In June, Superhuman acquired GPTZero, a New York-based startup best known for building one of the most popular tools for detecting A.I. in student writing. Its young co-founders and team have since joined Superhuman’s newly formed “authenticity” group, where they’re working to integrate detection capabilities across the company’s products, most notably Superhuman Go, an A.I. assistant designed to operate across everything a user does online.
The deal marks the latest step in a rapid transformation led by CEO Shishir Mehrotra, who took the helm in early 2025 after the company, then called Grammarly, acquired his productivity startup, Coda. That summer, Grammarly bought Superhuman Mail, a power-user email tool built around the idea of “zero inbox.” In October 2025, Mehrotra renamed the combined company Superhuman, signaling an ambition to build a suite of A.I. tools beyond a writing assistant.
Before founding Coda in 2014, Mehrotra built and launched products at YouTube and Microsoft. He is replicating a familiar Big Tech playbook—Google became Alphabet in 2015; Facebook became Meta in 2021—where a well-known product evolves into something more expansive.
“There’s been enough cases of that being done in a way that preserved the core brand,” Mehrotra told Observer at the time.
Grammarly, still Superhuman’s most popular product, is used by 40 million people every day. Students rely on it to refine essays; editors use it to catch typos and enforce style consistency—sometimes across sprawling internal rulebooks that can run hundreds of pages.
Using A.I. in writing—arguably one of the most human forms of work—has been contentious from the start: How much use of A.I. is too much? Is A.I. best used for generating a first draft or doing the final polishing? And increasingly, can anyone reliably tell the difference?
Mehrotra has had to navigate those questions publicly. Earlier this year, the company faced backlash over its “Expert Review” feature, which generated suggestions in the style of well-known writers and led to a class-action lawsuit over consent.
“There is a clear line, depending on the task, between things you should use A.I. for and things you shouldn’t,” he told Observer in an interview earlier this month. “In the case of a writer, it might be a great tool to do research or understand a topic. But if it sneaks into the actual writing and pulls away from the human doing the reporting, it reflects poorly on both the writer and the outlet.”
Teachers and editors have long complained that A.I. detection tools are unreliable, while many A.I. researchers argue the problem is fundamentally unsolvable. In any case, Mehrotra is already looking at a bigger picture. “Think of Grammarly as an excellent English teacher sitting on your shoulder all the time, helping you everywhere you work. But it’s actually a massive underutilization of the infrastructure we built,” he said. “The core infrastructure is the ability to bring immersive A.I. right to where you work.”
That idea underpins the company’s rebranding and, not long after, its launch of Superhuman Go, which extends Grammarly’s real-time assistance into a more general-purpose system of agents.
“We are enabling anybody to build an agent that works just like Grammarly, but does things far beyond grammar,” the CEO explained. “For a salesperson, that might mean having a sales coach next to you, warning you that you’re about to recommend the wrong product as you email a customer. For a support agent, it might mean a support coach reminding you that a customer had a big outreach yesterday and suggesting you mention it.”
Looking for human qualities in an A.I.-saturated world
Superhuman’s many departments–from engineering to sales to customer support–use almost every A.I. tool on the market, Mehrotra said. But when information reaches the CEO level, he faces his own test in drawing the line on A.I. usage.
“I often tell people: use A.I. to help you form your opinions, but not to write your briefings and reports,” he said. “When people send me something that clearly feels like it was written by an A.I., I’ll often tell them, ‘I’d rather just have the prompt.’ I would much rather see your unique, raw insight and start from there, rather than reading a polished A.I. output.”
The prevalence of A.I. has also changed the hiring process. Tech leaders like Bill Gates predicted early on in the generative A.I. boom that incorporating A.I. in work would be like having a digital personal assistant all the time. And that has changed what executives look for in job candidates, even for entry-level positions.
“A.I. has added another bar for almost every role; we’re turning everybody into a manager very early in their experience. You’re going to have a team working under you from day one, which is completely different from how it used to work in the pre-A.I. world,” Mehrotra said. “All of a sudden, we have to look for very similar skills [to managers] even for entry-level candidates. Your ability to work with A.I. to produce the right result is incredibly similar to your ability to work with a team of human employees.”
What’s left is looking for uniquely human qualities. Mehrotra and his Coda co-founder, Matt Hudson, have evangelized the idea of applying the “eigenquestion” discipline in decision-making—a borrowing from mathematics that reframes prioritization.
“We often find ourselves in situations where we have ten questions to answer and have to prioritize them,” he explained. “People typically rank them by severity or urgency. But if you use the eigenquestion test, you look at that list and ask: ‘If I answer this single question, how many other questions on the list does it answer?’ It might be question number six on a list of ten.”
This way of thinking is especially important in making strategic decisions. As A.I. becomes highly capable of generating plausible answers to everything, the scarce resource becomes deciding which problems are worth solving and which outputs to trust.
“When I think about the misuses of A.I., I think that people often try to substitute human insight with A.I. When they do that, they end up, at best, embarrassing themselves, and at worst, actually producing wrong outcomes,” Mehrotra said. “One of the things that is incredibly human is judgment with very little information. Ultimately, knowing what problem to solve also seems, at least for now, uniquely human.”
