Sell AI to Small-Business Owners as a $12-an-Hour Employee—What Breaks First Might Be a Phone Number

Greg Isenberg threw out a business idea on X: build Grok Bots for boring industries and small and medium businesses. Give each agent its own computer, email address, and phone number, then charge $12 an hour, as if hiring an employee for the owner.

The clever part of this pricing isn't the technology. Small-business owners may not understand agents, but they understand what it costs to hire a person. At $12 an hour, calculated full-time, one commenter immediately worked out that it comes to about $25,000 a year. For plumbers, salons, and repair shops, that's easier to grasp than 'subscribing to an AI tool.' Salons are especially natural: appointments, reminders, rebookings, reviews—half of what a front desk does is repetitive. One reply said all the owner needs is someone to show them once.

But the reply thread didn't stop at the winning pitch. Someone asked whether small and medium businesses would be willing to share data. Someone reminded that each agent's email needs to be warmed up first. More people focused on cost: one commenter noted that Grok Bot offers weekly usage pools at $30, $100, and $200 per month tiers; another reply quoted Elon saying the cost of serving Grok Bot is unsustainable relative to revenue. If you charge customers $12 an hour, write the cost-side ceiling into the contract first; otherwise, once the customer has locked in the price and the platform limits move, margins get thin.

Phone numbers are a more concrete problem. The Oz says that once an agent has its own number and the owner's cousin texts it at 9 p.m., no one has told it how to handle that. Write the process for what to do when it doesn't know what to do first, then build the stack. That reminder is more useful than model parameters. CodeGlitch asks more directly: after usage limits get cut, what happens when this $12-an-hour employee runs out midweek? Gideon Dilebo offered a selling point: send the owner a note on Friday that spells out what it actually did this week, because a plumber isn't going to read logs.

Travis Wright's take is closest to the actual sales conversation. He thinks Greg read the market right, but the product is underpriced. No one who runs a plumbing shop wakes up wanting 'a $12-an-hour agent.' What they want is someone to answer the phone at 2 a.m., invoices that actually get collected, and Google reviews that don't read like ransom notes. Charge by the hour, and the agent becomes a cheap intern; charge for outcomes, and it looks like the best hire the owner ever made—one that doesn't take vacation. The bot is a tool; the outcome is the business.

So selling Grok Bots to small and medium businesses can work. The selling point shouldn't stop at $12 an hour. First calculate how much work an agent can handle in a week and who answers the phone after it hits the limit; then put 'one fewer missed call, one more payment collected, one more normal review' on the quote. Greg put the link to this podcast episode in the replies: podcast link.

The first reply is a crocodile stage video, unrelated to the business model, but it fits this space well: the excitement arrives first, the ledger arrives later.