AI Impact on Small Businesses in India: What AI Actually Does
- Mahendra Rathod
- Jun 30
- 30 min read
What AI Actually Does to Small Businesses in India
Three stops, one country
A few years ago, driving on the Srinagar–Sonmarg highway, I stopped at a dry fruit shop called Kissan Kesar. Kashmir being Kashmir, I bought everything — dry fruits, kesar, whole spices, the works. Before leaving, I asked the shopkeeper if I could order again sometime. He said, "WhatsApp karo is number pe."
A few months later, I did. He sent me a product catalog — not a link, not an app, just a catalog inside WhatsApp, with prices and photos and an Add to Cart option I didn't know WhatsApp had. I placed an order. Paid on UPI. Three days later, the package was in Bangalore.
No e-commerce platform. No logistics startup. No middleman taking 30%. Just a man in Srinagar, a smartphone, and a payment system that India built for exactly this kind of transaction.
I thought about that shopkeeper for a long time afterward. He had, without knowing it, solved a distribution problem that business school case studies spend chapters on. And he had done it with tools that cost him almost nothing.
But here is the question I couldn't quite answer: what does AI do to him?
Now consider two other people.
Somewhere in Nagpur, there is a travel agent — let us call her Meena — who has been booking Char Dham packages for 22 years. She knows every bus operator, every dharamshala, every local guide between Yamunotri and Kedarnath. Her clients are not young tech-savvy travellers. They are retired couples, joint families doing their first big pilgrimage, elderly parents whose children live abroad and want someone reliable to take care of things. Meena does not just book tickets. She worries on their behalf. When the Kedarnath road washes out in July, she is on the phone at 6 AM rerouting everyone before her clients even wake up.
MakeMyTrip's AI can plan a Char Dham itinerary in 40 seconds. It cannot call the dharamshala in Gangotri to confirm the hot water situation.
And in Surat, there is an embroidery entrepreneur — let us call him Rajan — running a 10-machine unit in the textile cluster that supplies to garment exporters. His clients are fixed. His thread suppliers are fixed. His workforce of 14 people has been with him for years. Rajan's anxieties are the price of rayon thread, the rupee-yuan exchange rate, and whether the power cuts this summer will be worse than last year. ChatGPT does not feature anywhere in this list.
Three people. Three completely different relationships with AI. None of them are what the headlines say.
The headlines say AI is coming for everyone. The reality is more specific, more interesting, and — depending on who you are — either more reassuring or more alarming than the general claim.
This post is about the 7.34 crore small businesses that form the actual spine of the Indian economy. Not the startups. Not the unicorns. The restaurants and repair shops and distributors and weavers and CA firms and kirana stores that employ 35 crore people and produce nearly a third of India's GDP — most of them invisible to any formal data system, operating on trust, cash, and relationships built over decades.
AI is coming for some of them. For others, the real threats come from somewhere else entirely. And for a few — like the man in Srinagar with his WhatsApp catalog — AI is quietly working alongside them in ways no headline has captured yet.
That line is not as obvious as it sounds. In this post, we look at who India's small businesses actually are, what AI is doing to their world, and where the real threats and opportunities lie — sector by sector, ground up.
Section 1 — First, let's agree on who we're talking about
Before we talk about AI, we need to agree on something more basic — what exactly is a small business in India?
The question matters because the answer is not what most people expect.
The government classifies businesses by two numbers: how much they have invested in plant and machinery, and what their annual turnover is. As of April 2025, after a revision announced in the Union Budget, the classification looks like this:
A Micro enterprise invests up to ₹2.5 crore and earns up to ₹10 crore a year. A Small enterprise invests up to ₹25 crore with turnover up to ₹100 crore. A Medium enterprise goes up to ₹125 crore in investment and ₹500 crore in turnover. Together, these three tiers form the MSME sector — Micro, Small and Medium Enterprises. Anything above ₹500 crore in turnover is broadly a large enterprise.
Worth noting: this definition has been revised upward twice in five years — in 2020 and again in 2025 — which means businesses that were "small" last year may now be classified as "micro." The goalposts move, but the people inside them don't change.
Now for the numbers.
As of 2025, India has an estimated 7.34 crore businesses in the MSME category. That is 73.4 lakh tens of enterprises — more than the entire population of France or the UK, each running some kind of business.
Of these, 97% are micro enterprises. Not small. Not medium. Micro. A business with one machine, one shop, one skill. The data makes this concrete: in the most recent detailed survey, micro enterprises across India employed an average of fewer than 2 people each. The typical micro enterprise is not a team. It is a person — perhaps two — working at something they know how to do.
This matters for how we think about AI. When people imagine a small business owner, they often picture someone managing a team, running operations, worrying about processes. The reality, for 97% of this sector, is a man operating a lathe in Rajkot or a woman running a tiffin service in Pune. The business is the person.
The aggregate numbers are striking. This sector contributes 30.1% of India's GDP, 45.7% of exports, and employs around 35 crore people — roughly 5 crore enterprises with fewer than 2 people each, and the remainder ranging upward. Second only to agriculture as an employer. Every third rupee of India's economic output connects back to a business that most formal economic analysis overlooks.
