It might be the beginning of the AI Manager.
An MCP integration feature that connects AI agents to the PC log area has been released by "Mieru Cloud."
I believe this news is a major turning point. It feels like something I vaguely thought would take another two years has suddenly appeared before my eyes.
Accounting firms are made of "invisible work"
The work of an accounting firm is strange. The deliverables are visible: tax returns, financial statements, and trial balances. However, almost nothing about what happened until those deliverables were completed is recorded, other than the individual's self-managed information.
- Requesting documents from clients
- Internal document organization
- Information gathering and legal research
- Internal consultations
- Reviews
- Knowledge sharing
- Improving work tools like Excel
- DX initiatives and experimental AI use
These tasks do not lead directly to deliverables. Even if indirect operations lead to results, they are difficult to evaluate. However, in reality, it is this invisible work that supports the firm. What would happen if AI could understand all of that?
The era of "how many hours worked" is ending
Until now, management has centered on "how many hours were worked." But that's not what we really want to know. We want to know "why it took so long."
For example, even if monthly processing took 5 hours, the solution changes completely depending on whether the input was slow, document collection was delayed, client interaction was high, or the person was unfamiliar with Excel. When this part is visualized, what we used to think of as "it seems somehow difficult" can be clearly articulated as "this is the bottleneck."
This isn't just efficiency; it changes the understanding of the business itself. If customer and sales information is already accumulated in a CRM, the world expands even further. Personally, I believe the use of AI for big data—specifically behavioral information—is the final piece and the core. There is only so much you can do with PC log information alone.
But when business information and behavioral information are connected—through kintone as a CRM with built-in workflows, Notion with its database construction and AI, or even Google AppSheet—the view changes. Customer info, project info, sales info, man-hour info, and behavioral logs all connect. Then, AI can explain not just "which clients have low profit margins," but "why they are low" and even find commonalities among high-margin clients. Analysis shifts from looking at the past to creating the future. This is truly exciting.
Will AI replace managers?
I think the answer is NO. However, the work of managers will change drastically. Much of a manager's job consists of status checks, progress checks, man-hour checks, and delay checks. A significant amount of time is spent just "grasping the situation."
But what if AI grasped everything? When you arrive at work in the morning, the AI tells you: "This project is at risk this week," "This client is exceeding man-hours," "This person's workload is getting too high," or "This team's productivity is dropping." If you provide response patterns in advance, it could even think autonomously. Managers will be freed from verification work and can focus on what they should be doing: decision-making, training, and communication. This isn't about eliminating managers; it's about evolving them.
Personalized education for everyone begins
This is particularly interesting. Current training uses the same materials, content, and order for everyone. But in reality, weaknesses vary by person: some struggle with Excel, others with cloud accounting, reviews, client service, or AI utilization. Ideally, everyone should have a different learning menu. PC log utilization can visualize this. By looking at behavior from logs, AI can find weaknesses and suggest what to learn next. Education evolves from standardization to individual optimization.
The most interesting part: "Business Health Checks"
This is what I expect the most. Issues like turnover, quality decline, increased overtime, and client complaints seem to happen suddenly, but they always have precursors: decreased communication, skewed man-hours, increased corrections, or slower responses. These are invisible to humans, or noticed too late, because we could only evaluate behavior through subjective data like interviews or daily reports. But by connecting to behavioral information in real-time, we can objectively visualize the state of the organization—just like a health checkup. We can observe the firm's state at the level of "blood pressure rose" or "sleep decreased."
I don't think it's wise to use this technology for negative purposes like surveillance. The use of behavioral information requires ethics from management. Future accounting firms might notice problems before they occur rather than responding after the fact.
AI-powered management assistance: bookkeeping is just one part of the business, but management moves the entire organization. The impact of increasing the decision-making speed of the whole organization is greater than one person working faster. At the center of this is the use of behavioral history, CRM construction, and communication data. When AI connects these, it is no longer just a tool, assistant, or agent. It is a higher layer. I want to call this the AI COO (Chief Operating Officer).
Recruitment, placement, education, profit management, business design, and quality control—an entity that supports these cross-functionally. This is EBPM (Evidence-Based Policy Management) in its true sense, even for small businesses. To create an environment where people can focus on more creative decision-making. The entrance to such a future might unexpectedly be starting from the field of "PC log analysis." This technology innovation makes me feel a certain future is coming.
