AI as a business

Every answer 21 answers

• Can AI agents run a business on their own yet?

  Not yet, outside simple things. As of mid-2026 AI keeps getting better, but you can't just say "build this" and have it done, or I wouldn't be working 12-hour days. My goal is making human headcount optional wherever AI does the work reliably, using a small team of agents I can understand and trust.

• Can you make money building a niche version of old, boring software?

  Yes. Take a category that's been around 10 or 15 years, like review management or forms, and build the version for one industry you know deeply. A friend's review tool for self-storage owners reportedly makes about $12,000 a month. Sell it through that industry's conferences and Facebook groups.

• How can insurance agents use AI?

  Use AI to take over follow-up and admin so you have time for what agents are supposed to do and rarely do, like calling clients to review their policies. Captive agents are limited by the carrier's systems but can still use general AI for productivity. Independent agencies, with their own CRMs and manual processes, have far more room.

• How do I choose a niche for an AI consulting business?

  Pick one trade, one level deeper than a category, where you can reuse 90% of what you build. Look for fragmented, lightly regulated businesses with recurring revenue. "Texas business owners" is too broad, and so is "home services." Pool service or pest control is a niche. It helps a lot if you already know the industry's problems.

• How do I find clients for an AI consulting business?

  Go local and borrow trust. Most owners want to shake the hand of the person they hire, so chamber meetings, hosted events and referrals from accountants, fractional CFOs or coaches beat cold email. Publishing how-tos about connecting AI to software owners already use also brings clients to you. Workshops are the top of the funnel.

• How do I run an AI workshop for non-technical business owners?

  Make it hands-on and paid. I'm designing mine around $100 for three hours of prompt exercises instead of a passive one-hour talk. Lead with mindset before tools, promote it through an audience that already trusts you, and treat the workshop as the top of the funnel for implementation work.

• How do you run corporate AI training that actually sticks?

  Lead with mindset, not tools. Make it hands-on with laptops open and real work, standardize the company on one AI platform, and teach individual use before any shared team system. I run it as four touchpoints: an intro call, a prep call the week before, the in-person workshop, and a follow-up one to two weeks later.

• How long will the AI consulting opportunity last?

  My read, as of May 2026, is roughly the next 12 to 24 months, before business owners learn the tools themselves and prices compress. That's a forecast, not a guarantee. Rising AI token prices would help consultants, because owners will stop spending $400 testing tools and just pay someone to do it.

• How much should I charge for AI automation services for small businesses?

  For a productized AI 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. A full back-end operator retainer covering marketing, SEO, recruiting and automations supports $3,000 to $5,000 a month. A one-time Claude newsletter setup runs $500 to $1,000.

• How should a small business use AI without losing the human touch?

  Split by surface, not ratio. Anything a customer, reader or advertiser touches stays human. Back-office grunt work goes to AI, and people underrate how much it can now absorb. A business that's 100% human falls behind one getting 100x the output, and one that's 100% AI produces so much sameness that nobody cares.

• If anyone can build software with AI, what's the real competitive advantage?

  Audience. AI collapsed the cost of building software, so a great tool no longer protects you, and competition for attention is about to climb as software and companies multiply. An audience that trusts you can't be cloned. Build it first and you get options for how to make money from it.

• Is generic AI consulting a good business?

  No. Generic "AI for anybody" consulting is a trap because every client's setup is different, so nothing compounds. I talked to two people in two days doing it and both said it sucks. Pick one vertical, go one level deeper than "home services," and templatize 90% of the work so each new client is configuration.

• Should B2B software be self-serve or high-touch at first?

  High-touch first. I sell lightbreak as a high-cost, high-touch platform and onboard the first customers myself, with no Stripe checkout or automated onboarding. Build self-serve after the managed version works, the interface is solid, and you have revenue and referrals to show for it.

• Should I build AI software or sell AI-powered services?

  For most people right now, sell the service. AI-driven marketing and implementation use the same tech without the complexity of software, and demand for help grows as change speeds up. Build software only for a narrow problem you know, for an audience you can reach, and consider running it as a managed service first.

• Should I sell AI services as one-off projects or a membership?

  Both, in that order. Land the client with a one-off build, then move them onto a membership. Project work has no compounding and no floor, and when you build it for them you own every fix. A membership sells access, coaching, support and building help while the client keeps ownership of what gets built.

• Should I white-label GoHighLevel or build my own CRM with AI?

  Start on GoHighLevel's white-label plan. At $300 a month it already covers marketing, voice, forms, workflows and multi-niche templating. Building that from scratch leaves you with zero customers and ongoing maintenance. Your moat is the roughly 20% of workflows it won't do natively, which a custom layer can fill.

• What should a software MVP include, and what can you do manually at first?

  Only what the client needs to get value. Do the rest by hand. Billing, data gathering and edge cases can be manual for a couple of months. Sell it at a lower price, fix automation later, and in demos lead with what works while showing a manual workaround for what doesn't. Fast still can't mean sloppy.

• What should an AI membership for small businesses include and cost?

  My proposal is two tiers. About $400 a month gets everything the team builds, a Slack or Discord for questions, two group office-hours calls a month and discounts on custom builds, nothing one-to-one. About $1,000 a month adds a monthly 90-minute call, reserved office hours and a quarterly audit. These are untested proposals.

• Why do software products fail even when the product is good?

  Usually it's sales, not product. My insurance software stalled because nobody spent 20 to 30 hours a week in front of buyers, and in a mature market an incomplete product loses to competitors' feature lists. Define the sales motion first: something that exists, in front of a buyer, at a defined price.

• Will AI-generated content make trust more valuable?

  Yes, that's my bet. When AI floods every channel and fakes look real, people stop trusting what they see on screens and fall back on what they know is real: people they've met, things seen in person, sources they already trust. Local, verifiable and human gets more valuable. It's a forecast, not a measured result.

• Will OpenAI and Anthropic just build whatever AI product I make?

  Not if you go narrow. The big labs won't build hyper-specific, multi-step workflows for a niche like local media because it isn't worth their effort, and that's where an independent builder has a moat. What will get wiped out is simple one-feature software, like a basic AI receptionist.