Why Arun Jain of Intellect Design Arena sees banks missing the forest for the AI trees

Arun Jain doesn’t ease into a conversation. Within few minutes of settling down for an interaction with Fortune India, he’s calling out one of the most influential bankers in the world.

“Jamie Dimon says he’s investing $17 billion in technology. It’s bullshit,” Jain says, adding “You’re celebrating your inefficiency.”

The provocation is deliberate.

Jain's career in banking technology spans from the earliest days of bootstrapped software venture in India to the building of Polaris, one of the country’s earliest unicorns, and now to his role as founder and chairman of Intellect Design Arena. His irreverence is also on the back having migrated the Reserve Bank of India (RBI), a first for any central bank worldwide, to the cloud without a hiccup.

For the 65-year-old, the problem with technology in banking isn’t that the industry isn’t spending enough, it’s that much of the spend is patchwork: layering new systems on top of legacy pipes, outweighing the cost of serving the customer. And now, he cautions, that the same pattern is playing in the rush towards artificial intelligence.

Building blocks

As an entrepreneur, Jain has made decisions that would have been unthinkable to most business leaders: selling off a profitable services business in 2014 to go all-in on products. The services arm generated ₹200 crore in profits each year. The products side was loss-making. “If I had kept both, my attention would have gone 80% to services, 20% to products,” he says, adding, “You don’t get an entrepreneurship mindset in that comfort zone.”

From the ₹500 crore sale proceeds, he invested half back into Intellect and ploughed a significant portion into Mission Samriddhi, his rural development trust. “My consumption is less than anybody else's. So, I was happy to put it back where it could do better,” he says matter-of-factly.

In the following decade, Jain invested ₹2,000 crore in building a banking technology stack from scratch, codifying microservices so that every product function is modular, interoperable, and faster to deploy.

By 2020, Intellect had a “foundry-grade” library of microservices, each documented and numbered in what Jain calls his “Green Book.” It was, in his mind, the foundation for the next decade.

Intellect’s AI journey had, in fact, began in 2015, with labs in New York and London. “India had no data scientists at that time,” Jain says. The first years were frustrating. “The technology was fine. The problem was enterprise acceptance of getting different departments to use it.”

In 2022, the arrival of large language models (LLMs), what Jain describes as the capability of turning human language into mathematical vectors, was a turning point. “Language is vectorised now. Put into number form, it can be computed, just like numbers have been since 1945. That’s why there’s euphoria,” tells Jain.

Jain’s approach maps 13 core banking services, including lending, deposits, payments, cards, and wealth management, into a numbered microservice. “Finite number of APIs, finite number of entry points,” he says. The benefits are two-fold: lower risk and faster deployment. “This reduces the cost of implementation by 30% on our side only because we can implement the product in 12 months, which people are implementing in 24,” claims Jain.

His answer is neither a chatbot nor a suite of autonomous agents, the current buzzword in enterprise AI, but something more like a knowledge-first, human-in-the-loop AI architecture called Purple Fabric.

Leaning over a table, Jain shows a blueprint comprising a neat lattice of boxes, each with its own code. It looks less like a banking diagram than the blueprint for a micro-chip. “This,” he says, tapping the grid, “is all of banking. Forty-two core microservices. That’s it.”

Banking in pieces

In most banks, the technology stack is an accumulation of decades where the core platform of one era is linked to modules from another, each with its own database, and interface. APIs, that act as connectors between them, can number in thousands, posing a potential vulnerability.

In Jain’s architecture, every transaction, every account, every product a bank offers can be broken down, named and catalogued. Once done, the tangle of legacy systems at banks becomes manageable, and ready to roll out. But reinvention, he argues, is what most banks are not doing. “Europe’s already there where the cost of technology is bigger than cost of customer service. India will go the same way,” fears Jain.

When CIOs argue that such re-architecture is too disruptive, Jain points to the Indian central bank. “The RBI is the first central bank in the world to move to cloud,” he says. “We did it in 18 months, without disruption. A hundred million transactions a day, running on Rs 16 crore worth of servers.” If the central bank, with its systemic risk and compliance burden, can modernise seamlessly, he argues, so can banks.

Jain though tempers the excitement around AI, which, he says, is currently of “ninth-grade level”, that is, capable, but far from the PhD-level reasoning enterprises require. “It needs human support. You need human intelligence and AI together,” says Jain.

