Alex Hormozi's AI Strategy: What His Company Actually Does in 2026
Alex Hormozi froze his team's AI budget despite heavy usage. A breakdown of how Acquisition.com uses AI in support, sales, content, and hiring.
What is Alex Hormozi’s actual AI strategy?
Alex Hormozi runs AI through nearly every function of his company, Acquisition.com, from customer support and sales to content production, but he froze his team’s AI budget in late September after noticing that AI usage kept climbing while revenue stayed flat. His point isn’t that AI doesn’t work. It’s that most AI spend goes toward tasks that were never the business’s real bottleneck in the first place. He wants people using AI on the one constraint that actually limits growth, not on busywork that now just costs tokens instead of time.
TL;DR
- Hormozi froze additional AI budget for his team after finding that AI usage per employee had grown 16 times beyond the average tech-forward company while company revenue stayed flat.
- His support team built five AI agents that handled around 120,000 tickets during a book launch and resolved 90% without a human.
- An AI text-based sales agent reportedly sells around 200 units a month, but Hormozi says humans still outperform AI on live sales calls after years of voice AI development.
- He tells a cautionary story about a company that spent $350,000 building an AI system to replace 11 virtual assistants who were already doing the job for about $11,000 a month, on a task that wasn’t even the business’s real bottleneck.
- His test for whether AI is actually working: if your token bill rises but views, leads, or money don’t, you’re automating the wrong thing.
- He draws a hard line at decision-making, arguing that offloading judgment to AI models makes people “dumber,” even as he automates writing, editing, and support.
- His advice to employees is to automate their own job before someone else does, framing the future of work as “bring your own agent.”
How is Hormozi’s team using AI in support, sales, and content?
The clearest public example is customer support. During a book launch, his team deployed five AI agents that fielded roughly 120,000 support tickets, most of them simple requests like where to find a free bonus. Those agents resolved about 90% of tickets without a human stepping in, which is the kind of ratio most support teams would consider a success story on its own.
On the sales side, Hormozi has talked about this in stages over time. In January 2025 he predicted an AI appointment setter was close, a few months off. By April he said an AI sales development rep (SDR) was matching his human team’s performance. By the fall, he got more specific: a single AI text agent sells around 200 units a month, but he’s still hiring human SDRs because no AI system has yet matched a skilled human on live phone calls, despite roughly two years of work on voice AI at his company. That’s a useful data point for anyone building AI sales tools: text-based AI selling is working in production, voice still isn’t fully there.
Content is where the compounding effect shows up most. Hormozi says his media team’s output has tripled without any increase in headcount, because AI-driven workflows now handle 30 to 50% of what used to be manual editing work. He breaks this down by asking what the five or six discrete tasks inside a job actually are, then turns each one into its own automated workflow instead of hiring a person to do all of them end to end. He applies the same approach to video editing (an AI workflow finds the highest-tension moments in a transcript, trims dead air, and assembles a rough cut) and to his own writing, where idea generation and drafting happen faster with AI, even though he insists AI isn’t writing the final version. He says tweets that used to take him 90 minutes a day now take about 90 minutes a week, a roughly sevenfold efficiency gain, achieved not by automating the writing itself but by using AI to generate raw material he still edits and decides on.
What is the $350,000 mistake Hormozi keeps warning about?
Hormozi tells the story of a company he evaluated that had 11 virtual assistants doing data cleaning work for about $11,000 a month. The work was getting done fine. The company then spent $350,000 trying to build an AI system to replace those 11 people, and the project wasn’t even touching the actual constraint holding the business back.
That $350,000 is roughly three years of what the VAs were being paid, spent to automate a process that wasn’t broken and wasn’t the bottleneck. Every business, in his framing, has one thing capping its growth at any given time, usually leads, sales, or delivery. Spending heavily to automate something outside that constraint doesn’t move the needle, no matter how well the automation works technically.
Without shadow IT.
His stated fix is to question whether a task deserves to exist before automating it. He’s explicit that it’s more efficient to decide something isn’t a priority than to spend money automating something that was never a priority to begin with.
Why did Hormozi freeze his own team’s AI budget?
