to do to try something like this is to open Claude. Okay? Open up your connectors.
18:25I think it's customize on the left sidebar and then connectors, and then connect Gmail. Okay?
18:32Now once you've done that, open a Claude conversation and ask summarize my unread emails and flag what needs a reply today. Okay.
18:41So go ahead and do that. And in the process, you will have actually made it all the way to step three. Now proactive, if you wanna take it to step four.
18:50Right? So this this is already steps one, two, three.
18:54Step four in Claude CoWork, um, you can click schedule on the left sidebar and tell it to send me a daily email every single morning and summarize my unread emails, something like that.
19:09But this is proactive. Right? So that's like step four.
19:12Um, so it's it's not actually not crazy to go through steps one, two, three, four very quickly in a single example. It's just most people, like, don't realize it or they get confused or they get intimidated by the complexity of information.
19:26But I promise, like, you can pretty much bucket all AI education roughly into these four, like, kind of basic prompts.
19:34Like, oh, that's cute. Like, honestly, with my education with prompting, I just want people to realize, like, you can drastically change the outputs you get just by changing the prompts.
19:45Like, I want people to realize they're the ones actually in control and that they need to be careful with, like, what they're prompting because it will give a particular type of response. So how do you actually break things down?
19:56Like, what is the right way to think about it? In Hormozi's video, he talks about thinking in terms of workflows, not roles.
20:04So don't think of like a Facebook ads person. Break down what that person does into concrete tasks and then literally go to AI and be like, hey, how do I use you? Okay.
20:16So like, instead of thinking like this is a whole support person, okay, just think about in terms of the discrete tasks they do. So they read a message.
20:26They look up help docs. So let's say search for information. Okay.
20:30They draft a reply. Okay. Then they send the reply.
20:35And in some cases, they escalate to a human. Or in if it was a human, they would escalate to a manager or someone more senior.
20:43So instead of thinking, oh, how do I use AI to, like, create a support person? That's, like, really overwhelming because a support person actually does many things. So instead, break it down into concrete tasks.
20:56What does a support person do? They read the message. They search for relevant information.
21:00They draft a reply, review the reply, they send the reply, and in some cases, they ex escalate to a human. When you break it down into these steps. Okay?
21:10Then oh, it was too close there. Then the next thing is to go dump this list into chat or Claude and literally ask, here's all the things I have to do.
21:22What steps can AI handle today? And what tool do I use for each one? Okay.
21:28Like, that's it. So somebody who just asked about nonprofit fundraising.
21:33So I would challenge you to, um, create this list. By the way, AI can create this list for you.
21:39You can say, hey, chat. Interview me to figure out what are the things I actually do every day. Okay?
21:45It's gonna make an it's gonna interview, ask you a bunch of questions. It's gonna make this really nice list. And then you're going to AI and ask what are the areas you can help me with today?
21:55And what is the tool that I need to use for each one? Okay. So a lot of people overwhelmed with AI, especially business owners, they think about it at it too high an abstraction level.
22:05Like, how do I use AI to do sales? And I'm like, I can see how that's overwhelming. Um, because, like, uh, it's like sales is a complex task that involves, like, a lot of moving parts and you need to continuously give feedback, improve the system.
22:20It's never gonna work on the first try. So just don't think about it at that abstraction level. Think about it like, what does a salesperson do?
22:29So they find leads. Can AI help me with that? They qualify leads.
22:34Can AI help me with that? Or maybe I should plug in an existing API like Clay, like it already works. Um, number three, they reach out to leads.
22:42Okay. How can they reach out? Could they DM on LinkedIn?
22:45Is there an AI tool for that? Um, could they send or write a personalized email? Is there an AI tool that could help me for that?
22:52Then they book a meeting. They have a funnel. Oh, maybe AI could help us write the funnel and improve the copy so we get more conversions.
23:00Does the funnel have a video sales letter? Oh, maybe AI could write the script for our video sales letter. You can even use AI to edit your video sales letter, even though I think it's probably overkill.
23:11Um, so just break it down into really concrete tasks and then go ask AI, like, AI, be brutally honest with me. What can you help with today and what tool do I have to use for each of these tasks?
23:24Okay. So think in terms of workflows, not like entire roles.
23:29And generally, things become far less intimidating. Because a lot of that fear and, like, confusion is also because you're just you're, like, trying to use AI to replace what a real person does, but, like, a real person does a lot of things.
23:42It's not realistic. So, like, just think about it in terms of, like, the discrete concrete tasks, uh, that you do on a daily or regular basis for that role.
23:53Okay. Number five is what I call the TCCA stack.
23:58Uh, and the the idea here is, like, you wanna train AI just like you would train a person. We're gonna call this TCCA.
24:06It's a fancy acronym. So what does it even stand for? That's a good question.
24:10So whenever you're talking to AI, like, is useful to have some kind of framework. It doesn't matter which one, honestly. Just like whatever you can remember.
24:19So t stands for the task. Right?
24:22So you're going to explain the thing you want. Like, uh, I want you to write me an email. C is for context.
