In most restaurants, the front of house and the back of house may as well be two different businesses. The billing counter knows what sold. The kitchen and the store know what was bought and cooked. But the two sides rarely meet on the same page, so nobody can answer the question that actually decides whether the place makes money. What did it cost to produce what we just sold? Owners feel this gap every day. They watch revenue climb and still have no idea where the margin is going, because the tools they run were built to handle one side of the house, not connect both.
This is really a symptom of something bigger in how businesses buy software. For a long time the goal was just to digitise. Get off paper, get the numbers into a system. A billing tool here, an accounting package there, each solving its own little corner. That was fine when all you wanted was a record of what happened. But owners have stopped being satisfied with software that only stores their data. They want software that understands their work. And the minute you get into the actual work of a restaurant, the one-corner-at-a-time approach starts to crack.
The trouble with software that tries to fit everyone
Think about what a horizontal platform is being asked to do. One product that has to serve a law firm, a logistics company, a hospital, and a chain of restaurants, all with the same features. So it strips out everything specific and hands you a shell. Rename these fields. Configure these workflows. Looks great in a demo. In real life the buyer spends months, and a fair bit of money, with a consultant just to make it resemble their business. And even then the software doesn’t understand the industry. It only knows what you typed in during setup.
Take restaurants, which is the world I live in. A restaurant doesn’t behave like a generic business. Inventory is perishable and moves fast. A bottle of gin gets opened and poured out by the peg, so you’re counting stock in millilitres, not bottles. The cost of a dish comes from its recipe, and that cost swings with the market. Tomatoes this week are not tomatoes last week. A generic inventory module has no clue what a recipe even is. It sees a line item. It will never tell you that your food cost slipped three points last month because onion prices spiked and nobody repriced the menu. That knowledge was simply never inside the software.
Why the sector-specific players are pulling ahead
This is where purpose-built software has started to win. Instead of a blank shell, you get a system that already knows the shape of your business, with the vocabulary and the workflows and the weird edge cases baked in from day one. A restaurant operator doesn’t have to explain what a kitchen display screen is, or what happens between a purchase order and a goods received note, or why a stock count has to be done storage area by storage area. The software already knows. So the operator is up and running in days, not two quarters and a change-management project.
But there’s a deeper thing here, and it comes down to data. Most Indian businesses don’t have a shortage of data. If anything they have too much, just scattered everywhere. Some sits in the billing system. Some is a photo of an invoice on the supplier’s WhatsApp. Some lives in a spreadsheet on the manager’s laptop. A surprising amount lives in the owner’s head. None of it talks to each other. The POS knows what sold. The inventory system, if there even is one, knows what came in. But nothing connects the two. A sector platform can, because it was built around that connection from the start, not around storage. And that isn’t something you bolt onto a generic tool afterwards. It has to be the thing you begin with.
Now, about AI
Everyone wants to talk about AI, and fair enough. But there’s a lot of muddle about what it can actually do for a restaurant. So let me be blunt. AI does not create margin out of thin air. It cannot stare at a pile of disconnected records and magic up an insight. What it does do, and does really well, is surface the margin already hidden inside your operations. The waste nobody can see. The item gets over-portioned on the line every day. The vendor whose rates crept up 8% over six months and nobody clocked it. The dish that looks like a bestseller but actually bleeds money once you count everything that goes into it.
Here’s the catch though. AI can only pull that off if the data underneath is clean and connected. Feed it fragmented, half-entered numbers and you get fragmented, half-useful answers back. Feed it one connected record of what got bought, what got made, and what got sold, and it turns genuinely powerful. So the order matters, and it isn’t glamorous but it’s the whole game. First you connect the data. Then AI multiplies what it’s worth. Jumping straight to the AI part and skipping the plumbing is the mistake I watch a lot of businesses gear up to make. It’s also why the sector-focused players are placed so well for this moment. They already own the clean, connected dataset for their industry, and a horizontal tool with a hundred half-filled custom fields doesn’t.
What this means for India
India happens to have exactly the conditions that make this shift move faster here than most places. Our businesses watch every rupee, so the long, expensive implementation cycles of old-school enterprise software were always a bad fit. Our sectors are huge and fragmented, full of operators never big enough to justify a custom build but more than big enough to need proper systems. And digitisation has moved fast here, dragged along by cheap data and UPI, so an owner who ran everything on paper a few years ago is suddenly ready for software that speaks his language.
Put that together and you’ve got a big, underserved market for platforms that pick one industry and go deep. Not a jack-of-all-trades that does a little of everything for everybody. Something that does one sector completely. I think this is where a lot of the interesting building in Indian enterprise software happens over the next ten years. The winners won’t have the longest feature list. They’ll be the ones who understood a single industry so well that using their product feels less like configuring a tool and more like the software already knew how the business ran.
The generic era gave us digitisation, and that genuinely was a gift. But the next era is about comprehension. Software that understands the work, connects the data that work throws off, and then uses intelligence to hand you back the margin that was hiding in plain sight the whole time. It’ll get built one industry at a time. And a good chunk of it, I’d bet, gets built right here.
