AI Side Hustles Half-true — works only if you do the unspoken work

By the Income Reality Check Editorial Team · Updated 2026-10-16 · 7 min read

Selling AI agents for $2,000 setups: what the Koerner Office pitch skips

Verdict: Half-true — works only if you do the unspoken work. The service model is real and the unit economics check out, but the video hides the sales grind behind the word “automate.”

In “The Easiest Way to Make Money With AI Right Now,” Chris Koerner interviews Aaron Edwards, founder of the chatbot platform DocsBot, who says his software business passed $1 million in annual recurring revenue and that non-technical people are charging local businesses “$2 to $3,000 to set this up” plus “$500 to $1,000 a month to keep running it.” Is that real? Partly. The AI-agent agency model exists and some people are doing exactly those numbers — but the hard part isn’t the AI, it’s landing clients, and the person making the pitch sells the platform those agencies would resell.

What the video actually claims

Edwards tells a genuinely useful founder story. He tried eight AI businesses in six months while employed — a WordPress plugin, AI product photos, AI profile pictures (about $10,000 in two months before the fad died), a personalized children’s book that got him on Good Morning Texas and produced “not a single sale” — before DocsBot, an AI customer-support chatbot platform, stuck. Three and a half years later he reports over $1 million ARR, roughly 90% gross margins, 3% monthly revenue churn, and a lifetime value just over $2,000 per customer.

The opportunity he pitches to viewers is one step removed: become the implementer. Use ChatGPT or Claude (a $20/month subscription, he says) connected to lead-scraping APIs to find small businesses with heavy support loads, auto-generate a demo bot from their own help docs, and sell it with a $2,000–$3,000 setup fee and a $500–$1,000 monthly retainer. His justification math: if a support ticket takes five minutes of a $20/hour employee’s time and the bot deflects half of 2,000 monthly tickets, the client saves roughly $7,000 a month and happily pays $500. As proof, he offers his agency partners’ reported price points and his 19-year-old son, who just signed his first client — a local contractor who said the service could save him $100,000 a year.

The episode’s hook line is the tell: “you’re not even doing anything” for that $500 retainer.

What the method actually requires

Start with the parts that survive scrutiny. The value math is anchored in real labor costs: the U.S. Bureau of Labor Statistics puts the median customer service representative’s pay at $44,770 a year ($21.53/hour) as of 2025, with employment projected to decline 5% through 2035 — and the BLS attributes that decline partly to automation. Edwards’ “$20 an hour support agent” figure is, if anything, slightly conservative. The software costs are also roughly as described: DocsBot’s own pricing page lists plans from $49 to $499 a month, so an implementer paying $49–$149 per client and billing $500 has real margin.

What the video compresses into a sentence is the entire customer-acquisition problem. Edwards himself admits, almost in passing, that his inbound software sales “dried up” over the past year and that DocsBot had to pivot to white-glove service and courting agencies because self-serve customers stopped showing up. That’s the market telling you something: the product is easy; the selling is not. His own ranking of what works — in-person at chambers of commerce and conferences first, then phone calls, then cold texts, with cold email “down there” — describes classic door-knocking B2B sales. Koerner adds that cheap outreach converts at around 0.1% and that moving to roughly 12% requires expensive, personal gestures like mailing gifts to prospects. Neither man puts an hours number on it, so here’s a sober one: at even a few percent close rate on warm-ish outreach, landing your first three $2,000 clients plausibly means 10–20 hours a week of prospecting, calls, and demos for several months. The AI writes the demo bot in minutes. It doesn’t get a contractor who’s “installing water heaters” to return your call.

There’s a second omission. Edwards notes that “build it myself” cancellations at DocsBot rose from 1% to about 13% of churn in a year, and that the “AI receptionist” end of the market is already served cheaper by tools like GoHighLevel. Translation: the easy accounts are getting more self-serve and more contested every quarter. The defensible version of this business is the complicated one — manufacturers with spec-heavy product manuals, policy-bound support flows — which takes genuine discovery work per client, not a one-click demo link.

And one structural caveat: this is an interview with a vendor. DocsBot earns $50–$149 a month from every agency that resells it. That doesn’t make the numbers false — his pricing page matches what he says on air — but every “anyone can do this” in the episode is also a partner-program recruitment pitch, and you should weight it accordingly.

Who actually wins this game

Look at who’s actually succeeding in the episode. Edwards had years inside WPMU DEV watching thousands of support tickets, made his employer his first customer, and caught a viral distribution break in Japan. His son’s first client is “a family friend or someone I knew.” Koerner’s guests who make agency economics work tend to share the same traits: an existing network of business owners, comfort with face-to-face selling, and enough domain fluency to scope what a client actually needs. If you already sell websites, marketing, or IT services to local businesses, adding a $2,000 AI-agent SKU is a credible upsell. If you’re starting from zero contacts and zero sales experience, you’re not competing on AI skill — you’re learning cold outreach, the oldest and hardest part of any service business. We covered the same dynamic in our review of “proven ways to make money with AI”: the production step is automated; the distribution step never is.

What you’d realistically earn

The ceiling is real: ten retained clients at $500 a month is $5,000/month recurring plus setup fees, and nothing in the fee math prevents that. The floor is also real, and it’s zero. Survey data on side hustles consistently shows median earnings around $200 a month precisely because most people stall at the selling stage. A realistic first-year arc for a newcomer with no network: 2–4 months and dozens of pitches to land client one, perhaps $4,000–$15,000 in total first-year revenue from a handful of setups and retainers — decent money, nowhere near passive. For U.S. readers, there’s also a regulatory floor under the hype itself: the FTC’s Operation AI Comply has already sued schemes that promised five-figure monthly income from “AI-powered” businesses, with consumer losses topping $25 million in one case alone. And if you start selling this opportunity onward to others with income claims, the FTC’s Business Opportunity Rule requires written substantiation for any earnings claim you make. The “$7,000 a month saved” pitch to a client needs to be defensible math, not vibes.

One more honesty check on the retainer. “A robot writes the monthly report” is true — until the bot answers something wrong about a warranty policy and your client calls you, not the robot. Budget a few hours a month per client for maintenance, escalations, and relationship-keeping, or watch your churn do what churn does.

Who this is (and isn’t) for

This makes sense if you can commit 10+ hours a week for at least six months, are willing to pick up the phone or show up at local business events, and ideally already know business owners who’d take your call. Capital requirements are genuinely low — $50–$200 a month in tools — which is a legitimate point in the video’s favor. It’s a poor fit if you want income without sales conversations, if “passive” is the word that drew you in, or if you’d be pitching businesses in a market (or language) where you have no network at all. (The model travels globally, but the trust-building is local everywhere.) For a broader menu of service businesses where demand still outruns competition, see our look at nine one-person businesses.

What to remember

Half-true, in the most literal way. The fee structure, the software costs, and the labor-savings math all check out against public data — this is a real service business, not an FTC-sweep fantasy. What’s false is the frame: “easiest” and “you’re not even doing anything” describe the 20% of the job AI now handles, while the 80% — months of unglamorous selling to skeptical local businesses — goes almost unmentioned by a guest whose platform profits from your attempt.

View counts and platform fees reflect what we observed at the time of review and may have changed since publication.