The credit file was empty. The life behind it was not. In markets across Africa and Asia, Shivani Siroya met shopkeepers, food vendors and craftspeople who could describe every turn of their businesses but could not produce the paperwork a bank required. Their money moved in cash. Their reputations lived in relationships. Their ambitions were obvious at the kitchen table and invisible at the loan desk.
Siroya was there as an analyst with the United Nations Population Fund, studying how small loans affected economic progress. She interviewed more than 3,500 people across nine markets. In practice, this meant arriving in the morning, following people to work, recording what came in and what went out, and returning to their homes. She called herself a “walking QuickBooks.” It is a funny phrase for painstaking work, and a precise description of a founder before she knew she was one.
The official records said almost nothing. Daily life said plenty. A woman might run a food stall, pay suppliers, support relatives and manage uncertain demand without ever generating a conventional score. Siroya began to see the absence of data as an infrastructure failure, not a personal deficiency. When some owners needed money and had nowhere reasonable to obtain it, frustration defeated professional distance. She lent to them herself.
You have to be proximate to the problem to understand the pain point your customers are feeling.Shivani Siroya
The useful kind of obsession
Back in finance, the problem refused to stay politely in the past. Siroya emailed roughly 1,500 people on LinkedIn, looking for anyone attempting a solution. The responses produced no satisfying answer. Mentors told her the opening was real. She did not know the startup world or venture capital, so she approached both as research assignments.
She learned to code on the side. She built a prototype. She used her savings to test early models while keeping her full-time job. A mentor supplied a rule with none of the romance usually attached to entrepreneurship: do not quit until six to twelve months of rent is in the bank. Conviction may wear a leather jacket on conference stages, but at home it keeps a cash buffer.
The Siroya sequence - a founder playbook
That sequence gave her unusual leverage. By the time she spoke with investors, she was not selling a cloud of possibility. She had a prototype and results. She had also spent enough time with customers to explain why the product needed to exist. The first version collected financial information by text message. Later, the mobile phone itself became a source of permissioned signals - a record of consistency where a bureau saw only white space.
Tala, first called InVenture, began by generating scores and trying to persuade banks to lend. The banks moved slowly. Another founder might have defended the elegance of the original model. Siroya saw that a score without capital had not solved the customer’s problem. Tala became the lender.
Trust becomes a product decision
The business that emerged is more complicated than its clean origin story. Tala offers small, short-term loans to customers whose incomes can be modest and irregular. Each loan carries fixed operating costs, however small the principal. Defaults are real. Pricing can be high. A decision engine must judge risk quickly while the person receiving its verdict needs clarity, dignity and a useful option. Financial inclusion is not a halo placed over these trade-offs. It is the hard practice of improving them.
The customer test
Is the offer understandable, timely and more useful than the alternatives available?
The system test
Can the lender price risk, protect trust and remain durable enough to return tomorrow?
In 2020, that durability was tested. Tala’s regular default rate tripled during a sudden economic dislocation. Monthly lending fell from about $80 million to $3 million. Siroya worried about the company’s lenders and its roughly 600 employees. Tala cut costs and reduced staff. After a year, lending returned to its earlier level. The episode left no room for the decorative version of mission. Protecting access also meant protecting the institution providing it.
The sudden stop - monthly lending
The recovery restored lending volume. It did not erase the cost of getting there.
The company later rebuilt its underwriting. Its earlier system placed applicants into broad risk buckets. The newer approach aims to personalize decisions, drawing on how customers use the app and applying causal inference - a way to learn from many small tests when a neat laboratory control group is unavailable. In Mexico, Tala said the change raised approvals from roughly 40 percent to as high as 80 percent while defaults fell among customers with similar credit profiles.
The numbers matter because the central promise is not merely to approve more people. It is to see each person more accurately. A model that replaces one blunt category with another has only digitized the old problem. Better decisions require more relevant signals, careful consent, constant checks for error and a customer relationship strong enough to survive a refusal.
A global company with local ears
Siroya grew up between New York and time with family in Udaipur, Rajasthan. Moving between cultures became practical training for a company that could never succeed by treating one market as a translated version of another. Tala hired local leaders and built teams in the places it served. Siroya has said she still visits customers when she travels in those markets. The habit connects the chief executive to the analyst who once followed cash through an ordinary day.
