The first product Jesse Li and Isabelle Ilyia built together had the market penetration of a private joke. It was a social event-planning app, assembled when they were freshmen at Georgia Tech, and nobody used it. Its commercial afterlife was nonexistent. Its human afterlife became a company.
Li had arrived with an idea and a staffing problem. Ilyia did not know how to code. He offered a bargain: he would teach her if she helped him for free. She later joked that she got the better side of the deal. The app flopped, but the pair discovered something less obvious and more durable than product-market fit. They liked making things together.
So they kept going. A knowledge-graph tool followed, then more half-formed ideas, then internships and jobs without the projects ever quite stopping. They did not perform the familiar founder ritual of dropping out and declaring destiny. They graduated. Li earned his computer science degree with specializations in People and Intelligence, joined Intuit and moved to San Diego. The dream remained deliberately lowercase.
“A startup is a project you're working on until it becomes a company.”Jesse Li
The useful word was “project”
For almost four years, Li and Ilyia called their work “projects.” The word lowered the temperature. A startup came burdened with valuation, spectacle and the possibility of public failure. A project could be tried, abandoned and replaced before dinner. Li has written that the startup ambition once felt foolish to him, the sort of dream tempered by reality and statistics. “Project” gave the ambition somewhere quiet to live.
His time at Intuit supplied another education. He worked on software that matched customers with tax professionals, a problem involving conversion, trust and users with wildly different levels of technical comfort. Cedar's first public story would later describe this as learning the science and fun of making boring software. The parenthetical joke was “don't tell TurboTax.” It is a neat summary of Li's public register: earnest about the craft, allergic to making earnestness sound too grand.
Outside work, he described himself as “hobby-maxing.” Pottery, boxing, badminton and writing filled the calendar. By his own account, it was not a ferociously industrious period. The projects with Ilyia continued anyway. What looked from outside like delay was really an apprenticeship in choosing a partner before choosing the final problem.
The rejection with a return address
Their first Y Combinator interview was for an app that let people comment on and save anything they found online, arranging the web into a shared feed. The interviewer stopped them. YC had backed something similar before, and the outcome had not been good.
Li did a small but consequential bit of emotional accounting. Was this, he asked, their “life submission”? It was not. The interviewer probably knew more about this particular pattern than they did. He read the rejection as information rather than identity. In founder mythology, conviction is often portrayed as refusing to hear no. Li's version of conviction had better manners. He listened closely, then asked for another appointment.
They called the interviewer and received a rare re-interview, one week away. The two dropped everything and built a lawyer-matching app from scratch. They ran a trial, returned to the interview and got into the Winter 2025 batch. A tidy telling makes the week sound like a conversion scene. It was closer to proof of method: the idea could be discarded because the partnership knew how to make another one.
Result: a new product, a second interview and a place in YC W25.
Cedar kept moving
Admission did not freeze the product in amber. Cedar first emerged publicly as an embedded AI copilot for software. The system was meant to understand what a user wanted, guide that user through a complex application and help software makers see why people converted or left. It was an attempt to flatten the learning curve of powerful tools.
The founders built a multi-agent system around that idea. Specialized agents handled chat, interface state and step-by-step guidance. The technical work was real, as was the larger ambition. On Cedar's YC page, the founders said they wanted to make software agentic, moving from instructions toward systems that could execute tasks autonomously.
Then they moved again. Cedar's present focus is the account executive, a professional surrounded by expensive software and still condemned to copy transcripts, prepare meetings, update records, chase follow-ups and reconstruct the state of a deal across tabs. Li and Ilyia spent hundreds of hours with sellers. Ilyia later said they watched 200 account executives do their daily work before turning the observed routines into a product.
Today's Cedar connects to a company's sales data, maps the playbook and creates agents around the work between initial interest and a signed deal. Research before meetings, useful follow-up, CRM administration and multi-threading strategy become parts of one workspace. Cedar says account executives at Cursor, Warp, Corgi and Listen Labs use it. The claim is specific enough to reveal the company's bet: not a general chatbot hovering beside every job, but software shaped around how a particular seller wins.
Software as a game worth playing
Li's most revealing explanation of his motivation is not a market forecast. It is a question about basketball. Why, he asked, does anyone like software more than basketball? Basketball has no civilizational mission. There are balls and nets, rules and repetition, and people play because something about the game appeals to them. Software, for Li, carries the same intrinsic tug.
This matters because the Cedar story contains too many false endings for mission alone to explain the stamina. The event app failed. Other projects remained projects. YC rejected the first pitch. The product that entered the batch was not the product Cedar sells now. Anyone motivated only by being proved right would have found the experience terribly inconvenient.
“Why do you like software more than basketball? There's no mission, it's just balls and nets.”Jesse Li, on the pleasure of building
Enjoying the game does not mean playing carelessly. Li's public writing on sales is preoccupied with discipline: founder-led selling, two-week goals, paying customers and the danger of flattering growth that conceals churn. He argues that the best early sales knowledge cannot be outsourced because the founder needs direct contact with the objections, context and oddities of the customer. The engineer has become a student of the deal.
There is a tidy symmetry here. At Georgia Tech, Li offered to teach Ilyia to code because he needed someone to build with. Cedar now takes the working habits of good sellers and turns them into systems teammates can use. Both arrangements begin with tacit knowledge passing from one person to another. The software comes later.
A company, despite the ice cream
Li and Ilyia have worked from a San Francisco apartment overlooking the Bay Bridge. Their company rituals have included making ice cream every other day and taking a midday break to work out. Ilyia has said those habits preserve the feeling that they are friends working on a project. The difference is that customers now wait on the other end, and other people have joined the work.
In 2026, Cedar announced $3 million in funding and opened its sales workspace to the public. Funding is the sort of milestone that makes a project indisputably legible as a company. Yet the old vocabulary still fits. A project can change without staging a crisis of faith. It can be judged by what it teaches. It can remain playful while becoming consequential.
That playfulness has a practical edge. When a product is allowed to be provisional, customer evidence can outrank founder vanity. The name stays on the door while the answer underneath it improves. Cedar's history makes that flexibility visible: the team did not erase its earlier technical work. It carried the lessons about agents, interfaces and user intent into a narrower workplace where every saved hour can be counted.
The durable thread in Li's story is not the inevitability of Cedar. Nothing about the path was inevitable, which is why it is useful. A student needed a collaborator. A failed app produced one. A rejected interview produced a better question. A broad copilot narrowed into the daily work of sellers. At each turn, Li and Ilyia preserved the partnership and released the premise.
Some founders guard an idea like a family heirloom. Li appears happier treating ideas as sports equipment. Pick one up. Learn its balance. Play hard. When the game changes, reach for the thing that works. The company is what happens when the same people keep showing up to play.