Is generic AI consulting a good business?

I talked to two people in two days who were doing generic AI implementation consulting. Both said it sucks. Same reason both times: every client's setup is different, so every job starts from zero.

AI consulting is still the highest-probability way for most people to make money from AI right now, because faster change means more people need help. The generic version is the trap.

Why does generic AI consulting break down?

A friend of mine built an AI consultancy off YouTube in 2026. Clients came in. Then came churn and chaos, because each one wanted something different.

"Every single person is just doing something different. It's chaos. There's no compounding, there's no scale."

When every engagement is custom, nothing you built for client one helps with client two. A pest control owner I talked to described it the same way from the buyer's side: serving lots of industries with custom work is a services treadmill.

What does picking a niche actually fix?

• The work compounds. Go deep in one vertical and 90%+ of the solution is the same for every owner in it. Onboarding becomes configuration, not construction.
• Selling gets simpler. Imagine "we build AI agents for HVAC companies. That's it." You hit every HVAC company hard, you go to their conferences, and most of what you build applies to everybody.
• Your first client sells the next one. Build one general engine, change about 10% for the industry, and use the first customer's case study to close the second.

A founder I advised had seven customers in seven different industries. That's what generic looks like. He was hunting for one industry that would pay $2,500 a month, and by his own math 15 pool clients at $5,000 a month is a million-dollar business.

Where does a small consultant beat the big AI labs?

Right there in the niche. Anthropic and OpenAI aren't going to build hyper-specific, multi-step workflows for a narrow vertical like local media. It isn't worth their time. It's worth yours. More in will the AI labs just build my product.

How do I pick the vertical?

• One level deeper than the category. "Home services" is too broad. Pest control or pool service is a niche.
• Fragmented, light on regulation, seasonal. That's why I like pool service. Lots of small operators, and the software layer is hard for private equity to roll up.
• Recurring revenue. I spent eight years as an insurance adjuster working with roofers. Roofing has big jobs and almost no recurring revenue, which makes it a weak software customer.
• Bring more than AI. Marketing and Meta ads skills on top of AI are rare among AI-only operators.

The full checklist is in how to choose a niche for an AI consulting business.

What should a niche AI consultant charge?

Position yourself as the full back end, not the person who builds automations: recruiting, marketing, SEO, accounting help, the works. That supports $3,000 to $5,000 a month. For a productized system that's about 95% pre-built, I've priced setup at $2,000 to $10,000 plus $300 to $800 a month to manage it. Breakdown in how much to charge for AI automation services.

How do I land the first clients in a niche?

I still think trainings are the top of the funnel. You run a workshop, you show what's possible, and people ask how to do it. Your answer: watch 8,000 YouTube videos and spend forever, or pay me and I'll just do it. That same friend ran a paid workshop for 25 people and three became clients at $2,000 a month each.

The window is now. I think it's the next 12 to 24 months, before business owners figure this out and prices compress.