The invisible layer
Here is where the numbers get complicated.
The 7.34 crore figure comes from a SIDBI estimate that includes both registered and estimated unregistered enterprises. But the actual informal economy sits well beyond this. A significant share of micro businesses — pan shops, vegetable vendors, cycle repair stalls, home-based embroidery units — have no formal registration of any kind. No GST number. No PAN. No Udyam registration. They are invisible to every database, including this one.
When we talk about AI's impact on Indian MSMEs, we are largely talking about the businesses we can see. The ones we cannot see will feel the effects just as much — but no one is measuring them.
Two very different universes inside the same number
Not all MSMEs face the same world, and understanding the difference is important for everything that follows.
The first is the consumer-facing MSME — the restaurant, the kirana store, the travel agent, the salon, the coaching class. Their customer is a person. Their competition is visible. Their relationship with AI is direct — it shows up in how customers find them, order from them, and review them.
The second is the supply-chain MSME — and this is a world most people outside industry never see.
Take Rajkot, in Gujarat. To most people it is known for its food and its Navratri. But Rajkot is also one of India's most important manufacturing clusters. Around 60 aluminium foundry units operate here, supplying precision castings to OEMs including Mahindra & Mahindra, Kirloskar Oil Engines, and Godrej. There is a separate cluster of roughly 150 bearing manufacturers employing around 15,000 people. The auto components cluster alone has about 160 MSME units — 80 of them micro, 65 small, 15 medium.
None of these businesses sell to you directly. Their customer is another business. They have no storefront, no social media presence, no Swiggy listing. Their entire existence depends on a supply chain relationship — a purchase order from a Tier-1 supplier, a quality certification from an OEM, a price negotiated once a year that determines whether they run their machines or shut them down.
For these businesses, AI's arrival is not a question of chatbots or WhatsApp catalogs. It arrives through their buyers' requirements — and through macroeconomic forces that have nothing to do with technology at all.
The two types of MSME need to be kept separate. They face different threats, have different opportunities, and will need very different kinds of support.
Section 2 — What does this economy actually do?
If you drive from Srinagar to Kanyakumari, stopping at every town along the way, you will not find a single stretch of inhabited India that does not have a cluster of small businesses running. A row of welding shops behind a bus depot in Ludhiana. Silk weaving units in the lanes behind Varanasi's ghats. Printing presses in the bylanes of Mehrauli. Cold storage units on the outskirts of Nashik where onions wait for better prices. Embroidery frames in apartments in Surat. Fishing net repair shops on the Kerala coast.
This is the economy that the MSME number is trying to count. 7.34 crore enterprises. The table below organises them into categories — but a table cannot fully capture what they actually are, which is the accumulated economic ingenuity of a billion people solving their own problems with whatever was available.
A few things worth drawing out from the data that the table alone does not show.
Manufacturing is the backbone, but it is largely invisible
Textiles, food processing, auto-components, engineering goods, leather, chemicals — this is where the bulk of India's MSME manufacturing output comes from. And almost none of it has a brand name you would recognise.
The garment you are wearing almost certainly passed through an MSME at some point — a yarn spinner, a weaving unit, a dyeing house, a stitching unit, a packing shed. None of them put their name on the label. India's manufacturing MSMEs are deeply embedded in supply chains that end in someone else's brand. Their value is invisible by design.
This is important for the AI conversation because it means their fate is not entirely in their own hands. A Tiruppur knitwear unit does not decide to adopt AI-based quality control because it read an article about it. It does so because its buyer — an export house supplying a European retailer — demands a defect rate below a threshold that manual inspection cannot reliably achieve. AI enters the supply chain MSME from the top down, not the bottom up.
Trade is the widest category — and the most threatened
The kirana store, the wholesale distributor, the commodity broker, the stockist — these are businesses whose core activity is moving goods and information from one place to another. Buy here, sell there. Know what is available. Know who needs it. Take a margin for the knowledge and the effort.
India has an estimated 1.2 crore kirana stores alone. They are the most democratic retail network in the world — open early, open late, selling on credit to neighbours, stocking what the neighbourhood actually uses rather than what a planogram says it should.
The distributor sitting between the manufacturer and the retailer is a similar story. His value, historically, has been information and logistics — he knows which retailer needs what, when, at what price, and can move product efficiently across a territory. That value proposition is exactly what B2B platforms like Udaan and Jumbotail, and manufacturer-direct apps, are systematically dismantling.
Services are the fastest-growing category — and the most diverse
Restaurants, repair shops, CA firms, travel agents, coaching classes, salons, freight brokers, printers, small ad agencies — these are businesses that sell time and skill rather than a product. They are also the category most recently included in the MSME definition, which previously focused on manufacturing.
