In the age of AI, news floods every channel — often bringing information anxiety, and leaving many wondering whether this is all just an overhyped bubble.
That skepticism is valid. New technologies always come with bubbles. But the scale and scope of what AI can change has already far outpaced any previous wave of technology.
After a brief period watching from the sidelines, I quickly came to believe: those who master AI — whether individuals or companies — will be the ones who hold their edge in the next era.
That conviction led me to leave a stable job last year, build products with AI, and explore how AI can actually land in the real world.
What I found is that the hard part of adopting AI isn't technology — AI's capabilities have already surpassed most engineers. The hard part is clarifying your own workflows: which decisions should stay with you, and which repetitive tasks can AI take over.
Here are three angles to help you take that first step — approaches I've seen work repeatedly while helping companies evaluate where to begin.
How widespread is AI, and how much does it actually affect small businesses?
Two cases from Japan in the past six months have stuck with me.
One is Kobo, an office supplies management company that started using AI to write blog posts. Each post used to take three hours. Now it takes twenty minutes.
The other is tax accountant Kento Hatakeyama. Zero employees at his firm — yet using Claude Code, he handles the books for 60 advisory clients single-handedly. 130 invoices, ¥30 million in monthly revenue data, processed in fifteen minutes.
Neither is a tech company. Just ordinary small businesses with more work than people.
In Taiwan, only 7.4% of SMBs have adopted or are planning to adopt AI, according to the 2025 SME White Paper published by the Ministry of Economic Affairs.
Goldman Sachs research: employees using AI save an average of 40 to 60 minutes per day. For a 50-person company, that's 30 to 50 extra hours freed up every single day.
PwC's 2026 study: 74% of the economic value generated by AI flows to the first 20% of companies that acted. Not the highest budgets. Not the largest companies. The ones who started first.
Given all these benefits, should we roll AI out across the entire company right away?
AI may look all-powerful, but precisely because it touches so many areas, it tends to create friction with existing systems and workflows.
And like people, AI occasionally makes mistakes. ERP systems are expensive and clunky but relatively reliable — handing everything to AI raises legitimate questions about whether employees might gradually lose their own judgment.
That's why, when starting with AI, the counterintuitive move is to start where it's easiest — routine tasks that follow roughly the same pattern every time.
Kobo chose blog posts not because that was their most important work, but because the task had the right characteristics: consistent format, visible output, no cross-department coordination needed. Getting it working showed them where to go next.
Same for Hatakeyama. He didn't set out to AI-ify his entire firm. He started with bookkeeping entries — the most regular, most repetitive category.
Getting free from repetitive, tedious tasks does more than improve efficiency. It creates space to think about what actually matters.
Can you give me something more concrete — something that shows me what AI can actually do?
Here are two situations most companies face.
The first is onboarding new hires. Day one, something comes up — where do they go for answers? Usually they ask a colleague, dig through old emails, or give up if they can't find anything. Every time, it depends on someone remembering and passing it along. Slow.
The second is information alignment. A lot of company meetings exist just to make sure everyone knows the same thing. That kind of meeting is really a symptom: information scattered around, never organized.
Both problems are ones AI document organization can address. You don't need complete SOPs. Just take the things "everyone knows" and have AI organize them — what questions always come up before a quote, which clients you typically don't work with, how you usually handle a given situation. Build it up, and new hires have somewhere to look, there are summaries before meetings, and knowledge stops living only inside someone's head.
What's the most concrete way to start?
In your daily work, whenever you catch yourself thinking "this isn't worth my time" — write it down. Look back in a week or two. Your starting point is probably somewhere in that list.
When asking employees "where are the problems?" you usually don't get much. Nobody wants to say their work has problems in front of their boss. Try these instead:
- "If the company covered your AI subscription, what would you use it for first?"
- "What step is hardest to explain to a new hire?"
- "What questions from clients make you spend the most time looking things up?"
AI doesn't replace people — it amplifies them. Find the first thing worth amplifying, and start there.
Related solutions
If document organization resonates, see how we approach this with Zentropy, our AI knowledge management solution.
References
If you've been thinking about this
No software to sell, no package to push. Just a conversation about where you are, and whether there's something I can help with.
Talk to Barry