If there is a single point where Jain diverges sharply from the mainstream AI conversation, it’s about “agentic AI,” autonomous digital agents that can take action on a user’s behalf. The idea is popular: Microsoft sells copilots, Salesforce pitches agents, and many startups are built entirely around them. But Jain is unconvinced. “Agents are workers. They can do small tasks. But what you need is a digital expert, a set of agents that can dialogue with each other, reason, and know when to bring in a human,” says Jain.

Jain’s critique that autonomous digital workers will crumble under enterprise compliance, security, and governance demands. “They create an agent, don’t test its maturity, and don’t measure its hallucination rate or ethical reasoning,” feels Jain.

Purple Streak

The prerequisite that Intellect is pitching is an Enterprise Knowledge Garden — a structured repository of explicit knowledge (documents, processes), implicit knowledge (workflows), and regulatory knowledge. From there, Purple Fabric, Intellect’s AI platform, layers governance, domain-specific digital experts, and the ability to switch between multiple LLMs depending on the problem.

Purple Fabric’s four-layer architecture begins with the knowledge garden, then builds digital experts, wraps them in governance (security, traceability, ethics), and connects to a library of interchangeable LLMs.

Jain says, “We’ve built Purple Fabric at Indian cost.”

That “Indian cost” is $8,000 per core [a single processing unit] per year. Not per user, not per month. Competitor copilots often cost $40 per user per month, which can balloon into millions for a large bank. “I’m inspired by Jio that is make adoption inevitable, and monetise it later,” explains Jain.

The strategy also mirrors open-source models like Red Hat’s: get wide adoption first, then layer on premium services. “If 5,000 customers use my platform instead of 50, the value is on a different scale,” he says.

David of AI?

One of the slides in Intellect’s investor presentation shows Intellect alongside Palantir in one quadrant, a rather bold comparison that might raise eyebrows. “Indian banks are buying AWS or Microsoft like it’s the only game in town. They understand B2C, they don't understand large corporation issues. Palantir, which is a $300 billion US company, is the only company which understands big corporations because they work with US defence forces. They know what the security of the knowledge is, which level of security. That's what we applied over here [in Purple Fabric],” says Jain.

Intellect’s numbers suggest ambition but with some grounding: 85% of Intellect’s revenue from overseas clients like HSBC, Barclays, JP Morgan, Citibank, CIBC; $10 million single-licence deals with banks in Singapore and Canada. India, at 15% of revenue, is a smaller piece, but Jain believes AI could change that: 250 active leads, 3,000 registrations for his Purple Fabric masterclass, and seven bank leadership teams booked for bootcamps.

Core banking products are growing at 20% and clocking 30% profit annually, with incremental margins of 55 percent. Purple Fabric’s current revenue is ₹50–100 crore; Jain’s FY29 target is ₹1,000 crore, with a long-term goal of $1 billion from a single product. R&D spend is heavy — ₹1,000 crore into AI so far, one-third of total R&D — but, he says, justified: “We have 1,200 research engineers. Two million hours a year. At US rates, that’s a $2 billion investment. Cutting R&D would be foolhardy.”

Whether or not one agrees, Jain’s case has a coherent internal logic. His disdain for CIO caution comes from executing at the highest stakes — from RBI’s cloud migration to running the government eMarketplace (GeM) without a day’s disruption. “If the central bank with its systemic risk and regulatory complexity can migrate without incident, commercial banks have no excuse. I’m not saying SBI should rewrite its core overnight, but SBI Wealth has done it. Jio Financial has done it. It’s possible. The fear is manufactured.”

Jain doesn’t hide his frustration with domestic banking CIOs. “They’re mental slaves to the thinking process of America. They wait for Microsoft or GE to certify them before they believe in a solution,” he says.

Though Jain sees few leaders who combine deep technological and business understanding. One is Sanjiv Bajaj of Bajaj Finserv. “He’s an aberration,” Jain says. “He understands technology and business holistically. He designs AI for himself. I can trust what he says.”

Bajaj’s success with AI, Jain suggests, comes from design discipline. Most others, he says, “are buying copilots and thinking they’ve solved AI.”

For a revenue size of Rs 2,700 crore with a market value of over $1.5 billion, competition may see Intellect as punching above its weight. But the credibility of having migrated the RBI to the cloud with a single disruption till date is no mean feat.

Jain knows the odds when he says, “In 10 years, the cost of technology will be bigger than the cost of customer service. You [banks] never organised your backend properly…and then keep fixing the forest. Unless you simplify now, you’re walking into the same trap Europe fell into. But reimagination needs a cultural shift, and that will take time.”