During a recent conversation on The Executive Outlook, Isha Taneja spoke with Jerome Wittersheim, Founder of Hexagon Data Services, about data governance, artificial intelligence and what organisations must do before beginning an AI initiative.
Jerome shared a reality that many businesses discover only after investing in new technology. AI does not create value simply because an organisation introduces a powerful platform. It creates value when the data beneath it is accurate, governed, understood and connected to a clear business objective.
His perspective comes from years of managing information in the United Kingdom’s further education sector, where data quality directly affected government funding, regulatory confidence and organisational reputation. That experience shaped how he now approaches AI readiness.
He does not see AI readiness as a technology assessment. He sees it as a test of whether an organisation understands its data, responsibilities, business rules and intended outcomes.
AI may be the visible ambition, but data is the foundation that determines whether the organisation can trust its output.
The further education sector is a highly regulated industry that receives significant funding from public sources. Colleges and training providers collect hundreds of data points for every student and submit that information to government agencies to support their funding claims.
Accuracy is therefore more than a reporting expectation. It has a direct relationship with the income an institution receives.
External auditors, including firms such as KPMG and PwC, regularly assess whether the information these organisations hold reflects what is happening on the ground. They may spend weeks reviewing student records, funding evidence, systems and governance processes.
For him, this environment made the value of trusted data impossible to ignore.
Poor information could affect funding, expose an institution to scrutiny and damage its reputation. In serious cases, data failures could create consequences for senior leaders.
The experience taught him a lesson that applies across industries. Data quality is not merely a concern for analysts or engineers. It is a business responsibility with financial, operational and reputational consequences.
When leaders recognise those consequences, governance stops looking like administrative work. It becomes part of how the organisation protects value and makes decisions with confidence.
The discipline created by regulation also made education organisations cautious about emerging technology.
When cloud computing began gaining wider adoption, many institutions remained hesitant about moving from physical infrastructure to cloud platforms. They managed sensitive information, operated under strict funding rules and had little appetite for unnecessary risk.
At the same time, he could see the pace of technological change accelerating outside the sector.
Cloud platforms were transforming how organisations stored, processed and used information. They also provided computing power that would later support the rapid growth of artificial intelligence and large language models.
He wanted to move closer to that change. He stepped beyond education to understand how other industries managed information and how modern platforms could support wider business outcomes.
That journey eventually led him to establish Hexagon Data Services. Yet the discipline he developed in education remained central to his approach. New technology could create enormous value, but only when supported by reliable information, appropriate controls and clear accountability.
One of the strongest ideas he shared was that responsibility for data cannot sit with the data team alone.
Employees create, collect and update information every day without necessarily thinking of themselves as part of the data function. They may work in finance, sales, operations, customer service or human resources. Because they do not report to a data leader, they may assume someone else is responsible for the quality of the information they manage.
That assumption creates risk.
Every person who interacts with data influences whether it can eventually be trusted. A governance framework must make responsibility and accountability visible across the organisation.
This does not always require a complex system. It can begin with a shared data dictionary that explains what important terms mean and how they should be used.
A term such as revenue may appear clear, but different functions can define it differently. If those definitions are not aligned, reports will conflict, and AI may produce answers based on the wrong interpretation.
Governance brings those differences into the open. It creates agreement around definitions, ownership, quality standards and the rules guiding how information is used.
If accountability remains unclear, data quality will always be seen as someone else’s problem.
Formal responsibility is only one part of the solution. People must also understand why information matters.
Employees who collect or process data may see it as an additional task. They may not know how that information supports executive decisions, improves customer experiences or helps the organisation deliver a better service.
He believes data leaders must connect those dots.
When people understand how information moves through the organisation and influences wider outcomes, they are more likely to treat its quality seriously. Governance becomes easier to adopt because it is connected to a meaningful purpose.
Policies define what people should do, but communication helps them understand why it matters. That understanding turns governance from a document into a shared organisational behaviour.
The objective is not to make every employee a technical specialist. It is to help people recognise that their daily actions influence the reliability of decisions across the business.
The rise of tools such as Microsoft Copilot created enormous expectations. Many organisations believed these systems could scan files, interpret databases and immediately provide valuable insights.
The reality was often more complicated.
An AI tool does not automatically understand how an organisation works. It does not know which files are reliable, which definitions have been approved or why departments calculate the same metric differently.
Without a trusted data foundation, AI can produce answers that sound convincing but lack the context needed to make them dependable.
Employees try the technology, receive an inaccurate response and begin questioning whether it is useful. Leaders may then conclude that AI has failed to deliver on its promise.
Technology may not be a real problem.
The infrastructure may be fragmented. The data may be inconsistent. Business definitions may be unclear. The organisation may have asked AI to provide certainty from information that was never properly controlled.