The services MSME is where India's economic aspiration is most visible. The first-generation college graduate who opens a CA practice in a small town. The woman running a cloud kitchen from her home in Bengaluru. The mechanic who started his own garage after ten years at a dealership. These businesses represent genuine upward mobility — and they are also the businesses most directly in the crosshairs of AI-powered platforms eating their value proposition.
The rural-urban split that most analyses miss
In rural areas, manufacturing dominates. In urban areas, trade takes precedence. Services are relatively evenly distributed across both.
This split matters enormously for the AI conversation. A rural manufacturing MSME — the rice mill, the brick kiln, the agricultural processing unit — has almost no direct exposure to AI today. Its constraints are power supply, road access, raw material prices, and labour. AI is the least of its problems.
An urban trade MSME, on the other hand, is already living inside an AI-shaped world — even if it does not know it. The algorithm that determines whether its products appear on the first page of a marketplace search result. The dynamic pricing engine that a competitor is using. The chatbot that a customer is talking to instead of calling. The world has already changed around it. It just has not updated its own operations yet.
The geography of concentration
Maharashtra leads with 3.71 lakh registered MSMEs, followed by Tamil Nadu at 2.17 lakh and Uttar Pradesh at 2.03 lakh. Together these three states account for nearly 35% of all listed enterprises.
But concentration does not mean the rest of the country is empty. West Bengal has 1.8 lakh MSMEs — largely in trade and cottage manufacturing. Rajasthan and Gujarat each cross 1.5 lakh, with strong manufacturing clusters. The Northeast states have among the lowest absolute numbers but high per-capita MSME density, driven by artisanal and agricultural processing activities.
The state map matters because AI adoption will not arrive uniformly across this geography. Maharashtra, Karnataka, and Tamil Nadu will see AI tools reach their MSMEs faster. The MSME clusters of Bihar, Jharkhand, and the Northeast will follow years later, if at all. That lag is not a technology problem. It is an infrastructure, literacy, and awareness problem.
Section 3 — The AI impact on small businesses in India — what is actually happening right now
To understand the AI impact on small businesses in India, we need to start not with the headlines but with the ground.
Most writing about AI and small businesses falls into one of two traps. The first is breathless optimism — AI will empower every kirana store, every weaver, every street vendor, transforming them into globally competitive micro-enterprises. The second is apocalyptic alarm — AI is coming for everyone, and small businesses will be the first to fall.
Neither is accurate. And neither is useful.
What follows is not a projection. It is a ground-level account of what AI is actually doing, today, to businesses of different kinds — in India and in countries that moved faster.
What AI is best at — the global picture
Globally, AI has proven most effective at tasks that share three characteristics: they involve large volumes of repetitive work, they rely on information that can be digitised, and the output can be checked against a right or wrong answer.
This is why the first wave of disruption hit where it did:
Major outsourcing companies have reduced data entry headcount by 30 to 40% since 2024, as AI-powered document processing has made manual data entry obsolete for most standard document types
Media and communications job postings have fallen to 64.1% of their previous level — the lowest of any professional sector tracked — with 15,000 media job cuts last year, as AI writes articles, transcribes audio, generates images, and manages social media
Customer service representatives face an 80% automation probability for routine query handling
At the other end, jobs least exposed to AI automation tend to involve physical judgment, environmental variability, and human interaction — construction workers, electricians, plumbers, food service, and personal care roles consistently rank among the least vulnerable.
This is the global version of the garage mechanic insight. The mechanic's job involves diagnosing a problem on a car he has never seen before, in conditions he did not control, using tools held in human hands. AI can help him look up a fault code. It cannot turn the wrench.
What is actually available to Indian small businesses today
India's small business AI story is not about frontier models or enterprise software. It is about tools built on top of infrastructure that India already has — UPI, WhatsApp, GST, and a smartphone in almost every pocket.
WhatsApp as the operating system. With 48.7 crore WhatsApp users in India, the platform is not a messaging app for small businesses — it is the primary channel for customer communication, order taking, catalog sharing, and payment collection. AI now sits on top of this. AI chatbots responding in Hindi, English, and regional languages convert 40 to 60% more leads than manual messaging, while reducing customer service headcount from 3 to 5 agents to 1 agent plus AI.
GST compliance as the entry point. Every registered MSME above a turnover threshold must file GST returns. AI-integrated accounting tools cut GST compliance time from 8 to 12 hours monthly to under 2 hours. Small businesses that previously spent ₹15,000 to ₹30,000 annually on CA fees for basic compliance now spend ₹5,000 to ₹10,000. The tools doing this — Vyapar, Zoho Books, Tally with AI extensions — cost between ₹500 and ₹2,000 a month.
Marketing and content, democratised. A single-person saree business in Surat can now produce Instagram content, product descriptions, and promotional copy using Canva and ChatGPT at a combined cost of under ₹2,500 a month. Three years ago, the same output would have required a freelance designer and a content writer.
Platform intelligence flowing downward. Zomato and Swiggy now provide restaurant partners with AI-generated insights about their menu performance, customer ordering patterns, peak hours, and price sensitivity. The restaurant owner does not need to understand the algorithm. He just needs to read the dashboard and act on it.
