Most Indian entrepreneurs are about to make the wrong decision about AI. They will make it this year, or the next. They will make it without realising it is a decision.
AI has made part of the team's work easier. The natural instinct is to ask: can we operate with fewer people? It is the wrong question. The better question, and the one most entrepreneurs are not asking, is this — can we now take on work we could not handle before?
The numbers say which question matters. A 2025 EY report of 500 senior decision-makers found that of firms reporting productivity gains from AI, only 17 per cent used those gains to cut headcount. The other 83 per cent reinvested — into expanded AI capability, new service lines, R&D, price cuts, acquisitions. The cost-cutters made the news. The re-investors captured the market. Five years from now, those two kinds of companies will not be in the same conversation. One will have been quietly acquired. The other will have done the acquiring.
AI now does up to 80 per cent of the grunt work in most white-collar roles. A finance team that spent three weeks every month on bank reconciliations, GST returns, and audit working papers finishes the same work in three days. A sales team that spent half its time writing proposals and updating CRM entries now spends most of that time in front of clients. This is most of the working week being handed back to you. The biggest strategic decision you will make this year is what to do with that time.
You have two choices. You can use the saved time to operate with fewer people and book the difference as cost savings. Or you can keep the same people and relocate them to work that generates real value. Take your operations head, who today spends most of the week compiling Excels and pulling together reports. With AI, that work shrinks to a fraction of the time. The same person can now actually read those reports, draw insights, find the customer segments you are losing money on, find the markets you have not yet served, and act on them. The first choice shrinks the company. The second grows it. Both feel rational in the moment. They produce very different companies in five years.
Before you cut anyone, ask the demand elasticity question. If you could deliver more, possibly at a lower price, would your market actually buy more? In most businesses, the answer is yes. There are clients you are turning away because you do not have the time or the resources. If this is true, AI is a growth lever, not a cost lever. Use it to add new products, enter new geographies, or drop prices to take share from competitors who have not adopted AI.
The news is dominated by large companies laying off their employees in the thousands. Consider Oracle. They laid off 20,000 employees and redirected the savings into AI infrastructure. Large companies like Oracle have always been heavily staffed, sometimes overstaffed, and they can absorb decisions of that kind. This is not the model an Indian SME should copy. The SME owner has a much better option. The people whose time AI frees up are exactly the people who can be redirected to find new revenue, build new offerings, and serve customers you currently cannot. The Oracle move is a balance-sheet trade. The SME move is a growth trade. They are not the same play.
Most firms make the same mistake. They keep their existing process and add AI somewhere in the middle. The team has new tools but the work moves at the old speed. Margins do not improve. Leadership concludes AI does not work. The technology is not the problem. The process is.
A Kerala real estate developer we spoke to used to follow a six-week cycle for project launches. Marketing took two weeks to write brochure copy. Sales waited for the brochure before calling leads. The CRM was updated by hand. When AI arrived, the obvious move was to use it to write brochure copy faster. The brochure now took three days, but sales still waited for it. The cycle dropped from six weeks to five. A marginal improvement. The better move was to redesign the cycle entirely. AI was put to work across every step at the same time — brochure copy, visuals, lead-scoring against past customer data, first-touch WhatsApp messages, live CRM updates. None of these had to wait for the others. Sales started calling qualified leads on day one, with AI-generated context on each. Marketing kept refining the brochure in parallel. The CRM updated itself in the background. The cycle dropped to nine days.
Map your top three processes this month. Identify which steps AI can absorb. Redesign the rest.
Before AI, a manager spent most of their time checking whether the junior could produce decent work. After AI, the junior produces the work easily. The manager's time is now spent on something harder — checking whether the AI-assisted output is correct.
Take a junior in your accounts team. Before AI, they took two days to reconcile a vendor account, occasionally with a small error you would catch on review. After AI, the same junior produces the reconciliation in two hours. The output looks polished but is occasionally wrong — and confidently so — in a way that is harder to spot. The AI may have matched transactions that should not have been matched, or applied a GST rate that no longer exists. The junior has not noticed. The manager now has to check not the junior's work but the AI's reasoning.
AI does not free up time for managers. It intensifies it. Output volume goes up, expectations go up, and verification load goes up faster than either. The most experienced people in your firm become indispensable in too many directions. They burn out. They leave.
Protect your senior people's focus time. Schedule one weekly meeting where the team talks through what is happening on the ground, without AI tools in the room. This is what stops your senior team from breaking down.
There are two ways to use AI. One is to put AI alongside your people, so each person gets more done. The other is to hand a process the people used to handle entirely to the AI. Both produce productivity gains. They produce very different companies in five years.
The Anthropic Economic Index, which tracks how AI is used across countries, has found a striking pattern. Firms in developed economies use AI alongside their people. Firms in India tend to hand the work over entirely. In a year, the second pattern has risen sharply here.
Take a hospital appointment desk. The first hospital keeps its front-desk staff and gives them an AI assistant. Staff still handle the difficult cases — the elderly patient who is confused, the angry walk-in, the unusual case the AI does not recognise. AI handles routine bookings, reminders, follow-ups. The hospital now serves twice the patients with the same team. The second hospital replaces the front desk with an AI voice agent. Routine bookings get handled well. But the elderly patient is turned away, the angry patient is mishandled, the unusual case is misclassified. The hospital saves on salary in year one. By year three, it has lost a noticeable share of its patients. The leadership never quite figures out why.
For every major process, make this choice explicitly. The right answer for client-facing and judgement-heavy work is to keep the people and add AI alongside them. The right answer for repetitive back-office work is to let AI handle it.
AI productivity gains do not arrive in the first quarter. They follow a pattern — a short dip while the team learns and the processes are rewritten, followed by a sharp acceleration. Firms that fail at AI adoption usually fail in the dip.
A 40-person services firm in Kochi we spoke to adopted AI for proposal writing in early 2025. In the first month, proposals took longer to produce, not shorter, because the team was learning to prompt the tool. Margins compressed. Leadership came close to pulling the plug. They did not. By the third month, the prompts had been figured out. By the sixth month, proposal turnaround had halved. By the ninth, the firm was bidding on twice as many projects with the same team.
Plan AI adoption with a six-to-twelve-month productivity dip in your accounts. Tell your leadership team the dip is expected.
The biggest mistake Indian SMEs are making with AI is delegating the transition to the IT team, to a software vendor, or to whoever in the office is considered “the computer person.” This is a category error. AI is not a software upgrade. It is a strategic choice about what kind of business you are building over the next decade. It needs founder-level ownership.
Treated as an IT initiative, AI defaults to cost-cutting. Treated as a strategic initiative, it produces growth. Use AI to shrink, and you will run a smaller version of the business you have today. Use it to grow, and you will run a different business in five years. Both feel rational. Only one is.
The authors of this article will lead Dhanam Business Media’s hands-on AI workshop for business leaders in Kochi on Tuesday, 28 July 2026. For more details: Click Here
About the authors: Zubin Kabeer T is Partner at Art of Consulting and holds an MBA from IIM Kozhikode, while Arun S Babu is Director of Habit Theory and has studied Neuropsychology at the University of Cambridge Institute of Continuing Education.