We sell training in this field, so discount accordingly. The argument below is laid out so you can check it against what you observe rather than take it on trust.
What AI is genuinely making easier
Start with the honest part, because pretending otherwise would be silly.
- Building a workflow from a description. Increasingly you can say what you want and get most of a working automation.
- Writing the copy inside it: the replies, the follow-up sequences, the reminders.
- Debugging. Paste an error, get a plausible cause, far faster than searching.
- Learning. The gap between not knowing a tool and being functional in it has collapsed.
- Connecting things without a ready-made integration, because the code you occasionally need is easier to produce.
That is a real list and it is getting longer. Anyone whose entire value is operating the software should take it seriously.
What it is not touching
Now the part that matters, because this is where the money actually sits.
Deciding what to build
A clinic owner cannot describe the system they need. They can describe a feeling: the phone is busy, we seem to lose people, I am tired. Turning that into a specific build requires sitting in their business, noticing what they do not mention, and knowing which of five possible systems they will actually use.
AI can build what it is told. Somebody has to know what to tell it, and that somebody has to have understood a real business first.
Being accountable
At 11pm on a Saturday the booking system stops taking appointments. Who does the owner call?
This is the part people underrate most. A large share of what a client pays monthly is not for the software. It is for a specific person being responsible when it fails. That cannot be automated because it is not a task, it is a relationship with an obligation inside it.
Trust, which is the actual product
Small business owners have been burned. They have paid for websites that did nothing and marketing that produced nothing. They do not buy the best system, they buy from the person they believe will not disappear.
No model shortens that. It is earned by turning up repeatedly over time.
How the job actually changes
Not disappearance. A shift in where the value sits, and it has a direction you can prepare for.
| Getting less valuable | Getting more valuable | |
|---|---|---|
| Skill | Operating the tools | Knowing what a specific business needs |
| Output | Building a workflow | Owning an outcome over time |
| Position | Implementer taking instructions | Advisor who decides and is accountable |
| Moat | Knowing more features | Knowing one industry deeply |
| Defence | None, this is what AI does | Relationships and reliability |
That table is also a plan. Move deliberately along it and the changes work for you, because cheaper building means you can serve more clients with the same hours.
The version of this risk that is real
Being honest about what could actually go wrong.
The commodity end will keep compressing. Simple builds, generic setups, anything a business could reasonably do themselves with a good tool and an afternoon: that work will be worth less each year, and people whose whole offer sits there will feel it.
So the risk is not unemployment. It is that the easy version stops paying while the hard version pays more. Which is how most skilled trades have always worked, and it is a reason to get good rather than a reason to stay out.
What to do about it
- 01Go deep on one industry rather than broad on tools. Industry knowledge is the part that does not commoditise.
- 02Use AI aggressively in your own delivery. If it makes building faster, that is your margin, not your threat.
- 03Sell outcomes and ongoing responsibility, not builds. Recurring relationships survive tool changes.
- 04Be the person who answers. Reliability is unglamorous and it is the most durable advantage available.
- 05Keep one platform-independent skill, so a single vendor's decisions cannot end your business.
The honest bottom line
If your plan is to learn a tool and sell tool operation, that plan has a shortening shelf life and you should know it.
If your plan is to understand a type of business, decide what it needs, build it with whatever tools exist that year, and be the person accountable for it, then AI getting better is straightforwardly good for you. It makes your delivery cheaper while the part clients pay for stays scarce.