Credit, slowly unlocking. India currently faces a ₹30 lakh crore MSME credit gap. AI-based credit scoring using GST filing data, UPI transaction patterns, and ONDC sales history is beginning to change this — not yet at scale, but the direction is clear.
The honest assessment
65% of Indian MSMEs lack awareness of available AI tools
70% lack access to skilled AI professionals
59% cite budgetary limitations as the primary barrier
"The businesses that will benefit most from AI in the next five years are not the ones that develop an AI strategy. They are the ones that adopt one tool, use it consistently, and build from there."
The Kissan Kesar shopkeeper did not develop a digital commerce strategy. He said yes to WhatsApp.
For a broader view of how AI is reshaping employment globally, read: The Future of Jobs Report 2025 — What Lies Ahead
Section 4 — Where countries that moved faster ended up
The most useful thing about watching other countries adopt AI before us is not the adoption numbers. Numbers tell you how many businesses raised their hand. They do not tell you what happened next — whether the tool actually worked, whether the business model held, whether the jobs that disappeared came back in another form.
Here is what actually happened on the ground in three countries that moved faster than India.
Singapore — the compliance-first model
Singapore is the most cited example of successful SME AI adoption. The government did not wait for businesses to figure it out themselves. It subsidised specific tools, mandated sector-level digital plans, and built a grant infrastructure that made adoption financially low-risk.
What Singapore's SMEs actually used AI for:
Customer service automation — chatbots handling queries, appointment booking, after-hours responses
Finance and accounting — invoice processing, GST-equivalent compliance, expense tracking
Inventory and demand forecasting — particularly in food and beverage, where waste is a direct cost
Quality control in food manufacturing — defect detection reducing rejection rates by up to 32%
The top business functions where SMEs deployed AI were IT, customer service, and finance and accounting.
What changed for businesses: the SMEs that adopted AI did not fire their staff. They redeployed them. The person who used to answer the same 40 customer questions every day now handles escalations and relationships. The job changed; the person stayed.
"Even in Singapore — the world's second-highest AI adopter — only 14.5% of SMEs had actually deployed AI as of 2024. Broad-based adoption, even in the most digitally advanced economies, takes a decade, not a year."
Vietnam — survival pressure as the adoption engine
Vietnam's AI story is less organised than Singapore's and more instructive because of it. In 2024, five new Vietnamese enterprises adopted AI every hour, driving a 39% year-on-year increase. This was not driven by government grants or awareness campaigns. It was driven by margin pressure.
Vietnamese retail margins are thin. Waste in fresh food distribution runs at 30 to 40%. Any tool that reduced either problem paid for itself quickly. So businesses adopted:
AI demand forecasting to reduce inventory waste in fresh produce
Recommendation engines on e-commerce platforms to improve conversion
Computer vision for quality grading in agricultural supply chains
"Vietnamese businesses did not have an AI strategy. They had a waste problem, a margin problem, a query-volume problem. AI happened to be the cheapest solution available."
UAE — the logistics and customer service playbook
The UAE's SME AI story is shaped by its economy — a trade and services hub with high labour costs. The problems AI solved here were specific:
Real estate agents used AI chatbots to screen and qualify leads
Retailers automated inventory management and seasonal demand forecasting
Logistics firms used AI for route optimisation and delivery scheduling
The lesson for India: high national AI adoption rankings do not automatically translate into MSME benefit. The gains concentrate where transaction volumes are high, digital infrastructure is strong, and the business has enough margin to absorb the implementation cost.
The pattern across all three
Across Singapore, Vietnam, and the UAE, the same picture emerges when you look at what small businesses actually used AI for:
Answering customer questions they were tired of answering repeatedly
Filing paperwork that was mandatory, time-consuming, and added no value
Predicting demand so they did not over-order stock that would spoil or sit unsold
Sorting and grading physical products faster than human hands could manage
Notice what is absent from this list: strategy, creativity, relationship management, physical production, and trust. Every one of those remained human.
"AI replaced the tedious parts of running a small business. It did not replace the reason the business existed in the first place."
What this means for India
India has something none of these three countries had at scale when they began their AI journey: UPI, WhatsApp, and GST infrastructure already embedded into the daily operations of crore of small businesses.
The Kissan Kesar shopkeeper was already on WhatsApp. The kirana store owner was already on UPI. The CA firm was already filing on the GST portal. The infrastructure for AI to sit on top of already exists. What is missing is the awareness that the tools are available, the confidence to try them, and in many cases the language — most AI tools still work better in English than in Hindi, Tamil, or Gujarati.
Section 5 — The squeeze
There is a concept in physics called compression. Apply force from two sides simultaneously, and whatever is in the middle has nowhere to go.
That is what is happening to the middle layer of India's small business economy. Not slowly. Not eventually. Now.
To understand it, picture India's economy as a three-layer structure.