24:29Well, what is the email about? This is a reply to a customer support complaint. The other c is the constraint.
24:38Okay. This could be like our company is this. So only say information about our product, not anything else.
24:46Con constraints can come in all various forms. It could be like, um, I don't want em dashes in my response. Like, that's a constraint.
24:54Or, um, I want the reply to be less than 100 words. That's also a constraint. And then the last a is my favorite.
25:01It's the one I use all the time because I'm super lazy and I can't remember any frameworks. Um, ask me clarifying questions.
25:10This I love this one because you can just kind of skip a lot of the other ones, and AI is going to ask you. It's just it's literally gonna ask you, wait.
25:18Can you clarify what the task is? Can you give me some more context? Can you specify if you have any constraints?
25:23I guess, I'm gonna go through this exercise with you. Um, but, yeah, use some kind of framework like this so that, uh, you're giving AI useful information, sufficient information to be able to do its job well. So, yeah, as a next step, so I would challenge you to just pick, like, one task.
25:41So just choose one task that you do on a repeated weekly basis and open up AI. Okay?
25:48And then write it out using this framework. So, like, what is the task? K?
25:53Explain it. Add some additional context. If you have any constraints, if none come to mind, that's fine too.
25:59And then ask me clarifying questions. You just append this to the end of your prompt. So go to AI and do that.
26:05Like, here is a task I do on a weekly basis. Blah blah blah. Write it out, paste it into AI.
26:12And then if you like the output, you can save it as a reusable skill or GPT.
26:19Okay? So that way you don't have to repeat yourself every single time with all the same context, all the same constraints, etcetera. K.
26:26So this is a really simple prompt stack that I just like to recommend to people. Because, like, so many people are like, am I prompting AI correctly? There's no right or wrong.
26:39What matters more at the end of the day is to to me, is AI making you more money? All this other stuff around, uh, like, this is the right way to do it.
26:48This is not the right way. You're gonna get better outputs here. Like, yeah, you can get better outputs, but, like, at the end of the day, are you taking action on it?
26:58Um, if not, why? Like, was the plan you came up with too complicated, too overwhelming, not realistic given your constraints?
27:05K. So, like, ultimately, the only metric I actually look at is, like, is AI helping me scale my business or my brand with the only metric that matters in business that's revenue or revenue per employee.
27:18And for your brand, uh, that might be followers or something like that or watch time or something. Um, so point number six that, uh, Hormozi makes is that the last valuable thing a human can do is take risk. And like he goes on to say, like either double down on the AI native approach and build a business that's doing millions per employee per year or just like focus on a business where the human elements will never be removed.
27:46Okay? In in my case, like, I would phrase it as bet on yourself.
27:52Okay? And I briefly talked about this earlier. It's just that, like, I'm so passionate about teaching AI to individuals instead of big companies because I see the opportunity right now for one person or tiny team to build a meaningful amount of income without having to hire a traditionally big team and the headache that comes with that, by the way, um, without having to raise a ton of venture capital.
28:20K? So I'm a big believer in just, like, bet on yourself. And that also means, like, learn as much as you can, build income streams for yourself.
28:31And honestly, once you do this, like, once, you you will have the skill to be able to do to get. Like, you can build additional businesses if you want.
28:41And so I'm just really bullish on that. He gave the example in his video that his friend spun up a division inside his own company with one mission, which was to put the larger business out of business. Um, I think that's a super cool challenge.
28:56Um, I would just question, like, what you get out of it. Um, because like I said, at a big company, the gains don't really trickle down to employees. That's just the reality.
29:07Um, but they do provide, you know, stability kind of, security kind of. Um, but yeah. Anyway, so that's like, he talks about that, um, in terms of, like, building your own AI business or doubling down on the stuff that's very human and, like, just cannot be replaced by AI for, like, live experiences, um, like, uh, entertainment that involves, like, human actual humans and stuff like that.
29:33I'm gonna spin that one a little bit and just make the point, like, I believe you should bet on yourself. That means learning as much about AI as possible, trying to build AI businesses until you find one that you actually like and will stick to. And once you develop that skill, you'll be, like, much more confident so you can do it again.
29:48You can build another business if you want. So Hormozi's point number seven is every single day, audit what you're doing.
29:55K. Calls it daily task audit.
29:59So for example, just write down what you do every single day. So step one, yeah, literally write down what you do every day. Um, if you can figure out how to take tasks that typically take you thirty minutes and with the help of AI compress them down to two minutes, then that's huge.
30:16Right? That's like a huge, huge win w. Okay.
30:21Um, so in the Hormozi in Hormozi's example, he talks about Facebook ads in particular. So, like, if we actually break down what a Facebook ads person does.
30:31Right? There's like multiple steps involved. And I'm learning Facebook Ads right now from the amazing Mitch Barnum.
30:37And so it's it's really interesting for me to hear about his Facebook Ads examples. So break it down into break down what what it means to run Facebook ads, and you'll see all of these concrete steps. And then what you wanna do is, again, go to AI and figure out, hey.
