During a recent conversation with The Executive Outlook, Morgan Templar shared a simple reason many AI initiatives struggle.
The problem is often not the model. The organization was never ready for it.
A board approves the budget. A pilot gets funded. A capable team builds something impressive. Months later, the technology works, but nobody can clearly explain what changed for the business.
That is where the conversation about readiness needs to begin.
Morgan has spent twenty-five years working across healthcare data, information management, governance, and enterprise transformation. Her career has included Blue Shield of California, Cigna and Highmark Health, where she served as Vice President of Data and Information Management and led a large cloud-native data platform with a team of more than four hundred people.
In 2023, she is the CEO, CAIO and Founder of First CDO Partners, where she now advises enterprises preparing for AI. She also co-leads CDMC+AI and Data Product & Marketplace forums on AI governance work at the EDM Association and has written three books, including her latest book on Managing and Governing AI ecosystems, “The Cracked Egg: The AI FUTURE”.
Her view of readiness comes back to three questions.
• Do you know what you are trying to accomplish?
• Do you know where your data lives and what it means?
• Do you know who is accountable when a decision is made?
An AI readiness assessment should help leaders answer those questions before another pilot begins.
Healthcare was not part of a grand career plan.
Morgan needed a job and happened to find one at a hospital.
Her work involved physician credentialing, which meant validating and verifying information about doctors: Where did they train? Where did they work? Are their credentials correct? Were they qualified to care for patients?
On the surface, it sounded administrative. In reality it was the first defense for excellent patient care.
It showed her how much could depend on the quality of a single record.
That experience started a career built around data.
At Coventry Healthcare, mergers and acquisitions had created information that was inconsistent and difficult to coordinate. Her work focused on consolidating it and making it more useful.
At Blue Shield of California, she led the transformation of provider data from mainframe systems to the cloud.
At Highmark Health, her responsibilities expanded across the full data environment, including warehouses, APIs, data engineering, incoming information and outgoing information.
Across every role, one lesson stayed with her.
If you want AI to work, you first have to take care of the data underneath it.
Read the headlines and it can feel as if every large company is already deep into AI.
Major enterprises announce new platforms. Consulting firms publish success stories. Boards begin asking why their own organization is moving more slowly.
She sees a very different picture.
In the conversation, she argued that the highly visible companies making major advances represent only a small part of the market. Many other organizations still do not have a clear AI strategy, do not fully understand where their data live and have not decided how that information should be used.
That is why an AI readiness assessment matters before another proof of concept.
The issue is rarely a lack of ambition.
It is whether the organization is actually ready – financially and organizationally – to turn that ambition into something useful.
There is no shortage of ways to measure AI readiness.
Companies can use a variety of AI readiness assessments, vendor maturity models, or internal scorecards.
She starts somewhere simpler.
“What are you actually trying to accomplish?”
Without a clear answer, organizations can spend significant amounts of money on proofs of concept simply because an idea sounds interesting.
Of increasing concern, the pilot may work technically and still fail to contribute to anything the business actually needs.
Her AI readiness assessment framework begins with the outcome.
First, understand the target. Then make sure there is a real strategy behind it.
Assess the current state against the future. Identify the gaps. Build a roadmap around what the organization needs.
Those gaps often appear in technology, training, governance, or data.
Sometimes technology is missing.
Sometimes employees need training, so they understand what is changing and why.
Sometimes the information itself needs to be cleaned, mapped, structured, or connected through knowledge graphs and ontologies.
Often, she finds that several of these problems exist at the same time.
That is where good AI readiness assessment services provide more value than a simple score.
The objective is not merely to tell a company whether it is ready.
It is to show leaders what is preventing readiness.
AI readiness is often framed around opportunity.
New products. Faster decisions. Greater productivity. New revenue.
She offers another perspective.
At one organization, she worked on understanding where data were being shared internally and where that sharing created privacy exposure.
The potential risk was significant.
The organization invested thirty million dollars in fixing the problem. She described the resulting reduction in risk as roughly ten times the investment.
Her point is important for executive teams.
Data creates offensive value and defensive value.
The offensive side is easier to see because it promises growth.
The defensive side can be less visible until something goes wrong.
A useful AI readiness assessment therefore needs to consider both.
Readiness is not only about what AI might enable.
It is also about whether the information supporting AI creates risks the organization has not yet understood.
For years, enterprise data architecture followed a familiar pattern.
Take information from multiple systems. Move it into a central warehouse. Store it there. Then work out what it means and how people should access it.
Every movement creates additional storage and compute requirements.
She believes AI is changing that model.
Instead of moving every piece of data, organizations can increasingly map information virtually and move the context around it.
Where does the information live?
What does it mean?
How can it be accessed?
What is available?
The model begins to look more like a library.
You do not need every book sitting on the same table. You need to know what exists, where it belongs and how to retrieve it when needed.
That creates an important shift in how leaders think about readiness.
The goal is not necessarily to move all enterprise data into one place.
It is to understand the information wherever it already lives.
Ontology is one of those words that can quickly make an AI conversation sound more complicated than it needs to be.
She explains it by comparing it to a library.
Imagine that every piece of information has a place in the catalogue. You know the section, the shelf and where you should look for it.
Ontology brings similar structure to organizational knowledge.
At its foundation are simple facts.
She describes these through RDF triples, which use a subject, a predicate and an object.
“Big Ben ... is located in ... London”
That creates a fact with a clear meaning.
As more facts are connected, organizations can add business glossaries, technical metadata and context about how different teams understand the information.
