Doing it yourself is a real option. Here is how to tell when it stops being the smart one.
Let’s start with the obvious. You can figure out AI yourself.
The tools are accessible. The tutorials are everywhere. Many owners are already experimenting, and some of them are getting useful results. If you’re one of them, good. Keep going.
So why would anyone bring in help? That’s a fair question. It deserves a straight answer, not a sales pitch.
Doing it yourself is a real strategy
I’d rather say this plainly than pretend otherwise. For a lot of small businesses, DIY is the main alternative to hiring anyone. It costs little cash. It builds your own understanding. And nobody knows your business better than you do.
If you have a clear problem, some time, and a willingness to experiment, starting on your own can be a perfectly good move.
Where it tends to get stuck
In my view (this is an observation, not a study), the trouble rarely starts with the tools. It starts with everything around them.
Time is the first one. You run the business. Learning a new set of tools is the thing that slips when a customer calls.
Then there’s the choice of problem. It’s easy to try AI on whatever is closest. It’s harder to find the work where a change would matter most.
Fixing one part of a business changes something else. Sometimes that’s the problem. Sometimes it’s where the opportunity was hiding.
And finally, the follow-through. A clever experiment is not the same as a way of working your team actually uses.
Having access to AI isn’t the same as knowing what to do with it. Access is the easy part now.
What the learning curve looked like for me
I’ve used AI hundreds of times for sales and marketing work. Over four years.
The first time I did anything, I got a lot of it wrong. A lot.
Here’s what I figured out. We’re coding in English. And not all developers are the same.
Fast forward to today. I’m about 100 times faster at those same things. Not 2X. Not 10X. 100X. (That’s my own experience, not a benchmark.) It’s hard to imagine until you’ve done it.
That curve, from clumsy to fast, is real. You can climb it on your own. It takes a lot of repetitions. And it’s easier with someone who has already climbed it.
A few questions worth asking yourself
Not sure where you stand? Try these:
- Do I know which one problem would be worth the most to solve?
- Do I have the time to learn this properly, or am I fitting it between other things?
- If I build something, who checks the results and keeps it working?
- How will I know whether it actually helped?
If you have good answers, you may not need me. Really.
If a few of them make you pause, that’s useful information too. It doesn’t mean you did anything wrong. It may just mean the problem is bigger than a spare hour on a Sunday.
What help can look like
Help doesn’t have to mean handing the business over. The way I work, it’s the opposite.
We pick one real problem and solve it together. We measure what changed. You keep the skills and the tools we build, so the next improvement is easier for you to make. And then you decide whether we do anything more. Or nothing.
I come to this with more than 20 years of building and running businesses, and from using AI on my own work. But I’d rather you judge that by the result than by the résumé.
So, which is it?
Both can be right at different times. Start on your own if you can. Bring in help if you get stuck, or if the stakes are high enough that you’d rather not learn on the job.
If you’d like to think it through with someone, book a 30-minute, no-pitch call. We’ll talk about where your business feels slow, costly or stuck. If you don’t need me, I’ll say so.
If this was useful, you may also like Small Business AI Adoption: Started Isn’t Finished.
Appendix: Sources and claims
| Claim | Source | Status |
|---|---|---|
| DIY or internal implementation is a major alternative for small businesses | Sales and marketing, competitive analysis (Input 3) | Interpretation (not independently verified) |
| Tools and tutorials are widely accessible; many owners are experimenting | Sales and marketing, market trend priority 2; persona description | Interpretation (not independently verified) |
| Where DIY gets stuck: time, choosing the problem, follow-through | Adam’s observation, drawn from the buyer and problem definitions | Interpretation, not a study |
| Access to AI is not the same as knowing what to do with it | Sales and marketing, Story | Adam’s position |
| Four self-check questions | Written for this article | Recommendation |
| Engagement approach: one real problem, measured, skills retained, client decides next | Sales and marketing, buyer journey; business model | Adam’s description of his practice |
| 20+ years building and running businesses; AI used on own work | Sales and marketing, Credibility | Adam’s own experience |
| Hundreds of AI uses over 4 years in sales and marketing; many mistakes at first; now about 100X faster | Adam, supplied 2026-11-09 | Adam’s own experience (self-reported, not measured) |
| “We’re coding in English. Not all developers are the same.” | Adam, supplied 2026-11-09 | Adam’s own words (analogy) |