Her earliest memory of economic difference is set in her grandparents’ Udaipur courtyard. Another family lived and worked across from them, grinding flour. Siroya played and studied with the family’s children. As a child, companionship made the courtyard feel shared. Growing older made the unequal exits from it impossible to miss. The children occupied the same space and inherited different ranges of opportunity. Years later, the financial system would present her with the same optical trick at scale: people living within the economy while remaining outside the institutions meant to serve it.
This helps explain her resistance to language that shrinks customers. Tala uses “Global Majority” for people historically excluded from financial services despite their enormous collective economic power. The phrase is strategic as well as courteous. “Unbanked” describes an institutional absence; “majority” describes a market with agency. One points to what people lack. The other reminds product teams what incumbent finance has failed to build.
Siroya’s leadership vocabulary follows the same pattern. She looks for local expertise, curiosity and people willing to learn across cultures. Earlier in her career, she spoke admiringly of ClassPass founder Payal Kadakia for turning a personal passion into a business. Her own version is less about importing a clever system and more about constructing one around observed behavior. The distinction matters across borders. A global platform needs common engineering, but customer trust is stubbornly local: a repayment date, a clear fee, a support conversation in the right language.
By 2025, Tala reported an annualized revenue run rate of $300 million in the first quarter; later that year, reported annualized revenue reached $340 million. The company also had 1.8 million active, revenue-producing customers. Yet it was still losing money and choosing expansion over immediate profitability. Siroya argued that Tala could pull back and become profitable, but that its mission required continued growth. The statement is candid and contestable, as strategy should be.
Expansion gathered speed. Tala entered Guatemala and prepared launches across Latin America and Asia. In January 2026, it announced a $100 million partnership with CIMB for Vietnam, offering a flexible line of credit through a fully digital application. Tala said nearly 13 million people had used its app and that it had disbursed $7 billion over the preceding decade.
The company’s frontier now includes a $50 million tokenized credit facility, built with Huma Finance on Solana. In May 2026, Tala also announced an embedded partnership with Airtm, putting Tala-underwritten credit inside a stablecoin wallet and beginning in Guatemala. At that point the company reported more than 14 million customers and $8 billion distributed. Siroya has also discussed savings and international transfers. The vocabulary has traveled a long way from handwritten cash flows. The human question remains the same: does this rail give a customer a fairer and more dependable financial life?
I wasn’t trying to sell an idea, and that caught people’s attention.Shivani Siroya
What a founder can steal
A broadly transferable insight from Siroya’s work is a method. First, replace demographic distance with lived detail. Interview until the exceptions become patterns. Second, identify what the incumbent system is unable to observe. Third, make a small artifact that proves the missing thing can be measured or served. Finally, bring the proof to capital while preserving enough independence to walk away from a poor fit.
There is also a quieter lesson about identity. Siroya did not begin as a software engineer, and founding a company was not her declared destination. She became the person the problem required: researcher, coder, fundraiser, lender, manager and translator between a household cash box and global capital. “I’ve had to level up,” she once said. The phrase is refreshingly free of destiny.
Her public manner is energetic, but the operating philosophy favors proof over performance. She advises founders to bring a prototype and results, making it difficult for an investor to look away. That is partly fundraising craft. More deeply, it is a way to keep the person with the money from becoming the main character. Evidence returns the conversation to the customer, the behavior and the unmet need. A polished story can win a room. A working result can discipline it.
Tala’s work will continue to invite scrutiny. Credit touches power, privacy and the price of a bad month. Growth makes every mistake larger. New data can sharpen judgment or disguise an old prejudice in cleaner mathematics. The proper measure of the company is therefore not the sophistication of its language, but the quality of the options it creates and the trust it earns after the first loan.
Siroya’s original act was to notice a mismatch: capable people on one side, an empty field in a database on the other. She stayed long enough to understand that the blank belonged to the system. Then she began filling it, one conversation at a time. It is difficult work, occasionally untidy and still unfinished. Worthwhile infrastructure often is.