Something important is happening in enterprise AI.
AI agents are starting to move beyond answering questions and recommendations.
They are beginning to operate inside commercial systems.
One recent example highlighted in my AI strategy radar is Amazon opening Seller Central to third-party AI agents, including Anthropic’s Claude integration, allowing agents to help sellers with activities such as inventory, pricing and listings.
At first glance, this may look like another AI integration.
I think it represents a much bigger architectural shift.
Yesterday, the interaction looked like:
“Tell me how to change the price.”
The AI provides advice.
The human makes the decision.
The human performs the action.
Now the model can become:
Analyze inventory and demand
→ Recommend a price
→ Execute within defined rules
That is a different world.
Because the AI is no longer simply talking about the business.
It is starting to work inside the business.
And that creates a critical distinction.
AI as Advisor
“Here is what I think.”
AI as Operator
“I can do this for you.”
AI as Enterprise Actor
“I have authority to execute this business activity within defined boundaries.”
The jump from Operator to Enterprise Actor changes the architecture.
The chain becomes:
Model
→ Agent
→ Context
→ System
→ Authority
→ Action
→ Outcome
And suddenly a completely different set of questions becomes essential.
Who is the agent?
What does it know?
Which business system can it access?
What capability is it allowed to use?
What decisions can it make?
What requires human approval?
Who is accountable?
What happens when something goes wrong?
How do we measure the outcome?
This is no longer only an AI question.
It is a Business Architecture question.
Because once AI can change enterprise state, the enterprise has to define the boundaries within which that change is allowed.
That means connecting:
Capabilities → Processes → Rules → Decision Rights → Systems → AI Actions
And this is where governance becomes much more than a document.
Governance becomes part of the execution architecture.
Imagine an AI agent managing ecommerce pricing.
The agent may have access to:
Inventory data.
Demand signals.
Competitive information.
Pricing rules.
Promotion rules.
Margin thresholds.
Customer-segment policies.
And the pricing system itself.
The important question is not:
“Can the AI change the price?”
The important question is:
“Under what conditions is the AI authorized to change the price?”
That distinction is enormous.
Capability tells us what the AI can do.
Authority tells us what it may do.
Governance tells us how we control and observe what it does.
And Business Architecture tells us why that action exists in the first place.
This gives us a bigger enterprise AI value chain:
Infrastructure
↓
Data
↓
AI Models
↓
Agents
↓
Authority
↓
Applications & Business Systems
↓
Business Architecture
↓
Operating Model
↓
Business Outcome
This is the direction I believe enterprise AI is moving.
AI is moving down the stack.
Yesterday:
AI talked about the business.
Today:
AI works with the business.
Tomorrow:
AI increasingly acts inside the business.
And that changes the role of enterprise architecture.
Because when an AI system can actually change enterprise state, it can no longer be treated simply as another software feature.
It becomes an enterprise actor.
That is the sentence I would underline.
The moment AI can change enterprise state, AI becomes an architectural concern.
Before that moment, AI can often be treated primarily as a tool.
After that moment, the enterprise needs to think about:
Capability.
Context.
Authority.
Governance.
Accountability.
Outcome.
This is also where Grounded Strategy™ becomes particularly relevant.
The agent cannot operate in isolation.
It needs:
Enterprise Context
→ Business Meaning
→ Capability
→ Process
→ Rule
→ Decision Right
→ AI Action
→ Business Outcome
And then something even more interesting happens:
Outcome → Learning → Capability Evolution
The enterprise is no longer simply deploying AI.
It is beginning to create a mechanism through which AI can participate in continuous enterprise evolution.
That is a very different strategic proposition.
The future question is therefore not:
“Where can we put an AI agent?”
It is:
“Which enterprise capabilities should AI participate in, with what authority, under which rules, and with what measurable outcome?”
That is where AI stops being just technology.
And becomes part of the enterprise architecture.
Grounded Strategy™
From enterprise reality to AI-enabled action.
#AI #AgenticAI #BusinessArchitecture #EnterpriseAI #AIGovernance #AITransformation #OperatingModel #EnterpriseArchitecture #Strategy #GroundedStrategy