The freeze, announced September 30th, came from the same logic. He noted that his company’s AI usage per employee was running at roughly 16 times the rate of an average tech-forward company, while revenue stayed flat. For him, that gap is a signal that people are spending time and tokens on work that doesn’t actually matter to the business, not proof that AI itself is failing.
He’s framed the underlying problem this way: if you’re using AI constantly and not making more money, you had the same problem before AI. You were doing work that didn’t matter, and now it costs tokens instead of just time. He also makes a point about AI vendors directly: companies like Anthropic and OpenAI bill you whether what you built with their tools was useful or not. The meter runs regardless of outcome, so the discipline has to come from the user, not the tool.
Instead of expanding the budget, his team now gets training aimed at using fewer tokens for the same output and directing effort at tasks tied to the business’s actual constraint. He’s turned this into a simple self-check anyone can apply: since you started using AI, are you getting more views, more leads, or more money? If the token bill keeps climbing and none of those three move, the problem isn’t AI, it’s that you haven’t found your real bottleneck yet.
Is Hormozi replacing employees with AI?
Based on his public statements, no clear case exists where he’s said he let someone go specifically because of AI. He says Acquisition.com is still hiring, and points out that OpenAI and Anthropic themselves employ thousands of people despite building the most advanced AI tools available, as evidence that AI doesn’t automatically shrink headcount.
That said, he’s direct that AI raises the performance bar for everyone on a team. He’s said publicly that owners should raise standards company-wide, and that people who can meet the new bar stay while those who can’t, don’t. He’s also advised at least one business owner, running a $20 million website company, to spend six months rebuilding workflows around AI and cut headcount in half, a move that would roughly double margins. So his position isn’t “AI won’t cost jobs.” It’s closer to “AI will cost jobs for people who don’t adapt, but it doesn’t require layoffs by itself.”
His advice to individual employees is to get ahead of that shift by automating pieces of their own job before someone else does it for them. He calls this “bring your own agent,” meaning workers should train AI systems on their own tasks so they can offer a company automated capability instead of a traditional headcount hire. He suggests blocking off a few hours a week specifically to automate or speed up one task at a time, building momentum gradually rather than trying to overhaul a whole job at once.
Remy.
What should AI never be allowed to do, according to Hormozi?
Decision-making. He’s described testing the same straightforward business decision across multiple AI chatbots and getting wildly inconsistent answers, concluding that people who hand real decisions to AI get worse at thinking over time. His rule is to let AI handle typing, drafting, and research, but never delegate judgment.
He extends that logic to what he thinks keeps paying off for humans as AI gets cheaper: risk and stakes. His argument is that AI can generate infinite recommendations, but someone still has to own the decision and bear the consequences if it’s wrong. He uses content creators as an example: a creator who actually built ten real sales teams and then shares six tips will beat someone who got the same six tips from a chatbot, because the first person’s claims are backed by real stakes and real outcomes. He’s also openly critical of AI-generated writing, citing patterns like forced negation (“it’s not X, it’s Y”) as tells of low-effort content, and he’s described giving AI writing tools a fixed set of rules and real writing samples to make output usable.
Frequently Asked Questions
Did Alex Hormozi cut his AI budget or increase it?
He froze additional AI budget for his team on September 30th. He didn’t eliminate existing AI spend, which remained far above typical industry levels, but stopped approving increases because revenue wasn’t growing alongside usage.
How many support tickets does Hormozi’s AI handle?
During a book launch, five AI agents handled around 120,000 support tickets and resolved about 90% of them without human intervention.
Does Hormozi use AI for sales calls?
He says a text-based AI sales agent sells around 200 units a month, but human sales reps still outperform AI on live phone calls, even after roughly two years of voice AI development at his company.
Has Hormozi fired employees because of AI?
No public statement from him confirms layoffs tied directly to AI. He says his company is still hiring and points to AI labs themselves employing thousands of people as evidence that AI adoption and headcount growth aren’t mutually exclusive.
What does Hormozi think AI can’t replace?
Human decision-making and risk-taking. He argues AI can generate recommendations but can’t own a decision’s consequences, which is why he treats judgment as something to keep in-house even as he automates writing, editing, and support.