A brown envelope arrived in early April. Inside, the FSSAI manufacturer and retailer license for Scandinavian Food and Beverages Pvt Ltd, the company my co-founder Niklas and I incorporated last August to run Fika B’lore from a solar-powered farmhouse north of Bangalore. License number 11226302000578. The certificate, printed in the same officious typeface used by every regulator I’ve dealt with in this country, carried a small confidence at the bottom: Valid Until: 01 April 2027. It isn’t.
Three weeks earlier, on March 13, the Ministry of Health had approved perpetual validity for FSSAI licenses, ending periodic renewals for food businesses across India. The certificate in my hands, dated April 2, was one of the first issued under the new regime. The “Valid Until” line was a printing artifact. The form had not caught up to the rule. I filed the certificate in a folder, made a note in the company log, and went back to that morning’s WhatsApp orders.
I have started to think this is the actual texture of building a small business in India. Not the regulation itself. The gap between the regulation on paper and the regulation in your hand.
Malmö assumes the first ledger is the real one. Bangalore knows the second is.
Stewart Brand has a frame that has helped me more than any India-specific business book. He calls it pace layering. Civilisations, he argues, are made of six layers, each moving at a different speed. Fashion is the fastest. Then Commerce. Then Infrastructure. Then Governance. Then Culture. Nature is the slowest. Fast learns, slow remembers, he writes. Fast proposes, slow disposes. The tension between the layers is what makes a society work. If they all moved at the same speed, you’d have either a coup or a museum, not a country.
In Brand’s frame, Governance sits in the middle. It’s supposed to move slowly enough to be predictable, but fast enough to receive upward pressure from Commerce when the operating reality has changed. Sweden has spent the last fifty years pretending Governance is a slow layer. India is busy proving it’s a middle layer.
In our first eight months as a company, the regime around us has changed at least three times. Three weeks after we incorporated, the GST Council collapsed India’s tax slabs from four into two and moved bakery products from 18% to 5%. The cost stack of every order we’d ever priced changed in a single Council session. In November, India notified the rules under the DPDP Act, the country’s GDPR cousin, with a compliance runway that ends in May 2027. In March, the Ministry of Health made FSSAI licenses perpetual, ending the renewal cycle that had been a fixture of food regulation for two decades. We are a three-person operation, and the regime under us has been rewritten in three different domains in less time than it took us to set up our HDFC current account.
A Swedish operator reading that paragraph will read it as instability. I read it as a system doing what Brand says systems are supposed to do. Commerce moved fast in this country for a decade. Governance is now catching up. The regime is volatile because the regime is responsive. The fact that you can call your CA on a Tuesday, learn your tax category has changed, and be operating under the new rate by the next billing cycle is not a bug. In Brand’s frame, it is what a healthy middle layer looks like.
This is the part Malmö tends to miss.
I should be careful not to romanticise it. The pace cuts both ways, and the same year we benefited from GST 2.0 we have also been working alongside a registry that has gone in the opposite direction. India’s trademark examination queue lengthened from roughly thirty days in early 2024 to about 550 days now. We filed our marks in February 2026 through a Chennai advocate; we expect to hear back in 2027 at the earliest. The corporate filing system, MCA21, was modernised in the same window: routine forms that took days now clear in hours. The intellectual property registry didn’t move at all. Reform here is sectoral and asynchronous, not a single tide. A Swedish operator assuming the Malmö frame of “system-wide good or system-wide stagnant” will read MCA21 V3 and assume the trademark queue must be similar. It isn’t.
The harder version of this argument was made by Andy Mukherjee in Bloomberg this January, in a column called The Great Unease of Doing Business in India. His thesis: India has improved on every reform metric for a decade, and yet operators on the ground report no real change in friction. Permits still take longer than they should. Notices still arrive without warning. The reform looks tidy in a press release and stays untidy in a kitchen.
There is also the older shadow. The Cairn Energy retrospective tax case, in which India ran a 2012 amendment backwards to demand 1.6 billion dollars on a 2006 group restructure. The Vodafone case and the 2018 e-commerce FDI policy belong to the same family. The Indian state has the capacity, occasionally, to change its mind in a way that costs an operator hundreds of millions of dollars and several years of arbitration. None of this contradicts the pace-layer argument. It sharpens it. India’s Governance layer is fast because it is still under upward pressure from Commerce, and an immature middle layer sometimes overshoots.
The other complication is that the official regime is only one layer of operating reality. We hired a delivery driver in April. His name appears on no contract because, it turned out, he doesn’t have one with us. He is salaried by Ramesh, who runs a fleet of fourteen cars, and Ramesh pays the driver out of the daily rate we pay him. We negotiated the arrangement with Ramesh in person. The work will happen, was the phrase he used, and so far the work happens. The papers we have describe a relationship we do not actually have. There is a whole layer of Indian commerce that operates underneath the regime any reformist agency can see, and the official regime, however fast it moves, never quite reaches it. James C. Scott calls this metis: practical knowledge embedded in local networks that resists state-level legibility. You can modernise the MCA filing system every year. The fleet of fourteen will adapt around it.
I have come to think a Swedish operator looking at India needs two clocks running, not one.
The first clock is Malmö time. In Sweden, you build a business for the rules that are already written. The aktiebolag registration takes a week. Skatteverket’s processes are slow but predictable. The EU keeps producing directives, and they keep being slightly heavier than the last set, and you incorporate them the way a forest incorporates each new fall of leaves. You design for stability. The penalty for inattention is small, because nothing important changes on a quarterly basis.
The second clock is the one I have started to learn here. In India, you build a business for the regime that is coming next. You read your CA’s monthly newsletter the way Swedes read the weather forecast. You assume your cost stack will move once a year and your compliance stack will move twice. You accept that any official regime will only describe two-thirds of your operating reality, and the rest will live in a relationship you negotiate weekly. In this country, the penalty for inattention isn’t small. It is the difference between a business that runs and a business that doesn’t.
A Swedish company thinking about expanding into India almost always brings only the first clock. The risk register treats India as a noisier version of Malmö, and the noise gets translated into either over-caution (we’ll only do this through a JV, we’ll only do this with a 24-month legal review) or under-caution (we’ll set up like we would in Copenhagen and figure the rest out). Both are wrong. Both miss the question that running a small business here forces you to answer every quarter: which regime are you actually operating under, the one on the certificate or the one in the WhatsApp thread?
The answer is usually both, and neither, and you keep two ledgers. One is the statutory ledger that the CA, the auditor, and the Registrar of Companies see. The other is the operating ledger that the kitchen, the driver, the suppliers, and the customers know. Malmö assumes the first ledger is the real one. Bangalore knows the second is. The work of running a business here is keeping the two within walking distance of each other.
I keep the FSSAI certificate in a folder labelled Statutory Filings. The Valid Until: 01 April 2027 line is still printed at the bottom. I haven’t asked the regulator to issue a corrected version. I assume the form will catch up to the rule eventually, and in the meantime the certificate functions perfectly well as a piece of paper, which is most of what it needed to do.
The harder question I am sitting with isn’t whether India’s regime is more or less hospitable to a small Swedish-owned bakery than Sweden’s. It is whether I can encode enough of what we have actually built here, the part that lives in handshakes and a farmhouse lease and a driver who works for someone else, into a form that survives us moving back to Sweden. The paper company is easy to acquire. The other company, the one that runs on metis, isn’t on any certificate.