Thought Behind Things
Shut down overnight, Safepay came back as a licensed rival to Pakistan's banks
Safepay's Ziyad Parekh on losing his bank partner overnight, a licence that took three and a half years, Visa's fees, and why Raast could reset payments.
Contents
- A startup that survived the downturn
- A sheltered Karachi childhood, and the night the bubble burst
- New York, the Bay Area, and two Pakistani cities
- A Venmo for Pakistan that nobody paid for
- One merchant ID, an Excel file, and Y Combinator
- Shut down overnight, then three and a half years for a licence
- What Visa and Mastercard really charge
- Safepay today
- Payouts, Raast and the end of the card habit
- Crypto, tokenisation and Pakistan in 2050
A startup that survived the downturn
Muzamil opens with the context of the series. Thought Behind Things is working with Endeavor Pakistan to find the 50 most exciting startups and businesses in the country. He is curious about one group in particular. After the downturn that followed 2022, many startups with broken unit economics either shut down, left, or had to pivot. The ones still standing are the ones he wants to understand.
Safepay is one of them. Muzamil first heard of it two or three years ago, when Stripe invested and the company began to position itself as the Stripe of Pakistan. He has used Stripe in Dubai and says it changed how he does business, especially for people who sell services. His guest is Ziyad Parekh, Safepay’s co-founder and CEO.
A sheltered Karachi childhood, and the night the bubble burst
Ziyad was born and raised in Karachi. He describes a traditional family, a mother who pushed hard on studies, and a close circle of cousins who felt like siblings. He was a good student and played sports, trying to be an all-rounder.
Muzamil asks what Karachi in the 1990s and 2000s felt like, with terrorism, gang violence and weak law and order. Ziyad says his parents sheltered the family as much as they could. The tension was something he felt on the news before he understood it. Then the city came closer. At school, after a security scare, barbed wire went up on the walls. In his O level years came the attack on Benazir Bhutto’s convoy on Shahrah-e-Faisal. His mother was stuck at his grandmother’s house that night and could not get home.
That night became the subject of his college essay. He wrote about how the event broke the bubble he had grown up in and showed him how quickly violence could disrupt anyone’s life.
Muzamil presses on the bigger question. Ziyad had a strong degree, a job in the US and a clear path to an upper-middle-income American life. Why come back? Ziyad says something inside him was never satisfied. He had studied math and economics, added computer science, and wanted to build at the intersection of finance and technology. His father and his father’s cousins were businessmen, so starting something felt natural. And there was a sense of debt.
“I owed a lot to Pakistan and I owed a lot to Karachi and my community and the people over here,” he says. He had spent his first 18 or 19 years there, and those years made him. He chose to start from scratch instead of joining the family business, and his parents stopped supporting him financially after he graduated.
New York, the Bay Area, and two Pakistani cities
Ziyad spent four years at Cornell in upstate New York, which he describes as moving from a city of 18 million to a village. Then he moved to New York City. He loved the skyline, the pace and the feeling that he could walk anywhere and figure things out. He stayed about two years before moving to the Bay Area.
Muzamil quotes Baz Luhrmann’s advice to live in New York once but leave before it makes you hard, and Northern California once but leave before it makes you soft. He compares the pair to Islamabad and Karachi. In his view, Islamabad produces a lot of original thinking in an uncommercial way, and Karachi turns ideas into products because it knows the country’s heartbeat.
Ziyad agrees. When he moved to San Francisco, he says, it was the first time in a while he had heard birds. New York meant 8am to 8pm. In the Bay Area, people went home at five. He thinks that extra room is part of why the Bay Area is more creative. He saw Islamabad properly for the first time two weeks before the recording and says that if Safepay ever opens a second office, it would be there.
A Venmo for Pakistan that nobody paid for
Ziyad says few people know where Safepay started. His first idea was a Venmo for Pakistan, an app for sending money to friends and splitting bills. Part of the reason was to teach himself iOS and Android development. He got a merchant account at HBL, which was then the only bank offering a payment gateway, integrated its APIs, and put the app live.
Few people signed up. He learned two lessons. Pakistan was not ready for it, and the app had no business model, because it was free and he was the one paying the fees. Putting an app on the App Store, he says, is a small part of building a business. Distribution and product-market fit matter more.
He went back to his job for a couple of years and moved into back-end programming to learn Golang. Meanwhile his father kept sending him WhatsApp articles about Pakistani companies trying similar things. The fear of missing out worked. Around 2018 or 2019, single and still employed, he decided to try again and take it all the way.
Someone had once drawn him a triangle: a full-time job, a social life and a startup. You can only pick two. Ziyad kept the job because he needed the money and gave up the social life. He spent six to eight months building a way for businesses to accept card payments, then posted a short note in the Pakistan Startups group on Facebook asking for feedback. The product was a plugin for WooCommerce stores. It struck a chord. He spent late nights in the US walking Pakistani store owners through installation in their morning. One early customer, Smart Choice, still uses Safepay. It was also the customer that pointed out there was no dashboard to see payments, so he built one.