Those relationships begin to form a schema.
From the schema comes an ontology.
From that ontology, AI can begin to understand relationships and context instead of simply processing isolated pieces of information.
For executives, however, the most useful part of her explanation comes next.
Your organization may have already done much of this work without calling it ontology.
During digital transformations, teams often spend weeks discussing what individual fields mean, where information comes from, how it should be transformed and what it should look like in the next system.
That knowledge often sits inside source-to-target mapping documents and old project files.
Morgan's suggestion is practical.
• Find those documents.
• Use AI to learn from them.
Then add any missing context that was not captured originally. Ensure you have considered:
• Which APIs use the information?
• Which reports depend on it?
• Which dashboards include it?
• What decisions do people make with it?
That is how business knowledge becomes a useful context for AI.
There is another problem that technology cannot solve on its own.
A large part of how work gets done may never be documented. The actual “how to” is verbally passed through training or understanding gained from experience.
She describes processes where employees simply know what to do next.
→ Click here.
→ Move to this screen.
→ Enter this information before that information.
→ Check something before submitting it.
Those details may not exist in a procedure manual.
They live in someone's experience.
She warns that if AI learns only from the written process, it can miss the unwritten rules that make the process work.
That is why the human in the loop remains important.
Employees need a way to tell the AI when something happened in the wrong order, when a step was missed, or when the system misunderstood how the work is really done.
That feedback should not be treated as a temporary phase before deployment.
It is part of how the system improves.
The lesson for leaders is significant.
AI readiness is not only about documenting systems.
It is also about capturing the knowledge that people pass along verbally.
When she talks about the capabilities enterprises' need for AI, two stand out.
Architecture and Ontology.
→ Ontology gives AI context.
→ Architecture determines how that context, technology, security, policies, and permissions work together.
A business solution architect should understand how the process works today and what it should look like with AI.
Technical architects then examine the technology supporting that future state.
Is the right software in place? Do we have enough Compute?
Are the permissions correct?
Is security designed properly?
And increasingly, another question matters.
What is the AI Agent actually allowed to do?
Morgan suggests thinking about an AI agent much like an employee.
• Give it a role.
• Define its responsibilities.
• Set boundaries around what it can access and what actions it can take.
That is architecture becoming governance.
She also points to a common problem in enterprise transformation.
The business chooses a system first and tells the architect to make it work.
She believes the process should run the other way.
The business should explain the problem and the outcome it needs.
The architect should then help design the right solution.
Business leaders are trained to think about outcomes.
Architects are trained to determine how those outcomes can be delivered safely and effectively.
Both are necessary.
Traditional data governance has often been tied to job titles.
Data Owner.
Data Steward.
Data Custodian.
She takes a different approach.
Governance, in her view, should be connected to what someone is doing at a particular moment.
She describes three roles: Guardian, Maker, and Seeker.
Someone making decisions and applying policies acts as a Guardian.
Someone changing information acts as a Maker.
Someone looking for knowledge, building a report, creating a dashboard, or answering a question is a Seeker.
The role changes with the transaction.
So does the responsibility.
That idea becomes particularly important for AI agent governance.
A single AI agent may have permissions, security profiles, and instructions to search for information or make changes to system data or creating reports or a dashboard.
Governance therefore cannot exist only as a policy document reviewed later.
It needs to operate where and when the action is taking place.
That thinking is central to the AI agent governance framework Morgan discusses in her work and her book.
The broader principle is straightforward.
Accountability needs to exist at the point where the decision or action happens.
This changes the traditional picture of governance.
For many organizations, governance happens after the fact.
✓ Someone performs an action.
✓ The system records it.
✓ An audit reviews it later.
Morgan's approach moves governance closer to the transaction itself.
Policies, permissions, context and accountability should increasingly be engineered into the systems where decisions happen.
That matters even more as organizations give AI agents greater independence.
If an AI agent is allowed to act, the organization needs to control what it can do, which policies apply, what information it can access and who remains accountable for the result.
This is where an AI agent governance platform or framework becomes useful, but it must supports the organization's actual policies and operating model.
Technology should enforce the rules the business has already decided on matter.
It should not be the place where those rules are invented.
Morgan included an unusual experiment in her latest book.
She had agents from three AI systems discuss the manuscript, question one another, summarize their ideas and independantly wrote the foreword to “The Cracked Egg: The AI FUTURE”
She intentionally allowed the agents to do the work.
That was part of the point.
But the experiment also reinforces the deeper message running through her conversation.
AI agents can work with what an organization has given them.
They cannot define the company's strategic target.
They cannot automatically recover important business knowledge that was never documented.
They cannot decide what level of risk the organization is willing to accept.
They cannot determine their own boundaries of accountability.
Those remain in leadership decisions.
That is what AI readiness assessment should ultimately be.
Not simply whether the organization owns the right technology.
But whether leadership knows what it wants to achieve, whether the business understands its data, whether the architecture supports the intended outcome, and whether accountability exists now; decisions are made.
AI readiness is not a technology checkpoint.
It is an operating question.
The companies that make meaningful progress with AI will be the ones that can answer those questions before asking an agent to act.
Are you a CIO, CDO, data leader, AI executive, or technology founder with experience that can help global leaders make better decisions?
Share your perspective, leadership lessons and industry insights with an executive audience.
Get featured in The Executive Outlook and join the conversations shaping the future of data, AI and digital transformation.
Listen in for business strategies and interviews with business experts to help you build a business and a life that you love.
About Us
