At 18, John Ta walked into a bank near the University of Pennsylvania hoping to apply for his first credit card. He walked out with questions. Secured or unsecured? What was an APR? How, exactly, would this mysterious score follow him into the future? The encounter was mundane, faintly embarrassing, and productive. Confusion has a way of becoming curriculum when it lands on the right person.
Ta was a first-generation college student studying biochemistry and biophysics. Medicine initially looked like the proper road. He worked in a Penn medical-school lab and taught organic chemistry and partial differential equations. Yet the bank visit stuck with him. So did a conversation with a friend who did not know what a credit score was. Penn had no dedicated undergraduate group for sorting out these questions. In 2019, Ta and Vivek Olumbe started one.
They called it Common Cents, a name with the good manners to explain itself. It began as Penn's first personal-finance club and grew into a broader collaborative, adding chapters at Johns Hopkins, Cornell, and Florida State. By the time Y Combinator told the founders' later story, the nonprofit had reached more than 2,500 members across eight universities. A campus question had acquired an organizational chart.
There is a pleasing escalation here. First Ta needed one card explained. At 22, he delivered a Common Cents event called “The Power of Points - Why I Have 15 Credit Cards.” Two years later, Y Combinator's launch announcement put the collection at 30. His LinkedIn biography eventually compressed the whole preoccupation into six cheerful words: “i like credit cards & points.” The wallet was getting crowded, but the curiosity was becoming useful.
The untidy education of an operator
Common Cents taught Ta a part of building that profiles often edit away: administration. In a 2021 interview, he spoke about strategic vision and measurable impact, then immediately mentioned workers' compensation, nonprofit bank accounts, boards, and fiduciary duties. The paperwork took time from the educational work he liked. His colleagues carried more of the teaching while he kept the organization afloat.
That is not glamorous founder training. It is useful founder training. Institutions are made of missions, certainly, but also accounts, permissions, handoffs, and the occasional pain in the neck. Ta was learning to look for the spot where a worthwhile activity gets trapped inside a bad process.
“Don't close doors off early.”John Ta, on advice to his younger self
He had closed one door too soon by assuming the future was pre-med. Entrepreneurship, he later said, had barely been a category in his mind; even filing his first patent felt to him like pre-med work. Two years of exploration changed the map. He found that making an organization from zero offered a more immediate kind of satisfaction. His advice to his younger self was compact enough to fit on a sticky note: keep the doors open.
A career assembled from adjacent puzzles
Ta tested several doors. He spent a summer in M&A investment banking at Leerink Partners, became a venture fellow with Ripple Ventures, interned at Notable Capital, and joined Accenture as a strategy analyst. The résumé could look restless if read as a search for titles. Read as an education in financial systems, it looks more coherent. Research taught him how to interrogate evidence. Teaching showed him where explanations fail. Banking and consulting put him near regulated institutions. Nonprofit work supplied the operational bruises.
Then came Roame, a Y Combinator travel startup focused on points. In a founding growth role, Ta wrote more than 200 articles about credit cards and travel and managed affiliate relationships across the content. Here the hobby met the machinery.
The missing one percent
Affiliate marketing is a simple arrangement with complicated consequences. A publisher writes about a product, a reader follows a link, and the publisher may earn money from the resulting application or purchase. In financial services, every annual fee, rate, comparison, and disclosure matters. An old number beside an application button is not merely sloppy copy. The brand whose product appears on the page can face regulatory and reputational trouble.
The prevailing oversight process asked publishers to submit their own pages, often by email, so someone could review and label them. Ta knew the weakness personally. He later wrote that, as a writer, he submitted only about 80 to 90 percent of what he produced. Scale that tiny failure across a large publisher with 10,000 pages and the unmonitored remainder stops looking tiny. One percent is 100 pages. A spreadsheet cannot inspect what it does not know exists.
Keywords offered an incomplete rescue. A rules-based system searching for “American Express Platinum” might miss “Amex Platinum,” “Amex Plat,” or another perfectly human variation. A page could mention several cards, each with different terms. Context mattered. The work needed software that could first discover relevant pages, then understand what each passage was actually discussing.
Ta and Olumbe had already built together once. They had met while living two doors apart in Penn's Hill College House, and Common Cents had tested the partnership over years. Affil.ai turned them toward a commercial version of a familiar instinct: find the financial process people tolerate because nobody has made it legible.
Sell the promise before polishing it
Affil.ai joined Y Combinator's Summer 2024 batch in San Francisco. The product monitors affiliate and other third-party marketing content, finds brand mentions, identifies relevant products, and checks pages for disclosures and policy problems. Ta and Olumbe described a larger ambition at launch: compliance and auditing embedded in the affiliate lifecycle, eventually supporting a network for financial institutions.
From the open web to a fixable issue
The founders did not arrive with a sacred first idea. In a later interview, they discussed an earlier Y Combinator rejection and the discipline of selling before building. Ta's example involved a points-booking service. Asking people what they might pay yielded friendly hypotheticals. Charging $200 per business-class seat yielded a customer and a far more informative answer. Money, in this telling, is a truth serum with a receipt.
The distinction helps explain Affil.ai's shape. It did not spring from a generic wish to “use AI for compliance.” Ta had been the affiliate writer. He had submitted the URLs, managed the relationships, watched offers change, and seen how the oversight chain frayed. The product started with the customer's nuisance, not the model's parlor trick.
In May 2025, Ta announced that Affil.ai had hired its first full-time employee, founding software engineer Vishal Vinjapuri. His explanation was characteristically operational: customer demands were being missed, engineering work needed a dedicated owner, and the hire reduced the dependencies running through Olumbe. Growth appeared not as a victory lap but as a list of bottlenecks that had finally become too expensive to keep.
By August, Ta wrote that Affil.ai's compliance system reviewed approximately one-third of newly produced credit-card content. The statement is the company's own measure, but it gives scale to the original irritation. The student bewildered by one credit-card application had become an executive trying to keep a substantial slice of the category's publishing accurate.
The hobby was not beside the work. It was where the work had been hiding.
There is an easy temptation to tidy Ta's story into inevitability. Biophysics led to analysis; Common Cents led to financial education; Roame led to affiliate marketing; therefore Affil.ai. Real lives are less considerate. He explored medicine, venture capital, banking, consulting, nonprofits, teaching, and writing. The line became visible only after enough dots existed to connect.
What persists is his attraction to overlooked friction. A friend cannot explain a credit score, so build a club. Students lack personal-finance resources, so make a curriculum. Publishers cannot reliably submit and review everything they write, so build a monitoring system. Each project begins with someone saying, in effect, “surely there is a better way,” followed by the unfashionable work of proving it.
Ta's 30 cards make the memorable prop. The better detail is his willingness to count what slips through. Fifteen cards can become a lecture. Two hundred articles can become firsthand expertise. One missed percent can become a company. Curiosity gets interesting when it develops bookkeeping.