The top layer is large platforms, corporations, and AI-powered services. Amazon, Flipkart, MakeMyTrip, national banks, big CA firms, GST portals, online travel aggregators, B2B marketplaces. These businesses are getting smarter, faster, and more direct. They are using AI to reach customers and suppliers without needing anyone in between.
The bottom layer is the actual producer and the actual consumer. The Kissan Kesar shopkeeper in Srinagar. The aluminium foundry in Rajkot. The home baker in Coimbatore. The farmer sorting mangoes outside Ratnagiri. These are the people who make things, grow things, cook things, repair things. Their value is physical and specific.
The middle layer is everyone in between. And in India, the middle layer is enormous.
The distributor who carries goods from the factory to the retailer. The travel agent who assembles a package tour. The small CA firm that files your GST returns. The local printer who makes your visiting cards. The stockist. The commission agent. The booking clerk. The catalogue salesman who visits retailers every Tuesday with a physical order book.
These businesses share one fundamental characteristic: their core value is knowing something, or accessing something, that their customer could not easily get on their own. They are information intermediaries. And that is precisely what AI does — cheaply, at scale, without lunch breaks.
"The middleman's value proposition has always been: I know something you don't, and I will connect you for a fee. That sentence is a near-perfect description of what a large language model does."
The force from above
Large platforms are not waiting for small businesses to catch up. They are actively removing the need for the middle layer:
MakeMyTrip and Ixigo now offer AI trip planners that build complete itineraries, compare options, and book end-to-end in minutes
Udaan, Jumbotail, and manufacturer-direct apps are letting FMCG and pharma companies reach retailers without going through a distributor — the stockist who used to earn single digit margins for knowing which retailer needed what is being bypassed by an algorithm that knows the same thing better
Zoho, Vyapar, and Tally with AI extensions are automating the GST filing, invoice matching, and basic bookkeeping that kept small CA firms afloat
Canva AI and ChatGPT have democratised basic design and copywriting — the small printing and design shop that used to charge ₹3,000 for a social media kit is competing with a tool that produces the same output in three minutes for ₹200
Each of these is the top layer pressing downward.
The force from below
Simultaneously, the bottom layer — the small producer, the artisan, the specialty retailer — is discovering that they no longer need the middle layer for the basics.
The Kissan Kesar shopkeeper does not need a distributor to reach customers in Bangalore. He has WhatsApp, a UPI QR code, and a catalog he built himself. The home baker in Coimbatore does not need a marketing agency — she has Instagram Reels and Canva. The Rajkot foundry unit, if it gets on GeM or ONDC, can access government procurement or new buyers directly, without a commission agent taking a cut for making the introduction.
The tools that used to require a team — marketing, customer service, invoicing, catalog management — now cost between ₹500 and ₹3,000 a month and can be operated by one person on a smartphone.
Each of these is the bottom layer pushing upward.
What gets squeezed
The sectors under the most pressure:
Wholesale distributors — route-to-market is being rebuilt digitally by manufacturers and B2B platforms
Small CA firms and local marketing agencies — routine compliance and basic content work is being automated; only advisory work and complex cases remain defensible
Small web development shops — AI-generated websites and no-code platforms have commoditised basic web presence; only custom development and ongoing maintenance hold value
Standard-package travel agents — OTA AI handles straightforward bookings; only complex, relationship-dependent packages survive
Local printers and designers — basic design and print-on-demand is now self-service
Commodity brokers and aggregators — price discovery and matching is moving to digital platforms
"These are not businesses being disrupted by a superior competitor. They are businesses being disrupted by the disappearance of the problem they were solving."
What does not get squeezed — and why
The compression does not touch everything equally. Three types of businesses are largely insulated from it, at least for now.
The first is any business where the delivery is physical and irreplaceable. The garage mechanic. The dhaba cook. The wedding photographer. The embroidery unit in Surat. Rajan's business is not threatened by AI. It is threatened by power costs, thread prices, labour wages, and competition from automated looms. Those are real threats — but they are not AI threats.
The second is any business where the relationship is the product. Meena in Nagpur does not just book Char Dham packages. She is the person her clients call at 6 AM when the road to Kedarnath washes out. Research describes these as "phygital" journeys — where digital research is paired with human expertise. The planning happens on Google. The trust is placed in Meena.
The third is the authentic producer with direct reach. The Kissan Kesar shopkeeper is not being squeezed. He is being empowered. AI and digital tools have given him something the middle layer used to control — access to the end customer.
The irony at the heart of the compression
The same technology that is compressing the middle layer is also available to the people being compressed, often for less than the cost of a mobile data plan. The small CA firm that is losing basic GST clients to Zoho can use Zoho itself to serve those clients more efficiently and move up to advisory work. The travel agent losing standard bookings to MakeMyTrip can use AI to build hyper-personalised pilgrimage packages that no algorithm can replicate.
"AI does not destroy businesses. It reveals which part of the business was real and which part was friction that technology has now removed."
Section 6 — Three types of small business, three futures
Not every small business in India faces the same AI reality. India's 7.34 crore small businesses fall into three broad types — not by sector, not by size, but by the nature of what they actually do and where their value actually sits.
