I help companies with digital transformation, and AI is one of the tools I reach for regularly. By all rights, I should be out here telling you how powerful AI is and why your company desperately needs it.
But I'm going to tell you the truth instead: your problem might not need AI.
A Real Story
A client came to me wanting an "AI-powered smart customer service" system.
I asked how their current support situation looked. They said they got about 30 customer messages a day, handled by two people, and messages were often missed or responded to too slowly.
Thirty messages. Two people.
I did the math: even at 5 minutes per message, that's 150 minutes a day — two and a half hours. The problem wasn't volume. It was that there was no single place to manage these messages.
What I ended up building: a LINE official account connected to a simple ticketing system so messages wouldn't slip through the cracks and every reply would be on record.
No AI. Cost less than a tenth of the original budget. Problem solved.
The client told me afterward: "Thank god you talked me out of spending that money on AI customer service."
When You Don't Need AI
From my experience, these situations usually don't call for AI:
1. Your problem is a process problem, not an intelligence problem.
If your pain point is "we can't find anything," "data is scattered everywhere," or "nobody knows where things stand" — what you need is a good system, not AI. Get the process sorted out and the data centralized, and you've already solved most of the problem.
2. Your data volume is too small.
AI learns from data. If you're processing a few hundred transactions or a few dozen customers per month, there's a limit to what AI can do for you. That's not AI's failing — your business just hasn't reached that scale yet.
At low volumes, human judgment is often more accurate, faster, and cheaper than AI.
3. Your problem can be solved with rules.
"Automatically remind us about orders with no response for over three days." "Auto-reorder when inventory drops below 10 units." These aren't AI — they're if-else logic. A simple automation script handles it, no machine learning required.
A lot of products marketed as "AI solutions" are just a pile of if-else statements behind a polished interface.
When AI Actually Makes Sense
On the flip side, these situations are worth serious consideration:
1. You have large volumes of repetitive work that requires judgment.
Reviewing hundreds of medical images, thousands of customer feedback submissions, or tens of thousands of transaction records per day — humans can't keep up, and they get tired and miss things. AI doesn't get tired.
2. Your rules are too complex to write out as if-else.
"Which surgical approach is right for this patient?" "Is the sentiment in this piece of text positive or negative?" These problems can't be reduced to simple rules — they require AI to learn judgment from large amounts of examples.
3. Humans could do it, but the time cost is prohibitive.
Transcribing meeting recordings into structured notes with summaries, translating a long document into five languages, screening hundreds of resumes against a set of criteria — a person can do all of these. But the time cost makes it impractical.
A Simple Decision Framework
Next time someone brings up AI, run it through these three questions:
| Question | If "No" | If "Yes" |
|---|---|---|
| Is my process already digitized? | Digitize first | AI is worth considering |
| Is there high repetition in this task? | Probably don't need AI | AI can help |
| Can rules solve it? | Consider AI | Write the automation rules first |
If all three answers point toward "consider AI," then it's worth a serious evaluation.
If not — congratulations. You just saved yourself some money.
Why Would I Talk You Out of It?
You might be wondering: you work in digital transformation, so why would you tell someone not to build? Isn't that cutting off your own income?
Simple: I live on referrals.
If I convince you to spend a hundred thousand on a system you don't need, you're not going to recommend me to anyone. But if I help you solve the real problem with a tenth of that budget, the next time you have a genuine need I'm the first person you'll call.
And you'll tell people: "There's this person who could have charged me way more, but chose to tell me the truth instead."
That's the reputation I'm building.
So, Does Your Problem Need AI?
If you're not sure, here's what I'd suggest:
Write down your three most painful workflows. For each one, write three lines: how it's handled now, how long it takes, and how good the results are.
Read back what you wrote. Usually the answer becomes obvious. Some things need AI. Some just need a decent system. Some just need the process tidied up.
If you're still not sure after that, find someone who's done this before and have them look at it with you for an hour. An hour of honest advice might save you six months of going in the wrong direction.
I'm Young. I've worked on over a dozen digital transformation and AI projects across industries. My job isn't to sell you technology — it's to help you figure out what you actually need. Sometimes that's AI. Sometimes it's a good system. Sometimes it's just fixing the workflow.