Ask a department head what worries them about AI and you usually hear one of two fears. Either it will mean cuts they will be asked to deliver, or it will mean a lot of activity with nothing measurable at the end.
In September, the Indian IT services company Wipro made a claim that sits between those two. It is worth reading carefully, partly for what it says and partly for what it does not.
What Wipro said
Wipro’s chief technology officer, Sandhya Arun, told Reuters on 10 September that the company’s AI work has added productivity equal to the output of 20,000 employees. Wipro had about 243,000 staff in June, so that is roughly one in twelve.
The people whose work was freed, she said, were redeployed inside the company rather than let go. Alongside that, more than 100,000 Wipro employees are going through advanced AI training and certification, as the company moves to what Arun calls a “human-AI operating model”.
On what that means for individual jobs, she was specific. “It could be the same engineer managing a bunch of agents, deployed on other projects or being trained for some other role,” she told Reuters. “It doesn’t necessarily mean person-to-person replacement by an agent.”
The questions the number needs
The 20,000 figure is Wipro’s own. Reuters did not report a method for counting “equivalent output”, over what period it was measured, or how it was checked.
That matters because “equivalent output” can be counted in very different ways. It might mean hours saved on specific tasks, added up across the company. It might mean extra work delivered without extra hiring. Those can produce very different numbers from the same reality.
There is also a commercial context. Wipro is the only one of India’s four largest IT companies that does not disclose its AI revenue. Manoj Chandra Jha, principal analyst at Nord-IQ Research, told Reuters that Wipro is still in an earlier, cost-absorbing phase of making money from AI, and trails its larger peers on that front.
None of that makes the claim false. It does mean the 20,000 should be read as a company describing its own progress, not as a measured result. If your own leadership quotes a figure like this, the same three questions apply: counted how, over what period, and checked by whom.
Why the training figure is the one to read
The more useful number is the training one, because it is harder to inflate and it describes a choice.
More than 100,000 people is about four in ten of the workforce. Whatever the precise capacity gain, Wipro has decided that a large share of its staff need to learn new ways of working, and it is paying for that at scale. Arun put the challenge as making sure all engineers build the skills to work effectively with AI systems.
That is the part a department head can learn from, because it does not depend on the size of the company. The capacity gain, if it comes, is only worth something if the people whose time was freed can do something more valuable with it. That needs a plan for what they will do, and usually some training to get them there.
The decision behind the headline
Strip out the scale and Wipro’s claim describes a decision every manager using AI will face sooner or later. Some work that used to take a person’s time now takes less. What happens to that time?
There are three common answers, and only one of them is a decision.
The first is that nothing happens. The time leaks away into slightly longer meetings and slightly faster replies, and six months later nobody can point to a difference. It is the easiest outcome to drift into, and one reason so many AI pilots report activity rather than results.
The second is that someone else decides. A budget review notices the team seems to have capacity, and the headcount conversation starts without the manager in it.
The third is that the manager decides first. They name the work that will get more attention, the person who will move to it, and what that person needs to learn. Arun’s phrase for the test is useful here: “The shift has to be from productivity to outcomes.”
The third answer is also the only one that protects the team. A manager who can show where the freed time went, and what it produced, is in a far stronger position in the next budget conversation than one who can only say the team is busy.
Consider a finance team where AI now prepares the first draft of the monthly variance commentary. That used to take one analyst three days. If nobody decides anything, those three days disappear. If the manager decides, the analyst might spend them on the supplier contracts nobody has had time to review, after a short course on contract terms.
What to do before the time appears
Decide where freed time goes before you have it. When you start using AI on a task, write down what the person doing it would work on if the task took half as long. If you cannot answer, that is the gap to close first.
Pair every redeployment with learning. The new work is usually less familiar than the old. Budget the training at the same time as the change, not after someone struggles.
Report the outcome, not the hours. Hours saved is the number everyone quotes and nobody can check. What the team did with them is the number that convinces a sceptical finance director.
Wipro’s 20,000 may or may not hold up when the company’s AI work starts to show in its revenue. The decision underneath it holds up either way. AI frees time only on paper until a manager decides where it goes.