AI won't make you money — unless you settle these three things first
Buying the subscription is the easiest step, and the least useful one. The real gap is whether it's wired into your day.
Over the past year, nearly every business owner has asked me the same question: "Should we be using AI too?"
My answer usually disappoints them: yes, but not the way you're picturing. Buy a subscription, tell staff to "play with it when you have time," and three months later nothing will have changed except a new monthly bill.
1. Where you apply it matters more than which tool you pick
AI's value isn't evenly distributed. The same tool applied to the right process saves you a headcount; applied to the wrong one, it's just another toy.
There's a simple test for finding the entry point: repetitive, time-consuming, rule-bound work that still needs a bit of judgment. Common support replies, first-pass organization of large datasets, first drafts of content — that's the sweet spot.
Conversely, decisions that carry accountability and communication that depends on real relationships will only generate rework if you hand them over today.
2. It must live inside the workflow, not beside it
This is where most companies stall. If staff have to open another tab, copy data in, and paste results back out, that friction alone will kill the tool within two weeks.
Any tool that requires extra steps to use eventually goes unused.
What actually works: AI shows up where the work already happens. A form is submitted and the summary is already written. A customer gets in touch and their background is already assembled alongside. Users barely need to notice they're using AI at all.
3. Without your own knowledge, AI has no edge
Generic AI is available to everyone, which is exactly why it confers no advantage. The difference comes from what you feed it — your brand voice, your past cases, your product details, the mistakes you've already made.
When we implement AI for a brand, much of the work isn't technical at all. It's collecting knowledge scattered across employees' heads, old documents and chat histories, and shaping it into something AI can use. Only after that does AI go from "a generic assistant" to "someone who understands this business."
Back to the question
So "should we be using AI" is the wrong phrasing. The right one is: which of our processes is the most repetitive, the most time-consuming, and already has enough material to feed it?
Start there. Make it work. Then talk about the second one.
Frequently asked questions
How should a small business start adopting AI?
Start with one specific process rather than a company-wide rollout. The selection test: the task is repetitive, time-consuming and rule-bound, but still needs some judgment — common support replies, first-pass data organization, first drafts of content. Make that one process genuinely save time, then move to the next.
Why did our company buy AI tools that nobody uses?
Usually the friction is too high. If staff must open another tab, copy data in, and paste results back out, those extra steps alone will kill adoption within two weeks. What works is placing AI where the work already happens — a summary generated automatically when a form is submitted, for instance, so users barely notice they're using AI.
How much does AI adoption cost for a business?
Costs fall into three parts: tool subscriptions, integration and automation build-out, and knowledge organization. Subscriptions are usually the smallest piece; the real investment is shaping scattered data and processes into something AI can use. Starting with a single process significantly lowers the entry cost and validates the return sooner.
What can AI actually do for a business today?
The clearest wins right now: first-pass replies to common support questions, organizing and categorizing large datasets, first drafts of content, meeting notes and summaries, and preliminary data analysis. Decisions carrying accountability and communication built on real relationships are still not advisable to hand over.