Type A — AI empowers them
These are businesses where the delivery is physical, authentic, or deeply local — and AI handles everything around it.
The Kissan Kesar shopkeeper's core value is the quality of his kesar and the trust that comes from buying directly from someone in Kashmir. What WhatsApp, UPI, and digital catalogs did was remove the distribution barrier that previously made it impossible for him to reach a customer in Bangalore.
This pattern repeats across many categories:
Specialty retailers — dry fruits from Kashmir, silk from Varanasi, spices from Kerala, handicrafts from Rajasthan. Authentic provenance is defensible. AI helps them reach buyers they could never have accessed before.
Handloom and artisanal units — the Pochampally weaver, the Channapatna toy maker, the Kutch embroiderer. Authenticity is the product. Instagram discovery, ONDC marketplace access, and AI-generated product descriptions are net positives.
Food processing micro units — the pickle maker in Andhra, the papad unit in Rajasthan, the millet flour mill in Karnataka. WhatsApp B2B ordering and ONDC access open new markets without requiring a distributor.
"For the authentic producer, AI is the great leveller. It removes the advantage that scale and geography used to give to the middleman."
Type B — AI is largely a spectator
These are businesses where the real work is physical, the real threats are macroeconomic, and AI is at best a peripheral tool.
Rajan's embroidery unit in Surat is the clearest example. Ten machines. Fourteen workers. Fixed clients. Fixed vendors. His margins are squeezed by the price of rayon thread. His labour costs go up every year. His biggest existential risk is not ChatGPT — it is the automated loom facility in China producing similar output at a fraction of the cost.
This pattern covers a wide range of businesses:
Auto-component and engineering MSMEs — the Rajkot foundry, the Ludhiana machine tool unit, the Pune precision component manufacturer. Their threat is quality requirements moving beyond manual inspection capability, EV transition reducing demand for certain components, and global supply chain shifts.
Construction and fabrication units — physical presence, physical skill, physical output. AI can help with estimation and scheduling. It cannot lay a brick.
Agricultural processing — rice mills, oil expellers, cold storage operators. Their constraints are power, logistics, raw material prices, and monsoon variability.
Local repair shops — the garage, the electronics repair unit, the appliance service centre. AI can look up the fault code. The hands remain human.
For Type B businesses, there is one AI use case that should not be ignored: platform access. Getting on GeM, ONDC, or B2B marketplaces with AI-assisted onboarding can open new buyers without a commission agent in between.
"For the physical producer, AI is not the story. It is a footnote. The real story is whether global supply chains, input costs, and government policy will give them room to survive and grow."
Type C — AI is an existential question
These are the businesses that need to think hardest and fastest. Their core value proposition is informational — and that is precisely what AI does, at scale, for almost nothing.
The sectors in this category:
Small CA firms doing routine compliance work — GST filing, ITR preparation, basic bookkeeping, company registration. Zoho, Vyapar, and Tally AI are automating exactly this. The junior accountant billing hours for data entry is the most exposed person in the Indian professional services economy right now.
Small web development shops — basic websites, landing pages, and e-commerce setups are now buildable through AI-assisted no-code platforms. Only custom development and ongoing technical maintenance hold defensible value.
Local marketing agencies doing standard work — social media calendars, ad copy, basic graphic design. AI does this faster and cheaper. What survives is strategy, brand understanding, and client relationships — not execution.
Wholesale distributors — single digit margins, relationship-based territory management. B2B platforms and manufacturer-direct apps are rebuilding route-to-market digitally.
Standard-package travel agents — the agent booking a standard Goa package for a young couple who researched everything on Instagram is already losing that customer to OTAs. What survives is Meena — the Char Dham specialist who manages complexity, anxiety, and elderly clients who want a human being responsible for their journey.
Local printers and designers — basic visiting cards, banners, social media templates. Canva AI has commoditised this. What survives is production — the physical printing — not the design.
For Type C businesses, the question is not whether to change. The question is whether to change before the revenue disappears or after.
"The Type C business is not dying because it is bad at what it does. It is being made redundant because what it does is no longer scarce. The response is not to do it better. It is to do something adjacent that is still scarce."
A word on overlap
These three types are not clean boxes. Many businesses sit across two. The travel agent who books standard packages is Type C. The same travel agent who handles complex religious group tours for elderly clients is Type A. The Rajkot foundry that supplies standard castings is Type B. The same foundry that moves into precision aerospace components requiring AI-based quality certification is becoming Type A.
The businesses that will navigate this period best are the ones that consciously understand which part of what they do belongs to which type — and invest accordingly.
Section 7 — The squeeze also creates space
Every major economic shift destroys some businesses and creates others. The internet did not just kill travel agents and video rental stores — it created e-commerce, digital marketing, cloud computing, and millions of jobs that had no name in 1995.
AI is no different. The compression of the middle layer is real. But compression in one place creates space somewhere else.
