Part 2 of 3 - A series on operationalizing AI in enterprise marketing.
Today, someone inside your organization is already making operational decisions manually that AI systems will eventually automate.
Take something simple and common inside enterprise ecommerce environments: deciding which products appear on the homepage.
In many organizations, that responsibility sits with someone on the ecommerce or marketing team. Every morning, or maybe a few times a week, they review campaign performance, look at engagement metrics, check inventory manually, and decide which products deserve the most prominent placement.
The problem is not that the person is making bad decisions.
The problem is that no single person usually has access to the full operational picture while making them.
Marketing may understand conversion performance. Finance may understand margin pressure. Operations may understand fulfillment risk and inventory constraints. But the individual updating the homepage is typically working from only one slice of the business context available across the organization.
So, the homepage gets optimized using whichever signals are easiest to see.
That same limitation reappears repeatedly once organizations begin introducing AI into operational workflows.
AI Is Usually Optimizing Against Partial Business Context
Most enterprise AI systems today are connected to isolated datasets, isolated workflows, or isolated departmental priorities.
As a result, they optimize aggressively for whichever metrics they can actually access.
If an AI system primarily sees campaign performance data, it will optimize for campaign performance. If it mainly sees engagement metrics, it will optimize for engagement. If it only has access to short-term conversion signals, it may inadvertently create downstream problems involving inventory pressure, profitability, fulfillment instability, or customer experience.
The AI is not making irrational decisions. It’s making locally optimized decisions based on incomplete operational visibility.
In many ways, that is not very different from the manual decision-making process organizations already rely on today.
The difference is that AI systems can potentially operate against a much broader operational context than any individual human can realistically coordinate manually.
That’s where the workflow starts to change.
What the Workflow Looks Like Today
Most enterprise workflows still rely heavily on people manually coordinating information across disconnected systems.
Imagine the homepage merchandising example again.
A marketer may look at campaign performance dashboards. Someone else checks inventory availability separately. Finance reporting may exist in another platform entirely. If margin concerns exist, they often surface later through reporting reviews instead of influencing the decision in real time.
Even in relatively mature organizations, operational coordination frequently happens through spreadsheets, Slack conversations, reporting exports, email threads, and manual publishing workflows.
That process may technically work, but it creates an environment where decisions are constantly being made from incomplete context.
The person updating the homepage can only optimize using the information available to them at that moment.
What the Workflow Looks Like with Composition
Now imagine the same workflow operating differently.
Instead of someone manually reviewing reports every morning, an AI-driven workflow runs overnight.
Each night, the workflow reaches into Umbraco Compose and pulls structured operational context assembled from across the organization. That context could include:
- current inventory availability,
- product margin performance,
- fulfillment constraints,
- regional inventory conditions,
- active campaign priorities,
- customer demand trends,
- and operational rules defined by finance or supply chain teams.
Instead of simply promoting whichever products generated the highest engagement yesterday, the workflow evaluates which products should appear on the homepage based on broader business priorities.
For example, the system may determine that the homepage should feature the five highest-margin products that are currently well-stocked, operationally stable, and aligned with active campaign objectives.
That’s a fundamentally different decision model.
The important shift is not that the AI somehow became “smarter” than the person previously updating the homepage. The important shift is that the workflow now has access to a broader operational view of the business than any individual employee typically has access to on their own.
Where Umbraco Compose and MCP Fit Into the Process
This is where Umbraco Compose and Umbraco’s Model Context Protocol (MCP) orchestration environment start working together operationally.
Compose acts as the shared operational context layer. It assembles the business information, rules, constraints, and supporting data required for the workflow to make more informed decisions.
From there, Umbraco’s MCP orchestration environment coordinates execution across the workflow itself.
Once the AI determines which products should be prioritized, MCP can automatically update homepage content, trigger approvals when required, coordinate publishing workflows, and push updates across downstream digital experiences.
No one has to manually gather spreadsheets, compare reports across departments, coordinate updates in Slack, or publish homepage changes by hand the next morning.
The operational coordination is already built into the workflow itself.
That distinction matters because enterprise AI is increasingly moving beyond content generation into operational decision-making.
Organizations are no longer experimenting only with AI-generated copy or summaries. They are beginning to operationalize AI inside personalization systems, commerce workflows, campaign orchestration, inventory-aware experiences, and autonomous decision layers across customer experience platforms.
The Bigger Shift Organizations Are Moving Toward
This is the deeper transition many organizations are only beginning to recognize.
The long-term opportunity is not simply using AI to generate work faster.
It’s creating systems where operational decisions themselves can happen against broader business context than humans can realistically coordinate manually across disconnected systems.
That does not eliminate human oversight. It changes where humans spend their time.
Instead of manually coordinating information across systems every day, teams can focus more on defining business rules, governance models, operational priorities, and customer experience strategy upstream.
The workflow itself becomes more operationally aware because the context surrounding the decisions is already connected before execution happens.
The Real Enterprise AI Challenge
At this point, most organizations do not need dramatically smarter models.
The models are already capable enough to automate workflows, personalize experiences, generate recommendations, and support operational decisions in meaningful ways.
What most organizations actually lack is a reliable operational context layer that allows those systems to make decisions against the broader reality of how the business functions across teams, systems, workflows, and operational constraints.
Without that layer, every AI system ends up optimizing for one department’s metrics while inadvertantly creating friction somewhere else downstream.
The long-term advantage is not simply having access to AI.
It’s creating connected operational workflows where AI systems can make decisions against the full business context surrounding the work itself.