Dunzo founder Kabeer Biswas' new startup M wants to automate your home, starting with the kitchen

Dunzo founder Kabeer Biswas' new startup M wants to automate your home, starting with the kitchen

The cook is waiting in the kitchen. The refrigerator door is open. Someone is scrolling through a grocery app, trying to remember whether there are the requisite vegetables, species and cooking oil for lunch. A work call starts in five minutes.

It's a routine that plays out in millions of Indian homes every single day. For Kabeer Biswas, it also represents the next frontier in convenience.

The Dunzo co-founder believes the physical effort of running a household has largely been solved. Groceries, medicines, meals and almost everything else can now be delivered to the doorstep in minutes. The harder problem, he argues, is the invisible coordination that still happens inside the home — from deciding what's for dinner and replenishing groceries to managing household staff and keeping everyday chores moving.

That is the premise behind M, Biswas' new AI startup. The company starts by managing the kitchen, but its ambitions stretch far beyond. Over time, Biswas wants it to become an AI-powered operating layer for the home.

"The question now is: now that everything comes in 10 minutes, what's next in urban convenience?" Biswas told Moneycontrol. "We think it's the mental fatigue involved in managing a house. If Claude can run your work, can M run your house? That's potentially what we're chasing."

The startup recently emerged from stealth after raising Rs 102 crore in a seed round led by Peak XV Partners, with participation from Blume Ventures and fintech major CRED. It currently serves around 150 households in Bengaluru, employs about 45 people and aims to expand to 10,000 homes over the coming months.

According to Biswas, monthly retention is above 90 percent, daily active usage is around 82 percent and only four customers have churned so far.

Today, M lives inside WhatsApp.

Every new customer begins with an onboarding exercise that resembles less of a software installation and more of a household audit.

The team visits the home and maps how it functions: Who cooks? Who cleans? Who washes dishes? What time does breakfast need to be ready? Does someone follow a diabetic diet? Is another family member vegetarian? Which grocery platforms does the household use? Which brands do they buy? Which language does the cook speak?

That information becomes the household's operating manual.

Every evening, M generates the next day's meal plan, taking into account dietary preferences, existing inventory and perishables already sitting inside the refrigerator. The homeowner doesn't need to decide what will be cooked every morning. The cook receives instructions directly through WhatsApp, and the system can converse in multiple Indian languages, including Hindi, Bengali and Nepali, allowing domestic staff to interact with it naturally rather than through the homeowner.

If the cook sends a voice note asking for coriander, refined oil, bread and coffee, M understands what is being requested, identifies the family's preferred brands and remembers where they usually shop. Today, much of that workflow still requires human intervention. The long-term vision is for the AI to complete the entire process automatically.

"After you're onboarded on M, you shouldn't have to manage your kitchen at all. You shouldn't have to answer the question, 'What will be made today?' Everything after that — checking inventory, ordering ingredients and coordinating with the cook — gets taken care of," Biswas said.

The cook, interestingly, is the primary user of the product, not the homeowner.

Instead of replacing household staff, M attempts to become the layer sitting between them and the family, handling instructions, recipes, grocery planning and routine coordination.

The kitchen is only the starting point.

Biswas ultimately wants M to orchestrate every repetitive workflow inside the home by integrating with the broader ecosystem of consumer internet services.

The company is working towards eventually integrating APIs from apps across quick commerce, food delivery, e-commerce and payments, allowing the AI to move from recommending actions to executing them.

Grocery orders could be placed through Blinkit, Zepto or Swiggy Instamart. Household services could eventually be routed through platforms such as Snabbit. Purchases would ultimately be settled through an integrated wallet and line of credit, an area where its investor CRED's involvement could become strategically useful.

The WhatsApp interface, meanwhile, is intended to be temporary. As more services are integrated, M is expected to evolve into a standalone application capable of managing multiple household workflows from a single interface.

For Biswas, the objective isn't to build another chatbot.

"Consumers care about outcomes. They don't care about tools," he said. "You can give everybody a fancy tool and say, 'See what you can do with it.' Consumers at scale don't want tools. They want outcomes."

The ambition is sizeable. So are the challenges.

Unlike enterprise software, every household operates differently. Meal preferences change. Domestic staff vary from home to home. Families make decisions collectively rather than individually. What works for one customer may not work for another.

The business is also unusually high-touch. Every new household currently requires manual onboarding before the AI begins understanding routines. Scaling that process to thousands, or eventually millions, of homes will require far more automation than exists today.

Then there is the technology itself.

While large language models have become significantly more capable over the past two years, AI agents still struggle with reliability, memory and real-world execution. Payments, identity and permissions remain unsolved problems across the industry.

Biswas acknowledges those gaps but believes the technology curve is moving in M's favour.

"You don't start companies when all the technology already exists. You start companies when you know what the curve of technology is going to look like," he said.

He believes AI agents will eventually be able to hold identities, transact independently and interact seamlessly with digital services. M's job, in the meantime, is to build consumer behaviour before that infrastructure fully arrives.

If M succeeds, competition is unlikely to come from another Indian startup.

Instead, it could come from foundation model companies such as OpenAI, Google or Anthropic, all of which are steadily pushing AI assistants beyond simple conversations towards task execution.

Biswas believes the company's advantage will lie not in the underlying models but in the behavioural data it accumulates over time.

"We'll probably have to understand how five million households operate," he said. "Once you've trained a model on your household, the switching cost becomes very high. The only real competition eventually comes from the foundational models."

Whether that proves to be a durable moat remains an open question. M today serves a relatively narrow segment of affluent urban households with domestic staff. Expanding beyond that will require the product to work across different family structures, cities and income groups while convincing consumers to hand over decisions they have traditionally made themselves.

Quick commerce changed how Indians buy groceries by removing the physical effort of shopping. Biswas is betting the next chapter of convenience will belong to whoever removes the mental effort that comes before the order is ever placed.

Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!

Find the best of Al News in one place, specially curated for you every weekend.

Stay on top of the latest tech trends and biggest startup news.