The creator as a micro-enterprise
The most visible new category is the individual content creator. Strip away the influencer glamour and what you have is a micro-MSME — one person, one skill, one audience, generating income directly without an employer or a middleman.
India's creator economy is worth an estimated ₹20,000 crore in 2025 and projected to exceed ₹40,000 crore by 2027. India currently has around 20 to 25 lakh monetised digital creators who influence more than 30% of consumer purchase decisions.
These numbers include:
The Kannada-language YouTube channel teaching GST filing to small business owners in Tier 2 towns
The Hindi-medium educator explaining stock market basics to first-generation investors in Patna
The Tamil home cook with 80,000 followers who sells her own spice blends directly through Instagram
The retired engineer in Pune running a technical training channel for ITI students
Each of these is a business. Each generates income. None of them existed as a category five years ago.
"The creator economy is not a media story. It is an MSME story. Lakhs of Indians have become micro-enterprises by turning a skill or a passion into a direct relationship with an audience — and AI is the production team they could never have afforded to hire."
The one-person agency
One person with strong domain knowledge and a suite of AI tools can now:
Write, design, and publish a client's monthly content calendar
Build a functional website with e-commerce capability
Run Google and Meta ad campaigns with AI-assisted targeting and copy
Analyse campaign performance and produce a client report
What used to require a team of four or five now requires one person and ₹5,000 to ₹8,000 a month in tool subscriptions.
Across India's Tier 1 and Tier 2 cities, a new class of micro-agency is forming — individuals who left larger agencies or IT companies, took their domain expertise with them, and are now serving 8 to 10 small business clients at pricing that undercuts agencies but at margins that would be impossible without AI.
"The one-person agency is not a freelancer who got lucky. It is a new business model that AI made viable — one skilled person, the right tools, and clients who care more about results than headcount."
New service businesses serving the AI transition itself
The consultant who helps a cluster of kirana stores in Jaipur set up WhatsApp Business automation
The trainer running workshops in Surat teaching embroidery unit owners how to use ONDC
The vernacular AI tools specialist helping CA firms in Coimbatore transition from manual GST filing to Zoho
The local tech person who sets up and maintains AI tool stacks for small restaurants and retail stores
India has a long tradition of this — the person who sits outside the passport office filling forms, the person who navigates the railway booking system for those who cannot. The AI transition will generate its own ecosystem of guides, translators, and fixers. In a country with 7.34 crore small businesses and a massive awareness gap around AI tools, that ecosystem has a very large market to serve.
The honest constraint
It would be dishonest to present this new business creation as a clean offset to what is being lost.
The wholesale distributor being bypassed by Udaan is unlikely to become a content creator. The junior accountant whose data entry work is being automated by Zoho is not going to pivot into an AI tools consultancy next month. The skills required for the new businesses are different from the skills embedded in the businesses being compressed.
The risk of displacement is likely to be greatest for informal employees — making up almost 90% of India's labour force — who carry out low-skilled, repetitive jobs in manufacturing and retail. These are not people who will smoothly transition into creator economies or one-person agencies.
"AI creates new businesses at the top of the skill distribution and destroys old ones at the bottom. The distance between those two points is not crossed by motivation alone. It requires deliberate policy, time, and investment in people."
Section 8 — What the government can do
Policy discussions about AI and MSMEs tend to fall into one of two modes. The first is the grand vision. The second is the scheme announcement. Neither is what this section is about.
What follows is a ground-level assessment of five specific levers that could meaningfully change the AI reality for India's small businesses in the next three to five years.
1. ONDC — finish what was started
The Open Network for Digital Commerce is, quietly, one of the most important policy instruments India has built for small businesses in a generation. It is an open protocol that allows any seller to list on any buyer-facing app without being locked into a single platform's terms, margins, or algorithms.
What needs to happen:
Simplified onboarding for micro enterprises — today the process requires documentation and digital literacy that many micro businesses do not have
Vernacular language support across all buyer and seller interfaces
AI-assisted catalog creation so a small producer can describe their product in their own words and have it converted into a marketplace-ready listing automatically
Integration with Udyam registration so that a registered MSME is one step away from being discoverable on ONDC
"ONDC is not a government e-commerce platform. It is infrastructure — like a road. The government's job is to build it well and maintain it. What businesses do on top of it is their own."
2. The IndiaAI Mission needs a last-mile programme
The IndiaAI Mission, approved in March 2024, has a budget of ₹10,371 crore over five years, focused on computing access, innovation support, and dataset development. This is well-designed for startups, researchers, and large enterprises. It is not designed for a flour mill owner in Nagpur or a garment unit in Tiruppur.
What the micro enterprise needs:
Awareness that tools exist and are affordable — a ₹500/month WhatsApp bot, a ₹750/month GST automation tool
Demonstration in their language, in their cluster, by someone they trust
A subsidy or voucher that covers the first six months of a tool subscription so the adoption barrier is practical, not just informational
3. AI-based credit scoring — formalise it and fund it
The ₹30 lakh crore MSME credit gap is not going to be closed by traditional banking. Banks require collateral, financial statements, and credit history. Most micro enterprises have none of these in the form banks recognise.
