Conclusion First: The “Price Destruction” of Core Systems Has Begun

The cost of introducing core systems for small and medium enterprises typically ranges from 3 million to 10 million yen. Additionally, annual maintenance and operation costs can run into hundreds of thousands to millions of yen.

This structure is now on the verge of fundamental collapse.

AI-native ERP “Zynthoro” integrates 15 types of tools, including accounting, inventory, order management, CRM, and HR, into a single platform, offering it for a monthly fee of just a few tens of thousands of yen. The initial cost is virtually zero. The shift from 3 million yen to 50,000 yen a month is not merely a matter of being “cheaper”; it signifies that the rules of the game have changed.

In this article, we will explore how the emergence of Zynthoro, along with the latest paper “Agentic ERP” that outlines the concept of agent-based ERP, and the research on BatchDAG, which accelerates data analysis by up to 47 times, will shape the future of core systems for small and medium enterprises.

—

What Has Changed—The Shock of “15 Tools Integrated × Monthly Fee of a Few Tens of Thousands of Yen”

First, let’s clarify the current situation.

Traditionally, when small and medium enterprises attempted to digitize their core operations, it looked like this:

• Accounting Software: 5,000 to 15,000 yen per month
• Inventory Management: 10,000 to 30,000 yen per month
• Order Management: 10,000 to 50,000 yen per month
• CRM: 5,000 to 20,000 yen per person per month
• Attendance and HR: 5,000 to 10,000 yen per month
• Others (Project Management, Invoicing, Workflow, etc.)

If contracted separately, the total monthly cost could range from 100,000 to 200,000 yen. Moreover, the integration between tools often required manual work. Data would be copied into Excel, double-checked, mistakes would occur, and then corrections would be made. As the number of tools increased, the costs associated with the “connections” would balloon.

Zynthoro has consolidated all of this into one solution. It integrates the functionalities equivalent to 15 tools in an AI-native format for a monthly fee of just a few tens of thousands of yen. What’s crucial is not just that it is “cheap”; it’s that the connections between tools disappear.

When order data is entered, inventory is adjusted, invoices are generated, and accounting is updated—all within a single system. There’s no need for data entry, double-checking, or searching for “where did that data go?”.

For small and medium enterprises, the biggest cost is not actually the monthly software fees. It is the time people spend running between tools. Integrated AI-ERPs like Zynthoro drastically reduce this.

—

The Next Vision Presented by the “Agentic ERP” Paper—ERP That “Thinks and Acts on Its Own”

If Zynthoro disrupts current norms with its “integration and low cost,” the research on “Agentic ERP” points to an even further future.

This paper proposes an architecture where specialized LLM agents autonomously manage ERP operations based on their roles.

Specifically, it entails the following:

• A “Purchasing Agent” determines the timing and quantity of orders
• An “Accounting Agent” handles journal entries and detects anomalies
• A “Sales Agent” generates proposals from customer data
• These agents collaborate with each other, allowing humans to only handle “approvals”

Traditional ERP systems were merely “boxes for inputting data.” Humans input, verify, and make decisions. ERP systems only recorded information.

Agent-based ERP is different. The ERP thinks, proposes, and executes on its own. Human roles are narrowed down to “final judgment” and “exception handling.”

What does this mean for small and medium enterprises?

A world where “there is only one accountant” or “the president also handles sales and accounting” can still have agents managing operations becomes visible.

Large corporations can assign specialized personnel to each department. Small and medium enterprises cannot do that. Therefore, the benefits of agent-based ERP are even greater for small and medium enterprises. This presents a reversal in structure.

—

BatchDAG—What the “47 Times Faster Data Analysis” Means

Another noteworthy approach is BatchDAG.

This method structures operational procedures generated by LLMs as a DAG (Directed Acyclic Graph), enabling batch processing of data analysis all at once. Research has reported achieving up to 47 times processing efficiency compared to traditional sequential LLM analysis.

Forty-seven times. What does this mean?

For instance, if monthly sales analysis previously took two hours, it can now be completed in 2 to 3 minutes.

Consider what happens in the field of small and medium enterprises. When asked, “Please provide last month’s sales by product, region, and customer,” one would have to repeatedly adjust Excel pivot tables, create graphs, and paste them into PowerPoint. That’s half a day’s work.

If BatchDAG technology is integrated into ERP, simply giving a natural language command like “Analyze last month’s sales” would yield analysis results from multiple perspectives—at 47 times the speed.

As the cost of analysis approaches zero, the very decision of whether to analyze or not disappears. Just do everything. Analyses that were previously abandoned due to lack of time will run automatically.

This technology is particularly effective for small and medium enterprises that cannot afford specialized analysis teams.

—

So, What Should We Do?

Let’s summarize the discussion so far.

Three technological trends are moving simultaneously:

1. Integrated AI-ERP (like Zynthoro) → Destruction of introduction costs
2. Agent-based ERP → Destruction of operational costs (labor costs)
3. High-speed analysis like BatchDAG → Destruction of analysis costs

When these three converge, the core operations of small and medium enterprises will fundamentally change.

So, what should you do right now?

First, visualize your company’s “connection costs.”

How many types of tools are currently in use, and how many hours are spent each month on data entry and transfer between tools? By applying hourly wages, the “invisible costs” become quantifiable. For example, 20 hours a month at 2,000 yen per hour equals 40,000 yen. This continues every month for years.

Next, experiment on a small scale with integrated AI-ERP like Zynthoro. Start with one department or one operation, not a full-scale implementation. “Just try it out” is the most important step.

Agent-based ERP and BatchDAG are still in many cases at the research stage. However, the direction is clear. ERP is transforming from a “data input tool” to an “autonomous partner.” Whether you are prepared to ride this wave will determine the landscape three years from now.

—

The Real Question Lies Here

“Core systems costing 50,000 yen a month”—this is not just a story about cost reduction.

Companies that previously could not implement core systems will now be able to do so. Data that could not be analyzed before will become analyzable. Operations that could not be managed without hiring staff will now be managed by agents.

What will happen with reduced costs is that the boundary between “capable companies” and “incapable companies” will disappear.

Weapons that only large corporations could possess will now be available for 50,000 yen a month. At that point, what will determine the competitiveness of small and medium enterprises?

The answer is simple. “Whether to use it or not.” That’s all.

In a world where technology has become cheaper, whether or not you possess technology does not create a difference. What you do with it and how quickly you act will create the distinction.

The strength of small and medium enterprises lies in their speed of decision-making. There’s no need to pass through five levels of approval. If the president says, “Let’s do it,” it can start tomorrow.

Now is the time to leverage that strength.