Arrange by role
Role-bound agents
Market, Pricing, Sales, Operations, Implementation, Audit. AI will be deployed by role in the value chain, not by department.
A Keynote — 2026
01 / 14
No. 02 — Vision · Enterprise Agent Mesh / GPIE
Yaoyorozu's AI is
Stay.
The next competitive advantage will not be determined by the number of AI tools introduced. The company that will win will be the one that has AI agents in every aspect of management decisions, business processes, data, implementation, and customer contact points, and simultaneously improves decision-making speed and execution quality.
The Age of Enterprise Agent Mesh.
CoverGPIE
Chapter I — The Continuation02 / 14
§ 01
The issues of the AI era are:
Management OS instead of installationmoved to
And the answer I arrived at was
It wasn't about whether or not to use AI.
AI will change the role of people. The way organizations are organized will change. The distance between judgment and implementation changes. Up to this point, it's already a premise.
Last time, we looked at the entrance to the AI era. Check the previous
Where does enterprise value move?
Value moves because
decision-making centerIt is.
— In No.02, we will turn the abstract theory of the AI era into corporate value. AI will not be a time-saving tool for individuals, but will become a management OS that drives sales, gross profit, investment decisions, and risk management.
No.01 to No.02 · From Era to StructureGPIE
Chapter I — The Continuation03 / 14
§ 02
Winning companies will use AI
A group, not a monolithDesign as.
Corporate value cannot be determined by a single chat AI. AI that reads the market, AI that reads prices, AI that reads customers, AI that writes code, AI that audits, and AI that returns improvements. A bundle of roles changes the speed of management.
stay in one's work
Embedded operating layer
Not outside of Slack or ChatGPT. Place AI inside conference bodies, DB, BI, CRM, git, and operation logs.
Enterprise
Agent Mesh
Distributed AI as a moat
The essence of an AI Native company is that hundreds of AIs work together in a trail, accumulating company-specific learning assets.
One AI to Many AgentsGPIE
Chapter I — The Continuation04 / 14
§ 03
What is Yaoyorozu?
While dispersedthe whole thing movesIt's a worldview.
There is a mountain god in the mountains. There is a sea god in the sea. Tools, houses, and words all have roles and meanings. This feeling can be directly applied to corporate design for the AI era. Rather than a single central AI, AI located in various locations work together to drive the entire company.
Yaoyorozu will not change. This is not a matter of religion, but is a strong design philosophy for distributed AI agents.
dispersion
Distributed ownership
Don't rely on one central AI. AI resides at each work site and learns from on-site data.
interlocking
Evidence-connected
Observation, judgment, implementation, and verification are connected by a trail. This becomes company-specific learning data.
governance
Governance by design
Define roles, permissions, authorizations, and publication boundaries. So it won't break even if you move fast.
Yaoyorozu — Distributed IntelligenceGPIE
Chapter II — Where AI Lives05 / 14
§ 04
AI is not just a surface layer of management;
Where P/L movesreside in
AI that impresses managers is not a convenient demonstration. It is an AI that can see what will drive sales growth, gross profit improvement, SG&A expense reduction, decision-making speed, and risk reduction.
01
investment decision
Capital allocation
Continuation, withdrawal, redistribution. AI speeds up investment committees by aligning evidence, counterevidence, and alternatives.
02
Profit improvement
Margin expansion
Price, cost, selling, general and administrative expenses, occupancy rate. AI constantly detects areas where profits are leaking.
03
growth hypothesis
Growth intelligence
Market, competition, customers, channels. AI will reduce the next growth theme to units that can be tested.
04
Implementation trail
Execution evidence
Who changed what and on what basis? AI leaves traceability of results and responsibilities.
Where AI Lives — Inside the WorkGPIE
Chapter II — Where AI Lives06 / 14
§ 05 — the false start
Chatbot distribution is
Welfare benefits, not AI transformationIt ends with
Layer 01
personal use
Employees ask AI. However, it does not become a learning asset for the company.
→
Layer 02
local automation
Post sentences and code. However, it does not connect to P/L, liability, and trails.
→
Layer 03 — AI Native
Lives in the management OS
AI enters a cycle of observation, judgment, implementation, verification, improvement, and audit.
— AI will change from “something that listens” to “a structure that drives management.” Until this point is overcome, AI investment will remain a cost center.
Not a Chatbot StrategyGPIE
Chapter II — Where AI Lives07 / 14
§ 06
The AI agent isinvestment hypothesisCalculate backwards and place.
VCs aren't looking at flashy technology. Market, proprietary data, execution speed, reproducibility, moat. Decide first which AI will drive which KPI.
| Investment thesis | Agent placement | Value output |
|---|---|---|
| sales growthRevenue intelligence | 85 % | ICP, price, channel, opportunity temperature |
| Gross profit improvementMargin expansion | 90 % | Cost, man-hours, rework, utilization rate |
| development speedBuild velocity | 80 % | Specification, Code, Testing, Release |
| Control powerGovernance moat | 95 % | Authorization, trails, public boundaries, risk |
Agent Map — Role × Placement × OutputGPIE
Chapter III — How to Lead Agents08 / 14
§ 07
The role of management is not to give orders to AI.
Running a company with AIBecome.
Human-in-the-Loop
The faster AI moves, the more managers will become "designers" rather than "approvers."
