Srinivas Njay would like you to stop repeating yourself

Srinivas Njay found an early clue to the future of banking while watching his mother learn to text on WhatsApp. Here was a familiar problem for anyone who has ever become the family’s unpaid technology department: the software could do plenty, provided somebody first explained the software. Njay saw an opening for utility bots. A conversation might make a better starting point than another application to download.

In 2016, he and Bruce Kim were working on Recharge Bot, a service that let people top up prepaid phones through Facebook Messenger. The task was modest. The ambition behind it was considerably less so: make useful transactions available inside an interaction people already understood. A decade later, as co-founder and CEO of interface.ai, Njay is still working on that problem. The institutions have changed. The impatience with unnecessary steps has survived.

A family introduction to friction

The mother-and-messaging episode belongs beside another family connection. Njay’s father established a financial institution in India. Njay helped the business use technology as it grew, giving him a view of banking from inside an institution rather than solely through a developer’s screen. In his account of those years, the opportunity lay in the daily difficulties of serving customers and bringing fragmented information together.

That experience helps explain his eventual choice of customers. Community banks and credit unions have substantial responsibilities, even when their technology budgets are modest. They must answer questions, protect accounts and make products understandable. A small institution does not get to tell a customer that the complicated bits are reserved for larger banks. Njay’s business would aim to supply capabilities those institutions could struggle to build themselves.

His technical education ran through a computer science bachelor’s degree at Siddaganga Institute of Technology and a master’s at Northeastern University. He has described researching AI during his master’s. At Microsoft, he worked on Bing advertising products; at Electronic Arts, on mobile games. These were different settings for a product manager, but each put software in front of people who expected it to work without first acquiring a professional interest in its machinery.

There is a pleasingly practical progression here: study intelligence, work on products, then discover how often intelligence is needed simply to get through a product. Banking gave that frustration a particular weight. A difficult interface can stand between a person and something they need to do with their money.

Two founders, several first attempts

Njay and Kim began collaborating in 2015. Kim brought experience running Inovaware, a customer-management and subscription-billing company. Their partnership joined Njay’s product and banking experience with Kim’s background in the systems that keep transactions moving. Before interface.ai’s U.S. banking business took shape, they were testing ideas under the Payjo name.

The early work included cross-border transfers and then the Messenger-based recharge service. In April 2016, the founders were experimenting with a way for people abroad to pay for relatives’ phone credit in India. The attraction was straightforward: the customer could accomplish something useful through an existing messaging app. Every additional download is a small request for patience. Njay was trying to stop making that request.

These were working experiments, with changes of direction attached. By April 2018, he was discussing voice-assisted banking for credit unions in a NACUSO interview, still identified as Payjo’s founder and CEO. The later interface.ai business launched in 2019. The chronology matters because it gives the company’s current vocabulary some history. The work began well before the recent rush to put an AI assistant in every available rectangle.

The continuity is more interesting than a tidy founding legend. He moved between markets and use cases while keeping a fairly persistent question: could people ask for what they needed, and have software help them carry it out? Persistence sometimes looks less like staying on one road than repeatedly correcting the destination sign.

An investor courtship, on his own timetable

By March 2024, interface.ai had signed more than 100 credit unions and community banks. Its early financing combined $1 million of the founders’ savings, a $3 million credit line from the institution run by Njay’s father, and customer revenue. That last item is a useful character in any startup story. It tends to arrive with demands, but it also pays bills.

Njay was taking investor calls without treating them as a rescue operation. “I’m just dating,” he said. The line is amusing because it reduces a solemn Silicon Valley ritual to something with coffee, questions and the option of going home. His stated interest was in a partner who could help scale the company, rather than capital for its own sake.

That courtship eventually produced a deal. In October 2024, interface.ai announced $30 million in outside financing led by Avataar Venture Partners. The package consisted of $20 million in equity and $10 million in debt. It was a change in how the business would finance its next stage, following years of building with its own resources.

For Njay, the move was consistent with an earlier willingness to consider investors who brought relevant operating experience. Bootstrapping had been a way to build the company; accepting capital became another decision about its development. The interesting question was what the money and the relationship would let the team do next.

A large job for smaller banks

Njay’s commercial focus has stayed unusually specific. interface.ai develops banking software for credit unions and community banks, combining voice, digital and employee-facing assistance. His pitch starts with an imbalance: large national banks can devote substantial resources to AI, while smaller institutions may find the staffing and infrastructure difficult to assemble.

“Interface democratizes access to AI,” he said in 2024. In practical terms, that means banking-specific integrations, workflows and models offered as a product. An institution buying the system is buying work that it would otherwise have to organize itself. The software has to understand enough about banking to be useful beyond a pleasant greeting.

His ambition also extends behind the customer conversation. Employees need access to policies and procedures; institutions need consistent information across channels. In 2023, he was already describing a shared intelligence layer connecting customer and employee interactions. Today, the BankGPT platform carries that approach across the company’s products. The promise depends on those connections doing useful work together.

That is a rather less photogenic assignment than producing a clever answer on a screen. It involves the institution’s knowledge, its existing systems and its rules about who may do what. Njay has chosen a business in which the unglamorous details are part of the product. A banking assistant’s charm has limited value if the customer must then start again somewhere else.

His public discussions keep returning to the people responsible for running those institutions. He was listed for a March 2026 panel with Krystle Brown of Navigator Credit Union and Josh Henderson of Red Rocks Credit Union, examining efficiency ratios and operational impact. The subject suits his market: buyers need to connect a software decision to the cost of serving members. A persuasive demonstration begins the conversation; the institution’s daily operation supplies the longer examination.

The awkward question for a beautiful dashboard

In February 2026, Njay published an argument about how banking AI should be measured. He challenged the habit of treating containment - an interaction ending without transfer to a human - as proof that work had been automated. An ended conversation can still leave a customer’s request unresolved and an employee with a task to finish.

“Containment is a session outcome. Automation is a task outcome.”

Srinivas Njay, February 2026

His distinction supplies a useful test of his own ambitions. The right question is whether the requested work was completed, including what happens after the conversation closes. A dashboard can count the absence of a transfer; a customer cares about the presence of a result. The two deserve separate columns.

It is an exacting standard for a founder selling automation. It asks buyers to look past a convenient headline percentage and examine what their staff still have to do. Njay’s argument makes resolution, follow-up work and operational capacity central to evaluating a system. The consequence is a more demanding conversation about what the technology earns its keep doing.

Human judgment, without the grand tour

In April 2026, interface.ai introduced Nexus, a contact-center platform built on BankGPT. Its proposed arrangement keeps AI in charge of the customer conversation while routing individual requests for human judgment to staff. An approval or policy exception could be handled without passing the entire call to another person.

The company presents this as a way to avoid repeated explanations and hold queues. Those are product claims and intended outcomes, rather than a guarantee for every institution. The design itself reveals where Njay’s attention has moved: from making a transaction possible in a chat window to reorganizing the work surrounding a conversation.

There is still a place for human expertise in that picture. Nexus’s co-pilot mode supports employees who lead the interaction, while its agentic mode lets software lead and seek specific help. Njay’s next chapter depends on making those arrangements useful inside institutions with established responsibilities, systems and customers.

The distance from prepaid phone credit to banking contact centers is considerable. The original irritation remains recognizable. Someone wants to do an ordinary thing. The interface asks for extra effort. Njay has spent years trying to remove that effort, first for a phone top-up and now across a bank’s service operation. If the work succeeds, the customer need never develop an opinion about the technology. Getting on with the day would be quite sufficient.