How AI Agents Are Changing Workflows: What SMEs Should Start Doing Now in the Era of 15% Generative AI Adoption

“Isn’t AI ultimately just something for big companies to use?”

I heard this sentiment recently while speaking with a business owner. I understand how they feel. However, when you look at the actual data, a slightly different picture emerges.

According to a survey released by kubell in December 2025, the adoption rate of generative AI among small and medium-sized enterprises is 15.5%. While this is by no means a high figure, considering it was only a few percent two years ago, it is spreading quietly but steadily. A survey by Tokyo Shoko Research also shows that 25% of all companies are using generative AI in their business operations (Source: Tokyo Shoko Research, “Survey on the Actual Status of Generative AI Utilization”).

So, what are the remaining 75–85% of companies doing? And what kind of gap will emerge in the future between the 15% that started using it early and those that have not? Today, I would like to delve a little deeper into that question.

Why can’t we take the first step of “let’s try it”?

According to a survey by the Information and Communications Research Institute, the most common reason given by companies that do not use generative AI at all was “no use cases or scenarios (don’t know how to use it).” The reality is that only 19.8% of people use tools like ChatGPT on a daily basis, while 78.2% do not use them at all.

What do you think when you hear this? I suspect many of you are thinking, “That’s the same for us.”

However, I interpret this reason of “not knowing how to use it” a little differently. I believe the problem is not that people “don’t know how to use AI,” but rather that “the challenges in their own business operations have not been organized.”

Let me explain what I mean. AI will do work for you if you give it instructions, but to provide instructions, you need to have organized “what kind of work you are doing now, where time is being spent, and what you want to improve.” However, in many SMEs, this “organization of business operations” itself is being put on the back burner.

The daily work gets done. But in a state where you don’t really know what is taking how much time or who is doing what—you won’t know what to ask the AI to do either.

How “AI Agents” are changing work workflows

Entering 2026, I have had more opportunities to hear the term “AI agent (autonomous AI).” In a survey by Nikkei CrossTech, 29.7% of all Japanese companies responded that they are utilizing AI agents in their business operations.

An AI agent is slightly different from an “AI that answers questions” like ChatGPT. It is an autonomous AI that can judge situations on its own, gather necessary information, span multiple systems, and even execute the next action.

For example, you can use them in ways like this:

• Automating expense reimbursement: When you take a photo of a receipt, AI-OCR (technology that automatically converts paper documents into data using AI) reads the amount, date, and category, checks it against regulations, and creates a draft journal entry → the person in charge just needs to press the confirm button

• Primary automation of inquiry responses: The AI learns frequently asked questions and automatically generates initial replies for emails or inquiry forms → the person in charge checks the content and sends it

• Automatic generation of daily reports and meeting minutes: The AI transcribes Zoom recordings of meetings → automatically creates meeting minutes summarizing the key points and a list of next actions

Some of you might hear this and think, “That has nothing to do with us.” But I want you to pause and think: aren’t there tasks in your company where you repeat the same manual work every month?

End-of-month tabulations, email correspondence with business partners, creating estimates, compiling attendance records—these kinds of “routine tasks” are exactly what AI agents are best at.

SOMPO Japan is leveraging the AI agent platform "Heylix" to initiate efforts where frontline staff themselves design and automate their own work processes (Source: SOMPO Japan official announcement). You might think this is only possible for large corporations, but this platform is designed to be used without code (no programming knowledge required).

The first step is to "organize your work first"

As I work with SME owners at ShakeHands, I have been feeling something quite often lately. Before even discussing "wanting to introduce AI," the situation is almost always that "they don't really know who is doing what work or where tasks are concentrated."

The personalization of work—tasks that only certain people know about or procedures that only certain people can perform—if this is not resolved, introducing AI will not be effective. This is because, in order to entrust work to AI, the prerequisite is that "the procedures for that work must be verbalized."

"Personalization of work and lack of standardized procedures" accounts for 24.7% of the obstacles hindering productivity improvement in SMEs, according to a survey by kubell. Furthermore, "labor shortages and recruitment difficulties" at 41.0% is the top issue, but in fact, one solution to labor shortages is to "organize work and entrust it to AI."

In terms of order, this is what I would like to propose.

1. First, write down "what you and your employees do every day" (visualizing work)

2. From that list, identify "tasks that are the same every time" and "tasks that yield the same result regardless of who does them"

3. Try using AI from there

I believe this initial "visualization of work" is the true starting point for successful AI utilization.

Summary

The adoption rate of generative AI among SMEs is 15.5%. Conversely, 85% of companies have not yet fully utilized it. This is both a challenge and a great opportunity.

In the era of AI agents, companies that start using them early can dedicate the time freed up from routine tasks to sales, strategy, and customer service. On the other hand, if you fall behind, the gap will only continue to widen amidst recruitment difficulties and labor shortages.

"To master AI, first organize your work"—I believe this step will change the feel of management much more than you might think.

If you feel like "I don't know where to start organizing our work," please feel free to reach out. At ShakeHands, we provide support tailored to SME owners, from visualizing work to AI training and tool utilization support.

---

[Reference Data Sources]

kubell "Survey on the State of Digitalization in SMEs" (December 2025): https://www.kubell.com/news/2025/12/research.html
Tokyo Shoko Research "Survey on the Actual State of Generative AI Utilization": https://www.tsr-net.co.jp/data/detail/1201667_1527.html
Information and Communications Research Institute "Current Status and Prospects of Generative AI Introduction in Companies": https://www.icr.co.jp/publicity/5325.html
Nikkei Cross Tech "Introduction Rate of Generative AI Tools in Japanese Companies": https://xtech.nikkei.com/atcl/nxt/column/18/03314/090800004/

ShakeHands LLC

A business partner that organizes labor shortages from the structure
You are short on people.
But "recruitment" is not always the right answer.
ShakeHands LLC is a partner that faces the labor shortages and business expansion of SMEs,
organizes the options of recruitment, outsourcing, and AI utilization,
and designs the optimal approach.
Rather than proposals for the sake of selling, we think together about choices as management decisions.