For many organisations, the AI conversation has started with tools. Which platform should we subscribe to? Which model is better? Should we build or buy? What are other organisations using?
These are useful questions. But they are not the first questions leaders should be asking.
The more important question is: What do we want AI to do for our organisation?
AI is more than a technology purchase. It has the potential to change how work gets done, how information is used and where decisions are made. That requires a strategy, not simply a choice of tools.
An AI strategy should reflect how the organisation works, where AI can genuinely create value, what risks it is willing to take, and where human judgment must remain firmly in control.
In other words, don’t let the technology define your AI strategy. Own it.
It is easy to begin with what AI can do. A better place to begin is with what your people do.
Look at the work happening across the organisation. Where is time being spent? Which tasks are repetitive? Where are people reviewing large amounts of information? Where are delays occurring? Where does valuable knowledge sit with a few individuals rather than being easily accessible? And where could better information lead to better decisions?
These questions will usually reveal more meaningful opportunities than starting with a list of AI features. The objective should not be to find places to use AI. It should be to find problems worth solving.
Once you identify those problems, resist the temptation to simply automate the existing process.
If something takes ten steps today, the instinct may be to use AI to complete those ten steps faster. But the better question is: why are there ten steps at all?
AI gives organisations an opportunity to rethink how work gets done. Steps may disappear or combine. Information may no longer need to pass from person to person. Tasks that were previously sequential may happen simultaneously.
Redesign the process first. Then decide where AI can automate or accelerate it.
Making an inefficient process faster still leaves you with an inefficient process.
A good AI strategy is as much about boundaries as possibilities. AI may perform some work almost independently. In other areas, it may prepare a first pass for human review. It may provide analysis while the final judgment remains human. And there may be areas where an organisation decides AI should not be used at all.
Those boundaries will differ. A law firm, bank, hospital, manufacturing company and university will have different considerations around confidentiality, accuracy, accountability and risk.
There cannot be one universal AI strategy. Your strategy has to reflect your organisation: its work, risks, responsibilities and appetite for AI.
Buying AI tools is not an AI strategy. Neither is announcing that your organisation is “AI-first.”
What matters is whether AI is changing how work gets done. Is it improving quality? Reducing time spent on low-value activity? Enabling people to do things they could not reasonably do before? Helping them make better decisions, rather than simply faster ones?
Organisations also need visibility into how AI is being used. Employees will experiment, teams will discover tools, and people will find shortcuts because the technology is useful. Trying to stop all experimentation is unlikely to work.
Instead, provide clear guidance on what tools can be used, what information can be shared, what must be verified and when human approval is required. Good governance should make responsible AI adoption easier, not make useful technology difficult to use.
Much of today’s AI conversation revolves around models - which one is smarter or has the latest capabilities. Those differences matter, but as access to powerful models becomes more widespread, they may become less of a differentiator.
For organisations, a more interesting question is: What does our AI know about our business that everyone else’s AI does not?
The answer lies in context - internal knowledge, documents, processes, historical decisions, customer understanding, specialist databases and institutional experience.
Generic intelligence is increasingly available to everyone. Context is not.
AI strategy must therefore include a knowledge strategy: how organisational knowledge is captured, organised, protected and made available to AI securely. Organisations that have spent years building proprietary knowledge may find those assets becoming even more valuable. The advantage may not come from having AI. It may come from what your AI is able to work with.
AI can produce impressive work remarkably quickly. That can also be its danger. A polished answer can create confidence before it has earned it.
Someone still needs to ask whether the output is correct, where the information came from, what may have been missed and whether it can be verified. Most importantly: would I be comfortable putting my name behind it?
The more consequential the decision, the more important these questions become. Human oversight should therefore not be seen as a temporary safeguard until AI gets better. It is an essential part of responsible AI use and good governance.
No organisation will design the perfect AI strategy sitting in a conference room. Some of it has to be discovered through use.
Start with a few meaningful problems. Experiment and measure what happens. Speak to the people doing the work. Understand what saved time, what improved outcomes, what created problems and what people trusted - or chose not to use. Then adjust.
The objective is not to predict exactly what AI will look like three years from now. It is to build an organisation capable of learning and adapting as the technology changes.
Leadership should keep returning to a few fundamental questions: Where can AI genuinely improve the way we work? What should remain human? What information should AI have access to? Who verifies its output? How will we know whether it is creating value? And what capabilities will our people need as their work changes?
These may sound like technology questions. They are business questions.
Technology teams, consultants and AI vendors have important roles to play, but they cannot own the strategy. Decisions about how work changes, where judgment sits, what risks are acceptable and what creates value ultimately belong to leadership.
The organisations that benefit most from AI may not be those that adopt the most tools or move the fastest. They may simply be those that are clearest about why they are using AI, where they want it to create value and where they still want humans to decide.
Use the technology. Experiment with it. Learn from it. Adapt as it evolves.
But own the strategy.
About the author: Deepak Kapoor is the CEO & Founder at Manupatra.
Disclaimer: The opinions expressed in this article are those of the author(s). The opinions presented do not necessarily reflect the views of Bar & Bench.
If you would like your Deals, Columns, Press Releases to be published on Bar & Bench, please fill in the form available here.