Which market to pursue, what profit rate to aim for, and what risk to accept? It is man's job to set a purpose.
Distinguish between what should be left to AI and what should be left to humans.
Leadership of agents
AI group,Direct towards capital efficiency.
Sales, gross profit, payback period, LTV/CAC, development speed, audit cost. The results of the AI group will be linked to management indicators.
We look at whether capital efficiency is improving, not whether AI is being used.
Human Role — Lead, Bound, DecideGPIE
Chapter III — How to Lead Agents09 / 14
§ 08
At an AI Native company,Trail becomes an assetbe done.
The basis used for the decision, counter evidence, implementation results, remand, and improvements. If these do not flow, are not recorded, and are not reused, AI will end up being a one-off useful tool. If it flows, it becomes a company-specific data asset.
Bad loop.
AI answers. used by people. It ends. Nothing remains for the next decision.
In this case, even if AI spending increases, the company's moat will not increase.
Good loop.
The execution trail is
Company-specific learning assetsBecome.
Observation, judgment, implementation, validation, and improvement are connected in the same trail. This creates a competitive advantage that is difficult to replicate.
— Only companies that follow the trail will change from companies that use AI to companies that grow with compound interest through AI.
Evidence Flow — From Output to LearningGPIE
Chapter III — How to Lead Agents10 / 14
§ 09 — Interlude
The more AI dwells,Control is competitiveness, not costBecome.
Purpose · Boundaries · Trail · Responsibility
VC looks at speed. Big companies look at control. Management takes responsibility. AI Nativeization means achieving these three things at the same time.
Pillar 01 — Velocity
speed
— Decision velocity
Shorten lead times for review, approval, implementation, and validation. Speed is a management weapon.
The Bridge — Control
control
— Governance by design
Authorization, public, private, DB, git, external sending. Because we have boundaries, we can move quickly and with peace of mind.
Pillar 02 — Asset
Capitalization
— Evidence as data moat
What did the AI see and what did it do? By leaving a trail, organizations continue to learn.
Purpose × Boundary × EvidenceGPIE
Chapter IV — Risk and Governance11 / 14
§ 10 — Risk
In the world where Yaoyorozu's AI dwells,
proliferation without governancedestroys corporate value.
The scary thing is not that there are so many AIs.
The line between AI execution responsibility and disclosure is ambiguous.
AI that has strayed from its purpose continues to produce plausible results. It becomes impossible to keep track of who made which AI do what. Investors don't just dislike slow growth. This is an unexplainable AI risk.
It's off the KPI.
AI is fast. That's why you move quickly toward the wrong KPIs. AI that is not connected to sales, gross profit, and risk becomes noise.
Shadow Agents will increase.
I don't know who ran what. Only the artifacts remain, and there is no trail. This destroys auditing and accountability.
False Safety remains.
It appears that there are rules, but they are not actually enforced or verified. This false security is the most dangerous.
— So Hook, Permission, Sandbox, Run Ledger, and Approval Gate are not just operational aids. It becomes the very internal control of AI Native companies.
Risk — Drift, Shadow, False SafetyGPIE
Chapter IV — Risk and Governance12 / 14
§ 11 — Governance
Governance that compounds speed
Visible Investable Audited Compounding.
AI Native Governance
Governance as investability
Governance and guardrails are not meant to slow down AI. It's there to speed up the investment of capital.
You can see the role of AI, inputs, outputs, basis for judgment, and execution results. Dangerous operations, border crossings, unauthorized changes, and destructive practices will cease.
Failures, reversals, and improvements are recorded and reflected in the next AI operation. This is where it becomes moat.
Speed and governance are not at odds. Because we have governance, we can do it quickly. You can invest because you have boundaries.
Governance — Visible, Bounded, Audited, Learning2026—
Chapter V — GPIE Implementation13 / 14
§ 12
GPIE uses Yaoyorozu AI
Enterprise Agent MeshImplement as.
Domain 01 — Legacy to Native AI
Transform legacy structures into structures where AI can reside.
Don't just update old systems. Shift to a structure that allows you to incorporate AI agents into your business, data, APIs, operations, and trails.
Domain 02 — Agentic Operating Layer
Place an AI layer that connects revenue and implementation.
Don't separate sales, gross profit, operations, implementation, and auditing. Connect everything from management KPIs to implementation tasks with the same trail.
Domain 03 — Evidence and Learning Loop
Leaving learning assets, not artifacts.
The basis for judgment and execution results are returned to the next AI operation. This allows the company itself to become smarter through compound interest.
Domain 04 — Human Leadership
Management directs AI groups towards capital efficiency.
Vision, boundaries, responsibility, and ethics. The more a company is equipped with AI, the more the quality of human judgment will determine its corporate value.
GPIE — AI Native Operating SystemGPIE
End of Keynote No.0214 / 14 · GPIE 2026—
Colophon
From a company that uses AI,
Towards a company where AI resides.
Yaoyorozu's AI is a design philosophy that drives corporate value.
The era of treating AI as a tool is over. From now on, we will enter an era in which AI with a role will reside in various areas of management, humans will lead the group, and capital efficiency will change.
That's The Age of Enterprise Agent Mesh.
No.01 What is AI era? Return to To be Continued… No.03 How to create a company that houses AI. proceed to
Thank youGPIE