ARTIFICIAL intelligence has quickly become a country, economy and boardroom priority. Across different organisations and industries, leaders are asking the same questions: which model should we use? Should we deploy ChatGPT or Claude? Do we need an AI strategy? Where should we begin? How quickly can we implement it? These are important questions. But I believe they overlook a more fundamental one: is our organisation actually ready to benefit from AI?
Over the last few months as I have been thinking about how AI is changing organisations, this question has become increasingly difficult to ignore. Many organisations are treating AI as a technology challenge when, in reality, it is exposing a much older management challenge. That challenge is how organisations manage information.
Most discussions assume that AI itself is the limiting factor. I suspect the opposite is true. For many organisations, particularly in developing economies like Pakistan, the limiting factor is the organisation itself.
Consider a manufacturing company operating several production lines. Every morning, the production manager decides which supervisors should run which lines, which maintenance engineers should be assigned to planned shutdowns, which operators should work overtime and how production should be adjusted if a machine unexpectedly fails. These decisions rely on dozens of variables: employee skills and certifications, attendance, maintenance history, machine capability, quality performance, customer delivery commitments, inventory levels and production priorities. An AI system could analyse all these variables simultaneously and recommend the allocation most likely to maximise output, minimise downtime and meet customer commitments.
Two organisations can invest in the same AI technology and achieve different results.
Or take a call centre receiving thousands of calls each day. Traditionally, incoming calls are routed using simple rules such as the next available agent. An AI-enabled system, however, could optimise each interaction by simultaneously considering a far richer set of variables; the customer’s history and lifetime value, previous complaints, the complexity of the issue, the agent’s expertise, language proficiency, first-call resolution rate, customer satisfaction scores, current workload, average handling time and even the likelihood of customer churn.
Based on these factors, AI could assign each call to the agent most likely to resolve the issue quickly while maximising customer satisfaction and retention at the same time. But this capability depends on the availability of reliable organisational information. If customer histories are incomplete, agent skills are poorly documented, performance data is inaccurate or systems are disconnected, AI has little to optimise. It can only make better decisions when the organisation has already captured, maintained and integrated the information on which those decisions depend.
But there is one obvious question: from where does the AI obtain this information?
In many organisations, much of it does not exist in a form that AI can use. In my experience, working with different types of organisations, some information exists in an ERP system or other disconnected software. Some is stored in spreadsheets on individual laptops. A lot of it nowadays rests in WhatsApp conversations! And the most valuable knowledge frequently exists only in the minds of experienced managers.
Therefore, the production manager, not the organisation, knows which operator can recover a troubled production line, which maintenance engineer consistently diagnoses recurring faults and which supervisor performs best under pressure. The same with the call centre manager who relies on his personal knowledge of his teams’ capabilities to make critical decisions. This distinction is more significant than it first appears.
AI cannot optimise information that has never been captured, nor can it analyse knowledge that exists only in someone’s memory. Nor is it able to make better decisions when the underlying information is fragmented, inconsistent or inaccessible. In other words, it does not create organisational intelligence. It depends upon it.
This helps explain why two organisations can invest in the same AI technology and achieve completely different outcomes. One has spent years documenting processes, maintaining reliable operational data, integrating systems and making organisational knowledge accessible. The other relies on personal experience, informal processes and fragmented information. The technology is identical but the organisations are not.
For decades, experienced managers compensated for weak information systems. They knew where critical information was located, who possessed specialised expertise and how to make things happen despite organisational shortcomings. AI can’t be productive in the same way. It can only work with the information an organisation has chosen to capture, maintain and make available. This is why I believe we are asking the wrong question. Instead of asking whether organisations are adopting AI quickly enough, we should first ask whether they have built the organisational foundations that allow AI to create value.
That requires a different set of management questions. Where does our organisational knowledge reside? How much information exists only on individual laptops or in personal email folders or WhatsApp groups? Which decisions depend on a manager’s memory rather than organisational data? How reliable is our operational information? What information is continuously updated? And what information can AI actually access? These are not technology questions. They are management questions. This issue is particularly relevant for Pakistan. Many of our organisations have grown successfully despite fragmented information because capable managers developed ways to work around weak systems. But as AI becomes more deeply embedded in work, those workarounds become constraints.
The organisations that benefit most from AI are unlikely to be those that simply purchase the latest technology. They will be those that have invested in disciplined information, organised knowledge and management systems that make both available across the organisation.
Perhaps, then, the missing foundation of AI readiness is not more sophisticated technology, but greater organisational information maturity. That may ultimately prove to be AI’s most important lesson. Before AI can transform organisations, organisations must first transform the way they manage what they know.
The writer advises boards and CEOs on strategy, leadership and organisational transformation.
Published in Dawn, August 1st, 2026