One merchant ID, an Excel file, and Y Combinator
Muzamil asks how one person in the US could do this without a licence. Ziyad says they were naive. They did not know a licence covered their work. They did know KYC, so they built their own controls: the ability to switch a merchant on or off, and limits on how much and how often each business could accept.
The setup was simple. Safepay was a single merchant on HBL’s gateway, with one merchant ID. Every merchant’s money landed in Safepay’s account, and a ledger tracked who was owed what. Each day Ziyad collected cash-out requests into an Excel file and uploaded it to HBL’s corporate portal. For pricing, he copied Stripe: 2.9 percent plus 30 rupees.
On the back of early growth, Safepay applied to Y Combinator. The first application got an interview, and Ziyad admits they went in too confident. The questions humbled them, and the rejection came a few days later. YC encouraged them to reapply. The second time, a few things lined up. COVID had started, fintech was rising, and YC had just backed its first Pakistani company in the previous batch. They prepared an answer for every question they could imagine and got in.
During YC, Ziyad recalls, Safepay processed about $350,000 in one month and was growing 10 to 15 percent month over month, with a team of two or three. Muzamil asks whether anyone worried about tax when that much money passed through one account. Ziyad says they were too Americanised and too sheltered to think about it. Money came in and money went out.
By then he had a co-founder. In late 2019, a childhood friend joined him. The friend was a financial regulatory lawyer who had studied in the UK and worked at a law firm with clients such as the UK’s central bank and Barclays. That experience mattered for what came next.
Shut down overnight, then three and a half years for a licence
After YC, Safepay was flying high. Ziyad says it was the first Pakistani fintech accepted into the program, and it had raised from Stripe and other VCs. Then, in October, the bank processing its payments cut it off. Overnight, Safepay could not process a payment for any business. The bank pointed to the lack of regulation and licence. Ziyad accepts some of the blame. They had depended on a single entity that they were also competing with.
The industry helped. PayFast, a competitor, reached out and offered its APIs. Safepay switched to bank-to-bank payments, without cards, just to stay alive. It also applied for a licence, mainly for independence.
Muzamil has called Pakistani banking a club, and Ziyad partly agrees. He doubts the State Bank would have taken them seriously without the funding and Stripe’s name. He also credits luck. His grandfather had served as a governor for one year in 1990, and the family name still carried weight. The State Bank encouraged the application and let Safepay operate while it was reviewed.
“We didn’t realise it would take three and a half years to get the licence,” he says. Safepay was a guinea pig, as NayaPay and SadaPay had been before it. He says NayaPay’s EMI licence took four to four and a half years and SadaPay’s took three to three and a half. In 2025, he says, the State Bank is trying to fast-track applicants who show a sound regulatory posture.
He names the trait that got them through. “If there’s one thing that’s important for a startup, it’s determination,” he says. There were many moments when the founders could have returned the money and gone back to comfortable jobs in the Bay Area.
What Visa and Mastercard really charge
The licence lets Safepay work directly with Visa and Mastercard. Ziyad expected a simple fee he could mark up. He was wrong. “Visa and Mastercard have some of the most complicated, complex pricing that I’ve ever had to experience in my life,” he says. There is a monthly fee for membership, a fee per transaction, a fee per attempt whether it succeeds or fails, and cross-border fees on international payments.
The 30 rupees in Safepay’s original price never went to Safepay. It covered a fixed cost of about 11 cents per transaction. The job, he says, is to turn that tangle into one simple price for a merchant and absorb the variance behind it. Safepay now charges less for domestic cards than for international ones, because the networks do the same.
Muzamil shares his own pain. He tried to accept Pakistani cards through a Dubai Stripe account and found that most Pakistani banks flagged the payments as fraud. Ziyad explains why. Banks outsourced their e-commerce technology to vendors and invested in ATMs and point-of-sale machines instead. Before COVID, Pakistan was brick and mortar.
Refunds show the problem. Safepay, as the acquirer, can send a refund instruction, but the customer’s bank, as the issuer, has to credit the money. Often it does not, and Safepay ends up chasing the bank by hand. He recalls a merchant who saw a 70 percent acceptance rate at Alfalah, 60 percent at another bank, and 5 percent at a third. On the call to fix it, the third bank brought its technology vendor, and nobody knew what was happening behind the scenes. For a company whose bread and butter is payments, Ziyad says, you have to own the product and the technology and understand everything, or it cannot work. Banks have improved since fintechs arrived, he adds, but they juggle many businesses. Safepay only does payments.
Safepay today
Ziyad gives the numbers. Safepay processes roughly 1.5 to 2 billion rupees a month across about 400,000 transactions, for large domestic e-commerce brands, some global ones, and startups. It has raised twice, once after YC in 2020 and once in early 2022, about $5.4 million in total across a seed and a seed extension. It has not raised since. Ziyad saw the US about to raise rates and closed the second round fast. The goal since has been profitability, and he expects to reach it within six to eight months. The team is 30 people, mostly in Karachi, with a few remote staff in Lahore and Islamabad.
