Conclusion

To put it simply, the “cost of software usage” is beginning to collapse.

SaaS monthly fees continue to rise, AI agents are moving towards being free, and build times have shrunk from 90 seconds to 5 seconds. Local small and medium-sized enterprises (SMEs) should seriously consider the implications of these three simultaneous developments.

Why? Large corporations take years to change their cost structures due to approvals, transitions, and adjustments. SMEs can make changes starting next week. This “agility” is becoming a weapon at this very moment.

—

SaaS Has Shifted from “Convenience” to “Burden”

The essence of SaaS was a model where you paid a monthly fee for the convenience of not having to build things yourself. But what is the reality?

• Numerous unused features are included, and monthly fees rise year after year.
• Each tool costs between 10,000 to 30,000 yen per month. If you use five tools, it exceeds 100,000 yen per month.
• Struggles with tool integration leave manual tasks in Excel.
• Even when price increase notifications arrive, data becomes hostage, making it impossible to escape.

For a small business with 30 employees, implementing 5-6 SaaS tools for CRM, project management, chat, document management, and accounting can cost between 2 million to 4 million yen annually. Is this a fair price for “convenience”?

I want to ask this: What percentage of that SaaS do you actually use?

In many cases, the answer is less than 20%. You continue to pay full price for 80% of unused features. This is a structural problem with SaaS.

—

An Abnormal Situation Where AI Agents Are Offered at “Zero Profit Margin”

From late 2024 to 2025, something unusual is happening in the AI agent market.

Players like OpenAI, Google, and Anthropic have started offering agent functionalities at effectively free or ultra-low prices. Even the free plan of ChatGPT can accomplish a lot, and Google’s Gemini has a generous free tier. Models like Claude and GPT-4o can be accessed via API for just a few yen per use.

Why is this happening? The answer is simple: the battle for platform supremacy. To capture users, they are distributing AI functionalities without regard for profit. This is similar to how Uber continued to grow its passenger base despite operating at a loss.

Who benefits the most from this “zero profit margin competition”? It’s not large corporations. They are already paying high amounts for enterprise contracts. The beneficiaries are small and medium-sized enterprises that have not been able to invest in AI until now.

What can they do specifically?

• Automating customer inquiries: By having an AI agent read FAQs and internal manuals, 70% of customer interactions can be handled automatically, saving the cost of one part-time employee (150,000 to 200,000 yen).
• Automatically generating minutes and reports: By simply recording meetings, summaries, task extraction, and minutes can be created in minutes, eliminating 3 hours of administrative work per week.
• Data analysis: By uploading sales CSVs and asking, “What product categories are declining compared to last year?”, answers can be returned in 30 seconds. This analysis, which would have cost 100,000 yen if outsourced, is now available.

These services can be obtained for a few thousand yen per month or even for free. The cost structure is changing dramatically.

—

The Implications of Build Time Reducing from 90 Seconds to 5 Seconds

The announcement of “Turborepo 2.5” by Vercel has drastically reduced build times from 90 seconds to 5 seconds. While this may seem like news for developers, the essence is different.

The “cost of creating software” has dramatically decreased.

A reduction in build time to 1/18 means that the cycle of trial and error speeds up 18 times. Developers who could try things 5 times a day can now do so 90 times. This directly correlates to the speed of prototype completion.

Furthermore, with the evolution of AI code generation tools like Cursor, Replit Agent, and GitHub Copilot, a new world is emerging:

• Business applications that once cost 3 million yen to outsource can now be built for 50,000 to 200,000 yen using AI code generation and fast build environments.
• Development periods can be shortened from 3 months to 2 weeks.
• “Let’s build it and try it out” becomes feasible at a realistic cost.

Until now, it was financially unfeasible for SMEs to develop software in-house. However, that premise is now beginning to crumble.

—

A Concrete Example of How “300,000 Yen SaaS” Becomes “50,000 Yen In-House Build”

Let’s consider a local manufacturing company with 50 employees.

Traditional SaaS Configuration (Estimated Monthly Cost)

• CRM: 30,000 yen
• Project Management: 20,000 yen
• Chat Tool: 15,000 yen
• Document Management: 10,000 yen
• Accounting Software: 15,000 yen
• Other Integration Tools: 10,000 yen
• Total: 100,000 to 150,000 yen (1.2 to 1.8 million yen annually)

AI Agent + Composable Configuration (Estimated Monthly Cost)

• Notion (Document + Project Management): 20,000 yen
• AI Agent (Inquiry, Minutes, Analysis): 5,000 yen or free
• In-House Developed Simple CRM (Built with Cursor + Supabase): Server cost 3,000 yen
• Chat (Discord or Slack Free Plan): Free
• Accounting Software (like freee): 3,000 yen
• Total: 30,000 to 50,000 yen (360,000 to 600,000 yen annually)

Annual cost savings of 800,000 to 1.2 million yen. Moreover, you get a system perfectly tailored to your business workflow without paying for the “80% of unused features” in SaaS.

The initial setup cost may be around 200,000 to 500,000 yen, but it can be recouped within six months.

—

What Fruits Can Only Small and Medium-Sized Enterprises Harvest?

This is the crux of the matter. Why is it that only “small and medium-sized enterprises” can benefit?

The reasons large corporations cannot ride this wave are clear:

1. Compatibility with Existing Systems: They have invested tens of millions to hundreds of millions of yen in ERP and Salesforce. They cannot easily discard these systems.
2. Security Reviews: Implementing a new AI tool can take three months for approval from the IT department.
3. Organizational Inertia: Resistance from the field that “things are working as they are now.”
4. Vendor Lock-In: Long-term contracts prevent flexibility.

On the other hand, what about small and medium-sized enterprises?

• If the president says, “Let’s use this starting next week,” they can start using it next week.
• Their investment in existing systems is small, so the switching costs are also low.
• The distance between the field and management is close, so when something is deemed “convenient,” it quickly spreads throughout the company.
• They don’t need to strive for perfection. They can make decisions based on the judgment that it’s okay to operate at 70% efficiency.

The speed of decision-making and agility. This is the greatest weapon of small and medium-sized enterprises.

When technology costs decrease, the first beneficiaries are “organizations that can act quickly.” While large corporations are still in deliberation meetings, SMEs can complete implementation and achieve results.

—

So, What Should Be Done?

I will mention three things.

1. Take Inventory of Your Current SaaS Usage

Create a list of all the SaaS used across the company, aligning “the features actually used” with “monthly costs.” Visualizing the amount spent on unused features will provide a reason to take action.

2. Introduce One AI Agent into Your Operations Starting Today

You don’t need to implement it company-wide right away. First, the president can try using ChatGPT or Claude to “summarize today’s meeting minutes” or “analyze the trends in this sales data.” Experiencing it firsthand will reveal what changes can be made.

3. Keep the Option to “Build It Yourself”

With AI code generation tools like Cursor, even those with little programming experience can create simple business applications. Change the mindset that outsourcing is the only option. Start by replacing one of the “things being awkwardly managed in Excel” with a simple app.

—

This Trend Is Irreversible

SaaS prices continue to rise, while AI prices continue to fall. Build times keep decreasing, and the barriers to creating software are lowering.

This structural change is not a temporary trend. It is irreversible.

In three years, being in a state of “still contracting five SaaS at 100,000 yen per month” may sound as outdated as “still receiving orders via fax.”

If you are going to act, now is the time. While large corporations are unable to move, small and medium-sized enterprises can seize the fruits first. I hope you will start experimenting with restructuring your business systems for 50,000 yen per month starting next week.