AI does not repair weak foundations. It makes their weaknesses more visible.
Once trust is lost, employees become hesitant, executives question further investment, and valuable initiatives lose momentum. Organisations must establish the conditions for reliable results before asking people to trust what AI produces.
Governance establishes ownership and control, but AI also needs business context.
He describes this through terms such as context layer, semantic layer and business ontology. The language may vary, but the purpose remains the same. The organisation must give its systems a structured understanding of business rules, concepts and relationships.
Data rarely explains itself.
A customer record, revenue figure, funding claim or project status has meaning because of the rules surrounding it. AI needs access to those rules if it is expected to provide trustworthy answers.
Without that context, the system must be guessed.
It may infer what a field means from its name or from patterns in the available information. That may be useful in situations with limited risk, but guesswork is not sufficient for enterprise reporting, regulatory decisions or important operational actions.
A semantic layer helps AI understand not only what the data says, but what it means within the organisation. When that context sits above governed and reliable information, AI initiatives have a greater chance of producing accurate outcomes.
Cleaning data alone is not enough. AI must understand the business reality that gives the data meaning.
Looking back at projects that did not perform as expected, he identified a recurring pattern. Organisations often begin before they are ready.
Pressure to demonstrate progress can come from boards, executive teams and competitors. Leaders see other companies investing in AI and feel an understandable urgency to act.
That urgency can push teams into implementation before confirming whether the necessary foundations exist.
At Hexagon Data Services, Jerome and his team use a twelve-point framework to assess readiness before beginning a data warehouse or AI initiative. If an essential condition has not been met, they address the gap before proceeding.
Starting quickly may create the appearance of progress, but unresolved weaknesses eventually surface. By then, the organisation may have committed significant time, money and executive attention.
Preparation is not a delay in transformation. It gives transformation a realistic chance of succeeding.
He compares the process to building a house. No responsible builder would begin construction without a plan, a foundation and the correct infrastructure. Data and AI projects require the same discipline.
He has seen a significant gap in data maturity across industries.
Some businesses have established governance, modern cloud platforms and clear ownership. Others, including large companies managing complex projects, still depend heavily on spreadsheets.
These organisations may continue to operate effectively because experienced employees understand the business. They know how to interpret information, manage relationships and solve problems through accumulated knowledge.
The risk appears when those people leave or when the organisation needs greater efficiency.
Knowledge stored in employees’ minds is difficult to scale. Information spread across disconnected spreadsheets is difficult for AI to interpret consistently. The business may be functioning, but its ability to automate and innovate remains limited.
AI readiness cannot be judged by the size or apparent success of a company. The important question is whether its data environment can support where the business wants to go next.
Jerome’s advice to leaders begins with a practical question. What problem is the organisation trying to solve?
There is little value in launching an AI project without a clear objective. Leaders need a business problem that can be measured and connected to a meaningful outcome.
Reducing the time required to produce a report from ten working days to one is a clear objective. It gives the organisation a defined problem and a way to evaluate whether the investment created value.
Once the objective is established, the organisation needs executive sponsorship. Senior leaders must understand and support the initiative because data and AI projects often cross departmental boundaries.
The organisation must then confirm whether the required data is accurate, governed and fit for purpose. Definitions must be agreed upon; ownership must be clear, and business context must be available.
Only then should leaders select the technology.
They must also consider whether their data engineers, analysts and other team members have the skills to use it. A powerful platform will deliver little value if the people responsible for it lack the necessary training and support.
AI readiness therefore depends on business clarity, leadership commitment, trusted data, suitable technology and capable people.
Jerome does not argue against moving quickly with AI. He argues for moving with clarity.
His experience in regulated education taught him that data quality has real consequences. His work across industries has shown him how easily organisations overlook ownership, context and readiness when pressure to innovate becomes intense.
His message is practical. Begin with the problem. Establish accountability. Agree on business definitions. Give AI the context it needs. Secure executive sponsorship. Confirm that the data is ready and ensure the team has the skills to deliver.
These fundamentals are often the difference between an AI experiment and a trusted business capability.
AI readiness does not begin when an organisation purchases a platform or launches a pilot. It begins when leaders create conditions that allow intelligent systems to produce reliable outcomes.
AI can accelerate what an organisation already has. If the foundation is strong, it can accelerate insight, efficiency and value. If the foundation is weak, it can accelerate confusion.
The technology may shape the future, but trusted data will determine which organisations are ready for it.
Every leader has a story, experience and perspective worth sharing. If you have insights that can inspire the next generation of business and technology leaders, Get yourself featured on The Executive Outlook.
Listen in for business strategies and interviews with business experts to help you build a business and a life that you love.
About Us
