The useful thing about a grammar mistake is that it announces itself. A bad product idea is less courteous. It may sit in a pitch deck wearing sensible shoes, nodding at every market chart, until the money is gone. Alex Shevchenko learned the difference early. His first real signal was not a graph. It was a pocketful of business cards gathered while walking the floor of an American education conference, explaining a plagiarism checker to strangers.
The company was called MyDropBox, years before the other Dropbox turned the word into a blue icon. Shevchenko and his university friend Max Lytvyn had found the problem in Kyiv, in classrooms where instruction was in English and the internet was still new enough to feel like a secret passage. Some students discovered that an essay could be lifted from the web and handed to an instructor who did not yet imagine such a thing. Shevchenko and Lytvyn were the honest students in the room. Honesty, assisted by irritation, became market research.
The first draft was a detective
They began working on document originality in the early 2000s, built the technology, and tried to sell it from Ukraine. The software could travel instantly; trust could not. A young founder with an accent, pitching major American universities from Kyiv, carried several kinds of perceived risk before anyone opened the demo. So the founders moved closer to the customer. Shevchenko entered the MBA program at the University of Toronto. Lytvyn studied in the United States. The company took shape across Canada, Ukraine, and a collection of difficult sales conversations.
At that 2004 conference, doubt went in with Shevchenko and business cards came out. Potential customers wanted to try the product. MyDropBox grew to serve more than two million students in two years and was acquired by Blackboard in 2007. It was a tidy entrepreneurial ending, except the most interesting part of the problem remained untidy. Catching borrowed language did not explain why a person felt unable to write in their own.
“Happiness is not in money, but in making beautiful things that millions of people around the world use.”Alex Shevchenko, in a 2012 founder interview
By 2009, Lytvyn had completed his post-acquisition commitment to Blackboard. He packed his things in Washington and returned to Toronto. He and Shevchenko, joined by engineer Dmytro Lider, started again. Their new product would not police the finished paper. It would help with the sentence while the writer was still trying to make it behave.
A correction without a scold
Grammarly began as a paid online editor for students. It checked the mechanics: spelling, punctuation, grammar. The founders were creating a new category, but users supplied the map. People brought emails, applications, social posts, and professional documents. A product built near the end of an academic workflow was being pulled toward every place language could create confidence or embarrassment.
Shevchenko became the product lead, the role that suited his particular obsession. Writing assistance is technically difficult and socially delicate. The software must interrupt without heckling. It needs to be right often enough to deserve attention and restrained enough to survive being wrong. For years, he personally mentored and managed Grammarly's design team. The ambition was not merely smart software. It had to feel graceful, welcoming, empathetic, and engaging. A red underline, after all, is a very small piece of interface with the emotional range of a disapproving schoolmaster.
The decisive distribution change arrived in 2015. Grammarly released a free browser extension and moved from subscription-only to freemium. Instead of asking users to carry a document into a special editor, the product appeared where the sentence was born. Daily use passed one million. Versions followed for more browsers, desktop software, mobile keyboards, office tools, teams, and enterprises. The company had not made grammar more exciting. It had made help less inconvenient.
The shrinking distance between thought and help
This is a product-position diagram, not a usage chart: each step places assistance closer to the work itself.
For most of this early run, the company used its own revenue rather than venture capital. Grammarly was cash-flow positive while still a paid editor, then waited until 2017 to announce its first institutional round. Shevchenko has been blunt about the hierarchy: investment is a resource, not the goal. The distinction mattered because money could accelerate a model, but it could not decide which model deserved acceleration. The company had already learned that from subscription customers before opening the free door.
Freemium brought a different kind of evidence. Every writing surface became a small laboratory for intent. A suggestion in an essay did not carry the same stakes as one in a sales proposal. A correction in a private draft was not the same social act as advice in a public post. Grammarly gradually expanded from accuracy to clarity, effectiveness, fluency, and tone. It also expanded from individuals to teams, where one person's phrase can become a company's voice. The product's territory was no longer grammar in the schoolbook sense. It was the gap between what someone meant and what another person was likely to understand.
The brick walls get smarter
Success did not make Shevchenko more romantic about founders. The first exit taught him that charging through walls works only while the walls are thin. At higher levels of company building, he said, they become too solid for force. The answer was to surround himself with people who knew more. He has described the pleasure of pointing to anyone in the company and saying that person is smarter than he is.
This sounds modest, but it is also practical. Grammarly kept becoming a different company: education technology, consumer subscription, free extension, enterprise service, AI writing partner. Each turn retired a little founder certainty. Shevchenko's public comments repeatedly return to colleagues adapting, growing, and becoming experts. His official biography makes the point almost tenderly: he finds inspiration in watching team members thrive. The product manager's unit of progress is not only a feature. It is a person whose judgment has become sharper.
There is discipline behind that patience. One company biography noted that Shevchenko meditated every day for more than two thousand days in a row. It is a comic fact if you picture the streak counter surviving funding rounds, product launches, and browser bugs. It is also consistent with a career built around small interventions repeated at scale. One suggestion does little. Millions of suggestions alter how the internet sounds.
When the checker meets the generator
Generative AI complicated the old bargain. A writing assistant once corrected a sentence the user supplied. Now software could supply the sentence, the paragraph, perhaps the entire submission. The company had started in plagiarism detection and returned, by a much stranger road, to authorship. In 2024 Shevchenko introduced a set of originality features combining plagiarism checks, citations, and AI detection. His argument centered on transparency and accountability, especially in education.
That response reveals the continuity beneath the fashionable vocabulary. MyDropBox asked whether the words came from somewhere else. Early Grammarly asked whether the words conveyed what the writer meant. The AI era asks how much of the work remains legibly the writer's own. The detective, the coach, and the generator now occupy the same browser tab. They must learn table manners.
The higher the wall, the more important it is to hire people who know the way over it.The operating lesson after MyDropBox
The name changes, the question stays
Grammarly acquired Coda in late 2024, then the email company Superhuman in 2025. The combined company adopted the Superhuman name and arranged its products into a suite spanning writing, documents, email, and a proactive assistant called Go. Shevchenko now sits on the Superhuman board and works on product strategy and innovation within the Core Product team. He is no longer simply asking what belongs under a misspelled word. The canvas is the workday.
The suite promises help that arrives in context, sometimes before the user explicitly asks. This is the logical destination of the journey from editor to extension. It is also a delicate place. An assistant close enough to anticipate can become close enough to annoy. The old design problem returns in larger clothing: how should software intervene without taking command?
Shevchenko's career offers no slogan that resolves the question, which is fortunate because slogans are terrible product managers. It offers a method. Notice a real human difficulty. Stand near enough to hear whether anyone cares. Move the tool closer to the moment of need. Hire people who can see what you cannot. Let the product change while the human problem remains in view.
The copied essays in Kyiv were the first draft of that method. MyDropBox caught them. Grammarly tried to make them less necessary. Superhuman now wants to help with the work around the words as well. The scale has changed, and so has the technology. The blinking cursor is still there, waiting with perfect manners for someone to decide what ought to happen next.