30:54How can AI help me with this one? So for example, when it comes to running creative, I recently ran some, like, bottom of funnel Facebook ad campaigns and just used Nano Banana to make, like, a poster.
31:08Uh, so I have, like, three posters on Facebook. They're just Nano Banana made. One of them really does not look like me, but they're doing really well on Facebook and they just made them with nano banana.
31:19Um, so that's an example, a real example of using AI for creative to drive real revenue. Um, and then you can also use AI for analysis.
31:29Like, I'm still pretty new to Facebook ads. So I ask Claude all the time, like, what does this mean? Like, I don't know what this means because I'm in the process of trying to develop my own intuition around these numbers.
31:40So I'm always having Claude, like, explain, what does this mean again? Like, is this good? I don't know.
31:45Like, just just keep asking it for help with all of these things. And, of course, um, I use AI a lot for helping me with copywriting, um, whether it's email copywriting, headlines, for example.
31:57When it comes to, um, headlines, I just help use Gemini to help me write some email copy to reactivate churned users for my product. Because I am not very good at, like, like, revenue generating email copywriting.
32:12Like, it feels, like, sales y to me, and I don't wanna write that way. At the same time, it does convert very well.
32:20So it's something I'm experimenting with, and that's where AI is teaching me. Like, I'll write a draft. Then it writes its version.
32:26And I'm like, well, why do you think yours is better? And it explains to me like, well well, Sabrina, mine is, like, actually harping on an emotion, and yours just describes your tool.
32:35And I'm like, oh, that makes sense. Right? Like, emotions.
32:39That's a good one. Um, but, yeah, just like figure out break down what the role is. Like I said before, discrete tasks.
32:47Literally go to AI, ask it how could it can help with each one. And you just do this on a weekly basis. So every day, you just write down the tasks, look for something that currently takes you thirty minutes but can be compressed in two minutes.
33:00Like if I had tried to make this poster in Canva, it would easily take me thirty minutes. Like probably two hours because I'm that bad at visual design.
33:09Um, but with Nano Banana, it took me literally two minutes. So now I can, like, create lots of different creative and test out lots of different ideas in a fraction of time it would have normally taken. Right?
33:20So it goes back to, like, speed and compressing the cycle. Like, it's always about speed and compressing cycles of data and iteration and experimentation.
33:29So if you wanna take the next step here and, like, try something like this, go to YouTube and go to my YouTube and look for my Claude CoWork tutorial. I think it's called I think it's called five insane use cases or something.
33:46So look for this on YouTube. This one was really fun because it basically walked through building a personal AI email assistance and then setting up an email brief.
33:58It also walked through creating social media call contents, scheduling out your content, and also making unlimited videos for free inside Claude CoWork, uh, using open source video generation libraries. Okay.
34:10So this one is really fun to go through, and this one will very much level you up if you're trying to understand how can I use AI in my real world day to day? Okay. So, yeah, that was it.
34:22So we covered how to win with AI in 2026. Just going through each of Alex Hormozi's talking points and then adding my perspective. If you like this, make sure you hit like, hit subscribe, and hit the notification bell so you don't miss the next training.
The Hook
The bait, then the rug-pull.
Sabrina Ramonov opens by torching the most common AI mistake in one line, then backs it with receipts: zero to two million followers solo, a profitable SaaS with thousands of customers, no team. The promise is a seven-point operating playbook for the one-person business that wants to run like a company of ten.
Named ideas worth stealing.
23:54acronym
TCCA Prompt Stack
- Task
- Context
- Constraint
- Ask clarifying questions
Four-part prompt structure. The A (ask clarifying questions) is the lazy-genius move.
Steal forAI education content, onboarding, prompt templates for JoeFlow or MCN+
08:34model
4-Step AI Leverage Staircase
- Basic prompts
- Context / Projects
- Tools / MCP connections
- Proactive / scheduled agents
Progression model for AI capability. Most users at step 1-2. Steps 3-4 are where leverage compounds.
Steal forPositioning JoeFlow: show where it sits on the staircase and what it unlocks
14:58model
3-Agent Support Stack
- Primary (answers + closes at 96% confidence)
- Cleanup (processes open tickets)
- Adversarial (double-checks closures)
Multi-agent customer support that auto-handles 70% of tickets. Adversarial agent catches false positives.
Steal forSaaS support automation; great demo format for AI agent content
29:48concept
30-to-2 Compression Audit
Find one task that takes 30 min, find the AI tool that compresses it to 2 min.
Steal forWeekly content hook or newsletter section
11:13concept
Revenue-per-Employee Benchmark
Buffer = ~250k/employee (traditional SaaS). AI-native = millions/employee.
Steal forPositioning argument for any solo-operator or tiny-team product
How they asked for the click.
VERBAL ASK
33:34next-video
βGo to my YouTube and look for my Claude CoWork tutorial. I think it's called five insane use cases.β
Soft CTA to related tutorial, not a product pitch. Subscribe CTA bookends video at 01:04 and 34:25.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
