In November 2012, Merrill Lutsky was in a Mountain View motel room when the call came: his startup had been accepted into Y Combinator. He and his co-founder would leave Harvard and build their company in California. For a sophomore interested in technology, it was an invitation with considerable gravitational pull. A year later, he published an essay suggesting that they should have thought harder about saying no.
That small reversal makes a useful introduction to Lutsky. The acceptance letter mattered. So did his willingness to reconsider what it meant. His essay had a title that left room for circumstances: “Don’t drop out. (until you have to)”. He eventually returned to Harvard, finished his degree, and met the two people with whom he would build Graphite.
Years later, the same question returned in a different form. He and his friends had funding, engineers, and a developer-tools company. They also had an internal project that seemed to interest people more than the product they were trying to sell. Changing direction would mean giving up work they had already done. Continuing would mean ignoring what was happening in front of them.
Before the motel room, Northfield
Lutsky grew up in Northfield, Minnesota. In a 2011 profile of local National Merit finalists, he described an early interest in mathematics and science, encouraged by his parents, teachers, and the surrounding Carleton College community. Asked how he had reached that point, he gave credit to educational opportunity and luck alongside his own interests. The answer contained an unusually long thank-you list for a school-leaver.
He was still weighing Harvard against Princeton then. He imagined applying mathematics, economics, and science to contemporary problems, perhaps through research, perhaps through a company. The uncertainty was explicit. A doctorate featured in his plans; a Silicon Valley startup was a possibility. The teenager had interests before he had a job description.
Robotics for Science Olympiad helped introduce him to programming. A visiting-weekend encounter with Hack Harvard, where students talked about computer science and startups, helped draw him to the university. The attraction was partly social: here were people who wanted to make things and discuss how companies worked. Before Graphite had a product, Lutsky had found a conversation he wanted to keep having.
A survey, a sale, a return ticket
His first company, Posmetrics, began with a Harvard computer-science project. Lutsky and Erik Schluntz were interested in a familiar absurdity: businesses kept asking for feedback through surveys that the two of them never completed. Putting the survey on a tablet, directly in front of the customer, removed some of the effort between being asked and answering.
They joined Y Combinator’s winter 2013 batch. Posmetrics focused on hotels, where feedback could lead to more specific action. Schluntz later recorded that the company reached six employees and more than 50 installed customers before its acquisition in July 2013. It was a short company history with a fairly complete set of first-time-founder experiences: making a product, finding customers, hiring, and selling the business.
Lutsky returned to Harvard. By 2015, he was using advanced standing to graduate with his original class. In the meantime, he had learned what an accelerator’s instruction to talk to customers could mean in practice. He described sales calls spanning the working day from the East Coast to the West Coast. There are glamorous versions of the college-founder story. Eleven hours on the telephone supplies a useful corrective.
and is acquired
completed
Reimers and Foster
Friends who read the filings
Back on campus, he met Tomas Reimers and Greg Foster. They were younger computer-science students, already accustomed to working together. Lutsky brought experience from Y Combinator; they brought questions. The friendships outlasted graduation and separate jobs. Their shared interests included reading companies’ S-1 filings, an activity unlikely to trouble the campus party calendar.
They subsequently saw how software was built inside larger organizations. Lutsky worked in product at Square and later in other product and engineering roles. Reimers worked at Meta; Foster’s experience included Google and Airbnb. Those workplaces had tools and habits that were easy to miss after leaving. A developer can take a laptop to a new job. An entire internal engineering platform is harder to pack.
The three incorporated their company in March 2020. Their early experiments included a tool for capturing software bugs and a way for native iOS applications to roll back problematic changes. The broad interest was consistent: give teams access to capabilities that large technology companies had built for themselves. Finding the particular capability that customers urgently wanted took longer.
The useful thing in the corner
The code-review tool grew out of work on the iOS product. Engineers joining from Meta missed their previous workflows, so the team built something to make its own work easier on GitHub. Other developers became interested. Here was an awkward development for a small company: its internal convenience was beginning to look like the business.
Lutsky has described the period as difficult and uncertain. The existing product had users and commitments from companies willing to pay. Abandoning it was therefore a decision with consequences, rather than a tidy insight available on demand. Interest in code review nevertheless kept arriving. The team followed it, and Graphite became the focus.
The distinction mattered personally. With Posmetrics, Lutsky had learned that he wanted to understand and care about the people using his software. Developers were colleagues, friends, and the founders themselves. A complaint no longer required a research expedition. They could encounter it while doing their own work, fix it, and feel whether the fix helped. That proximity gave their next product a different starting point.
“We personally use the product every day.”
Merrill Lutsky, on building for developers
The queue after the clever part
Code review sounds like a narrow subject until you consider how software reaches a customer. Writing a change is one stage. Someone must inspect it, tests must run, and the change must be combined with other work. A faster author can still spend the afternoon waiting. Add AI-generated code and the pile awaiting attention can grow faster than the team’s capacity to examine it.
Graphite worked on that space between creating a change and shipping it. Its approach included stacked pull requests, which let developers divide related work into smaller changes, alongside review tools and a merge queue. The underlying appeal was practical. Small pieces are easier to examine; organized dependencies help work keep moving. The person waiting for approval gets something better than another opportunity to refresh a browser.
By March 2025, Lutsky was announcing a $52 million Series B led by Accel, with participants including Menlo Ventures’ Anthology Fund with Anthropic, Shopify Ventures, and Figma Ventures. The announcement also introduced Diamond as a standalone AI reviewer. The product was intended to identify problems and suggest changes within the review process. Later that year, the Diamond name was retired and the capabilities came under Graphite Agent.
A Friday appointment with unfinished work
Lutsky’s account of running Graphite gives the company a more ordinary rhythm than its funding announcements suggest. He described two-week building cycles and a Friday show-and-tell, where people shared what they had made. Work that could not fit into a cycle needed smaller milestones. The team used Graphite itself, including its AI reviewer, in daily development.
That arrangement puts product judgments close to consequences. A confusing review interface becomes an interruption in your own afternoon. A useful improvement reaches the people making the next improvement. It is an appealing feedback loop, although it also requires remembering that customers’ codebases and working habits can differ from your own.
Away from the company, Lutsky’s interests include running, climbing, electronic music, and contemporary art. He has DJed, and his Harvard years included cycling. A university cycling archive places him in a photograph at the Northeast Velodrome in New Hampshire. It is a welcome change of scenery in a career often described through software: wheels, helmets, a bend in the track, and no pull request awaiting approval.
The next change of course
In December 2025, Lutsky announced that Graphite had signed an agreement to join Cursor. He described an ambition to bring creating, reviewing, and merging code into one collaborative process for people and agents. The announcement said Graphite would continue as an independent product and that the entire team would join Cursor when the transaction closed.
The fit was straightforward to understand. Cursor helped generate changes; Graphite helped teams examine and combine them. Linking those activities could reduce the repeated handoffs between tools. In March 2026, Cursor Cloud Agents arrived inside Graphite, allowing developers to start agent work and request revisions from the pull-request page. A direction described in an acquisition announcement had acquired buttons people could actually use.
It is tempting to arrange Lutsky’s career as a sequence of vindications. The details make it more interesting. The first company produced an exit and a return to college. The second found its direction in something built for internal use. The friendships survived separate employers and uncertain product ideas. Each decision depended on noticing that circumstances had changed.
The motel-room call remains a good opening because it catches him before any of those outcomes were settled. He had an opportunity worth taking seriously, and later enough distance to examine his response. Graphite would ask for that capacity again. For a founder whose work concerns helping people review changes, there is something fitting about a career repeatedly shaped by reviewing his own.