But they have something else — three years of UPI transaction data, a GST filing history, ONDC sales records. These are, in aggregate, a more accurate picture of a business's health than a balance sheet prepared to minimise tax.
What is needed:
RBI policy that explicitly recognises alternative data credit scores for MSME lending under priority sector norms
Incentives for banks and NBFCs to use these scores rather than defaulting to collateral requirements
A public credit data utility that aggregates MSME transaction data with consent and makes it available to lenders in a standardised form
"A kirana store with five years of consistent UPI transaction history is a better credit risk than any document it can produce. The financial system has not yet learned to read what it already knows."
4. Reskilling — specific, sector-linked, and honest
What useful reskilling for MSME-adjacent workers actually looks like is much more specific than current programmes suggest:
The small CA firm needs training in AI-assisted audit, advisory services, and financial planning — not generic prompt engineering
The travel agent needs training in curated experience design and complex multi-leg itinerary management — not how to use MakeMyTrip
The wholesale distributor whose territory is being digitised needs training in demand planning, inventory analytics, and category management
The junior accountant whose data entry work is being automated needs to learn exception management and AI output review — the supervisory layer above the tool
Skill India and PM Vishwakarma are large programmes with genuine reach. The gap is in sector-specific curriculum designed around what the job will actually look like in three years, not what it looked like five years ago.
5. The unregistered economy — bring it in, do not just count it
97% of India's MSMEs are informal enterprises that neither fall within the GST framework nor have business PANs. These businesses cannot access credit schemes, subsidy programmes, AI tool vouchers, or ONDC onboarding support — because they do not officially exist.
The policy question is not how to count the informal economy. It is how to make formality worthwhile:
Guaranteed credit access within 30 days of Udyam registration, even for micro amounts
Protection from retrospective tax demands for businesses that formalise voluntarily
Local-language, low-documentation onboarding onto ONDC and GeM as an immediate benefit of registration
"The small businesses that remain unregistered are not hiding from the system out of ignorance. They are making a rational calculation that the cost of visibility outweighs the benefit. Change the calculation."
Closing — The question Kissan Kesar did not ask
We started on a highway in Kashmir.
A dry fruit shop. A WhatsApp number written on a piece of paper. A catalog that appeared on a phone screen in Bangalore three months later. A UPI payment. A package that arrived in three days.
At no point did the shopkeeper ask: "How do I build a digital commerce strategy?" He asked: "Do you want to order again?" And then he solved the problem in front of him with the best tool available.
That instinct — solve the specific problem, use the affordable tool, do not wait for a strategy — is the most useful thing any observer of small business can take from the AI moment we are living through.
Through all of this, one distinction keeps returning.
Every small business has two layers. The first is the information layer — how customers find you, how you take orders, how you file compliance, how you manage inventory, how you communicate, how you price. This layer is largely replicable. AI is coming for it regardless of sector, size, or location.
The second is the delivery layer — what you actually make, cook, repair, weave, advise, or build. The kesar that actually tastes like Kashmir. The casting that actually fits the engine. The Char Dham package that actually gets the elderly couple home safely. The embroidery that actually takes twelve hours of skilled human work to produce. This layer is largely irreplaceable.
The businesses that use AI aggressively for the first layer and invest in quality and relationships for the second are the ones that will find the next decade manageable.
"Use AI for the layer that is replicable. Protect the layer that is not. Know the difference between the two."
The irony that nobody is writing about
The Kissan Kesar shopkeeper is already living with AI. Not in his business systems. Not in his supply chain. But his daughter uses Claude to study for her board exams. He reaches for ChatGPT when he needs to draft a message in English to a new customer in Mumbai.
AI has entered his life the way electricity entered his grandfather's life — not through a policy announcement or a digital literacy programme, but through a problem it was useful for, at a price he could afford.
The AI transition for India's small businesses will not look like a corporate transformation programme. It will look like this — incremental, informal, driven by specific utility, spreading through families and clusters and WhatsApp groups before it ever appears in a government survey.
The shopkeeper on the Srinagar highway did not wait for anyone to tell him that WhatsApp could be a business tool. He figured it out because a customer asked him a question and he had a phone in his hand.
That is how this will unfold. Tool by tool. Problem by problem. Person by person.
The only question worth asking is whether the infrastructure, the policy, and the awareness will be ready when the 7.34 crore get there.
Happy Reading!
Sources: SIDBI Understanding Indian MSME Sector 2025; Ministry of MSME Udyam Portal Feb 2026; Ministry of MSME Annual Report 2023-24; IMDA Singapore Digital Economy Report 2025; Vietnam AI Annual Report 2025; CII MSME Employment Data 2021-22; PHD Chamber Structural Analysis 2023; SAMEEEKSHA Rajkot Cluster Profiles; WEF AI Playbook for India SMEs 2025; OECD AI Adoption in SMEs Dec 2025; IndiaAI Mission Budget Notification March 2024.
Comments