The bank card took six weeks to arrive. It belonged, in theory, to Alok Kumar's 15-year-old brother in Saudi Arabia. In practice, their father's name was printed on it, and every purchase set off an SMS to Dad. The arrangement had all the privacy of a family group chat with push notifications. One transaction eventually revealed that the teenager had a girlfriend. The card's brief spell of glamour was over.
Kumar could see a decade of technological progress everywhere except in this small domestic ritual. When he had been his brother's age, he too had asked his parents for cards and cash. The world had moved from the iPhone 4 to the iPhone 13; teenage money still traveled through a parental checkpoint. He called Nuha Hashem, a friend he had known since they were teenagers. She was pursuing a doctorate in the United States. Together, they began studying the gap.
The result was Zywa, founded in 2021 as a social banking app and prepaid card for young people in the Middle East and North Africa. The product let them receive, spend and understand money in an environment designed for their age rather than borrowed from their parents. Hashem built while still in graduate school, then left her program and moved to Dubai. Kumar brought an engineer's regard for systems, a consultant's instinct for institutions and a problem he could explain without a slide.
The homework before the hustle
Kumar and Hashem did not begin by romanticizing speed. Before the first line of code, they wanted clear answers: why this company, why this market, why the existing system failed and where their product could go. They interviewed young people, banks and other participants in the financial system. Then they bootstrapped and assembled a list of 300 investors whose experience, networks and values might actually help.
The first cheque required four months and 189 customized emails. The pair researched each recipient, described a specific case for working together and asked other founders about the investors on their list. Every meeting produced notes. Every set of notes produced a new pitch deck. Kumar was treating fundraising like product development: ship, listen, revise. Rejection was information with poor manners.
“We were telling our own story when pitched to investors.”Alok Kumar, on founder-led fundraising
The method began to compound. A second cheque followed three months after the first. Zywa entered Y Combinator's Winter 2022 batch and raised a $3 million seed round that year. By 2023, it had recorded more than 130,000 signups and over 40,000 accounts. Kumar was named to Forbes Middle East's 30 Under 30 list in Commerce & Finance. The original family irritation had become a funded regional fintech with recognizable partners.
It also carried a mission larger than convenience. Kumar had grown up in Saudi Arabia and worked on strategic projects there. He saw how public efforts toward financial inclusion could stall before reaching young customers because the segment looked unattractive to traditional banks. Zywa's answer was to make learning part of the product: money management presented through the language and mechanics of the generation using it.
When the side project answered back
Success made the next decision harder, not easier. Zywa had spent roughly two years inside regulated finance. Licenses jammed, rules changed and the market beneath the plan weakened. Meanwhile, growth created a simpler nuisance: too many support conversations for a small team. The founders tried available voice-AI products. They found them robotic, generic and forgetful. So the company trained its own models and built an AI employee to handle support and inbound calls.
The internal fix behaved like a flare. During a demonstration, a partner interrupted with the sentence founders spend years hoping to hear in some form: forget the bank, I want this. The software created to keep Zywa afloat had found demand beyond Zywa. The company could defend the story it had told investors, employees and customers, or it could believe the evidence occurring in the room.
They killed the bank and kept the AI. The line sounds brisk. The experience was not. Kumar has spoken publicly about the pain of letting 86 people go and the emotional cost of making brutal decisions in view of everyone affected. A pivot is a jaunty little noun for a company-scale admission. It asks a founder to detach identity from direction while other people's livelihoods sit in the room.
Yet the team did not discard every human asset along with the old plan. Kumar and Hashem bet on the learning capacity of the people who remained. An iOS engineer became a backend engineer. An Android engineer moved into AI. A backend engineer aimed for machine-learning architecture. The founders gave themselves a crash course in voice activity detection, deployment pipelines, B2B sales and enterprise decision cycles.
“There's nothing more powerful than gritty generalists, people who are willing to unlearn and relearn everything they know, again and again.”Alok Kumar
This is the less glamorous part of adaptation. The website can change overnight; competence cannot. Kumar's bet was that a team already practiced in trust, speed and shared adversity could cross the knowledge gap together. Products are repositories of code. Teams are repositories of memory. He chose not to throw out both.
A founder assembled from several industries
Kumar's path had quietly prepared him for a company that lives between technology and consequential operations. He studied engineering at King Fahd University of Petroleum & Minerals, graduating in 2017. He worked in wireline field operations at Schlumberger, where reliability is not a decorative value. At Strategic Gears, he worked on strategy and public-sector transformation in Saudi Arabia. Consulting taught him how large institutions decide; fintech taught him what regulation does to a timetable.
His university years also contained an early rehearsal for leadership through language. Kumar became a founding president of the ALOHA! KFUPM Toastmasters Club and reported growing paid membership by more than 120 percent. He earned advanced communication and leadership credentials. A later profile offered a sweeter detail: he wished he could speak every language, simply to connect with people more naturally.
That wish now has a technical echo. Cozmo's product has emphasized multilingual voice agents, including Arabic and Gulf dialect understanding, for regulated markets. The company first described a broad enterprise workforce able to see, speak, think and act across calls, messages, images and documents. Its current Y Combinator profile narrows the wedge to property claims: taking a loss report, entering claim data, dispatching contractors, chasing service-level deadlines and drafting estimates from field photos.
Write down why the market fails before writing the first line of code.
Use every investor or customer conversation to improve the next one.
Pay attention when a workaround attracts more pull than the flagship.
A trusted team can change disciplines when the company changes industries.
In April 2026, Kumar previewed another layer called Pods: one runtime for voice, vision and text agents, with shared memory across channels. He said ten million conversations were already running on it in production. The practical insight is easy to miss beneath the AI vocabulary. A customer's problem rarely respects software categories. A call becomes a message; a message needs a photo; a photo creates a decision; a decision requires an audit trail. The job is the continuity.
Kumar's public ambition is to turn Cozmo into a system of action for enterprises rather than another database that merely records what happened. For claims, that means software that can observe damage, speak to the people involved, reason about the next step and perform the work while keeping a traceable account. It is a far journey from a teenager's prepaid card, though both products begin with the same impatience: why should a person wait while institutions pass information from one silo to another?
Mentors become customers
In November 2025, Kumar announced a pleasing bit of narrative symmetry. Strategic Gears, the firm he had left four years earlier to start Zywa, signed up with Cozmo AI. Nasser, his first mentor there, was now a partner. Kumar credited him with teaching the habits of building quickly and setting a high bar. Careers often preserve their old relationships even when their business cards do not.
The episode reveals something stable beneath all the changes. Kumar talks about companies through people: the brother whose frustration exposed a market, the childhood friend who became his co-founder, the investors who received individual notes, the engineers trusted to reinvent their work, the mentor who later became a customer. Even the technology is described as useful when it releases people from repetitive labor and lets them handle higher-value conversations.
There is no tidy moral available to every founder. Some side projects deserve to remain side projects. Some pivots are evasions wearing strategy's expensive coat. Kumar's record offers a more demanding test: do the homework before committing, collect evidence while building, and remain willing to change the conclusion. Conviction is valuable. So is noticing when reality has edited the brief.
His first company asked young people to start early. His second asks enterprises to move sooner. Both ideas belong to a founder who seems uncomfortable with the bureaucratic pause between knowing and doing. The useful thing to steal is not the industry or the fashionable noun. It is the loop: observe closely, tell the story yourself, listen for the interruption, then keep the part that works.