Stripe, he says, has become an active, helpful investor that never meddles. Stripe staff have joined Safepay in meetings with the State Bank and written letters to the regulator on its behalf. Other backers had their own links to Pakistan: Global Founders Capital, the investment arm of Rocket Internet; HOF Capital, founded by the Sawiris family, who started Mobilink; and Fatima Gobi Ventures.
He names the top players in the gateway market as Alfalah, HBL, Meezan Bank, PayFast and Safepay. At one point, he says, Safepay was doing the same number of transactions as HBL. Its licence is a payment service provider licence, the same category as PayFast, 1Link and NIFT. NayaPay and SadaPay hold EMI licences, which let them open customer accounts and take deposits.
Payouts, Raast and the end of the card habit
Ziyad does not plan to lend or offer buy now, pay later, though Safepay can power BNPL providers. The next product is payouts: helping a business pay its vendors or drivers. His example is InDrive, which collects from riders and must then pay out to many drivers. Muzamil notes that this is the Excel problem Safepay once had, turned into a product. Ziyad adds that merchants now ask for APIs, documentation and SDKs, which was rare five years ago.
Then comes Raast. Muzamil cites a Profit Magazine piece on government targets for 2026: 2 million digital merchants, 120 million online banking users and 15 billion digital transactions. He asks whether uptake will suddenly snowball.
Ziyad names two brakes. Cards are expensive in a price-sensitive market. And merchant onboarding is slow, because payment providers must follow bank-level KYC. A street vendor selling mangoes will never be onboarded, he says, if he is asked for a bank maintenance certificate. Laws have to suit the market they serve.
Raast addresses the cost. The State Bank does not charge per payment the way Visa does, but joining is expensive: routers, firewalls, a direct connection to State Bank servers, primary and backup sites, all hosted in Pakistan. Providers may charge a small fee to cover it. Ziyad gives 0.4 percent or 20 rupees as examples. Raast also lets Safepay enter offline payments for the first time, through QR codes at shops.
He says Raast handled about 3.9 trillion rupees of person-to-person payments in the third quarter. Banks resisted at first because free transfers killed their IBFT fee income, and the State Bank made them comply. The same resistance is now meeting P2M, the merchant side. Meanwhile the industry is testing in three or four WhatsApp groups, swapping test payments and QR codes. Ziyad is confident about the result. When P2M launches, he says, “it will drastically shift how people accept payments online.” A shopper would enter a phone number or IBAN and approve a push notification in a banking app instead of typing a card number.
His last ask is about onboarding. If a business already has a bank account in good standing, the bank has done its KYC and keeps doing it. That should be enough, he says, with sanctions screening automated on top. He wants Safepay to work like Stripe, where a merchant uploads documents and gets an answer within an hour.
Crypto, tokenisation and Pakistan in 2050
Ziyad has long been a fan of crypto and wants to explore it as Pakistan begins to regulate cryptocurrencies and stablecoins. He calls it unfair that Bitcoin went from zero to about $110,000 in around 17 years while so few Pakistanis benefited. He sees stablecoins as a route into global markets, and tokenisation as a way to offer fractional ownership of assets such as Karachi plots worth 20 crore. He also wants his parents to be able to research a stock on their phone instead of calling a broker.
He is optimistic about the decade ahead. The government wants digitisation, which makes the State Bank easier to work with. Nearly everyone has a smartphone and the internet, so people know what exists elsewhere and want it at home. He compares Raast to UPI in India and Pix in Brazil, which changed the course of both countries. The State Bank has built the infrastructure, he says. Now it is up to the banks and fintechs to build on it.
Muzamil says it has been one of his favourite conversations. He recalls Adnan Shafi of PriceOye telling him that fintech comes first in an economy, because businesses cannot be built before payments work. After four years of hearing many theories, he has concluded that the core problem was payments and Raast.
For 2050, Ziyad expects Karachi to be bigger and more chaotic. He hopes for serious investment in infrastructure, technology and education. He will be about 60 by then. He hopes Pakistan can again become a model, as it was after independence, when visitors from Singapore and Malaysia came to study its factories. This time the audience might be the GCC and Central Asian countries such as Uzbekistan, Azerbaijan and Kazakhstan.
His deeper wish is about self-reliance. Pakistanis ask why PayPal and Stripe do not come, he says, instead of building their own. “If we don’t help ourselves, no one will help us.” Pakistan is the only country where his generation will be first-class citizens. The government’s job is to keep the train on the track and not rock it, and to let the people on board build. More new companies have been registered in the past five years, he says, than in the whole of the 2000s. He is bullish on Pakistan, he says, and that is why he is here.
Never miss a conversation.
New episodes and the thinking behind them, straight to your inbox. No hype, no spam, no pitch.
