For Unproven Startups, Public Narrative Attracts Talent and Capital

Amjad Masad says Replit’s public voice was not a marketing layer added after product-market fit. For much of the company’s decade-long life, it was part of how the company sustained itself before its recent commercial acceleration.

“Replit would have probably died if I wasn’t telling a story that is larger than the company itself,” Masad says. The company was not initially commercially successful, in his account, but its broader narrative helped it fundraise and recruit.

Masad does not present founder-led communication as a universal requirement. A company that succeeds commercially from the outset may not need its CEO to build a public narrative around it. But when a product’s potential is not yet evident to the market, he sees storytelling as a practical way to attract capital and talent during the period when the company is still proving its case.

At the time, Masad says, much of what he was saying about Replit was “a dream,” even if it later proved accurate. The public work was to make the company’s intended direction intelligible before the business results fully supported it.

Amjad Masad traces his comfort with public communication to work that began well before he built an audience on X. When he first arrived in New York, he says, severe stage fright could push his heart rate to 150 and leave him sweating while speaking. He took improv and storytelling classes, performed shows, and adopted exposure therapy as a working method: do more of the thing that produces fear until it becomes less destabilizing.

Improv was useful because it required regular embarrassment. Masad’s public profile also grew gradually rather than exploding overnight, which gave him room to make mistakes with relatively little attention. He points to a communication dispute with a former Replit intern that he says he handled badly; the controversy remained largely within the Hacker News community rather than becoming a much larger public event.

He compares the process to progressive overload in powerlifting. The weight does not become easier, but the person becomes more able to carry it. That is why he sympathizes with founders whose companies become highly visible before they have developed the instincts to handle mistakes in public.

Masad’s practical recommendation is to turn communication founders already do into public practice when there is a legitimate choice between public and private. CEOs write to employees and investors; rather than producing a separate internal version, he says he will sometimes write something publicly and paste it into Slack. The approach lets outsiders follow the company’s work while giving the founder repeated practice explaining what it is doing and why.

The timing matters. Replit benefited from a long incubation period in which Masad could make smaller mistakes before his audience became large. For a company already under heavy scrutiny, he says, practicing in public may be a riskier way to learn; coaching earlier may be the better route.

Amjad Masad describes a shift away from the CEO whose public voice is tightly filtered through public-relations teams and journalists. He points to Elon Musk and Mark Zuckerberg as leaders who built their own public platforms, and says Zuckerberg changed the narrative around himself by talking directly to people after journalists became more adversarial.

Masad’s emphasis is not simply on informality. He argues that people increasingly respond to a public figure who sounds recognizably like a person rather than an institution. He cites Donald Trump and Kanye West as examples of figures whose short, idiosyncratic posts made their voices memorable, whatever one thinks of them otherwise.

That does not mean a CEO should manufacture eccentricity or seek conflict for its own sake. In Masad’s account, a durable public voice has to be grounded in views the speaker can sustain when the response becomes hostile.

He points to Anthropic as an example. Whatever outsiders make of its actions, Masad says he believes the company genuinely holds its safety and mission commitments. That conviction, he argues, gives its people strength to continue under pressure. Principles, in this formulation, are not only a moral posture; they give a leader something durable to communicate when critics interpret an action as cynical marketing.

Amjad Masad draws a practical distinction between low-impact criticism and an issue that can materially affect the company. Founders do not have to answer every reply, he says, and treating every hostile comment as a demand for immediate defense is a trap. But a consequential misunderstanding, widespread story, or problem with clear business impact calls for judgment—and, where the company made a genuine mistake, accountability.

Masad frames “being canceled” as, at least partly, a choice about whether to retreat from public life after backlash. Continued visibility can eventually reduce the force of opposition, he says, because critics may move on when they conclude they cannot remove someone from the conversation.

Erik Torenberg connects that view to So You’ve Been Publicly Shamed, saying the book’s examples suggest that people most thoroughly defined by public shaming were often those who went quiet, while others rebuilt an image or simply became associated with something new. Masad agrees with the broad point but stresses that the mental cost is real. In his evolutionary account, exclusion from an ancestral community could be materially dangerous; online backlash can trigger fight-or-flight responses severe enough to disrupt sleep and appetite. Exposure helps, he says, but does not make the experience trivial.

The rule is not to be impervious or constantly combative. Masad says a company should correct a consequential misunderstanding, explain what people have wrong, and state its actual position without becoming overly defensive. The relevant question is whether the issue will affect the business.

His account of Replit Agent’s database-deletion incident is his clearest example. SaaStr founder Jason Lemkin, whom Masad says was new to coding and “vibecoding,” used Replit Agent and had a database deleted after the system misunderstood him. At the time, Replit did not provide development-production separation by default. Masad says backups and snapshots made recovery one click away, but Lemkin did not know that. The agent’s visible reaction compounded the episode: Masad says models at the time, especially Claude, could become anxious, and the agent appeared to be panicking in screenshots Lemkin posted.

A Times of India headline shown during the discussion described Masad apologizing after an AI agent deleted a user’s data and “lies about it.” Masad’s account was that the data could be restored, but that the absence of production-development separation was still a real product weakness.

Masad says Replit rolled out the separation two days later. By default, Replit Agent could read from a production database but could not write to it. He says his response received positive engagement because it paired an admission with a concrete product change. He adds that Lemkin later became a major Replit user, running SaaStr’s multimillion-dollar business with himself, another employee, and roughly 10 agents based on Replit Agent.

Amjad Masad rejects the idea that every CEO needs to become an influencer. There are many paths to success, he says, and forcing an executive with no aptitude or interest to post can become net-negative for the brand.

He cites Groq as a case where someone other than the CEO has been the visible public personality. In Masad’s telling, the company had trouble fundraising in Silicon Valley and looked to Saudi Arabia, but its technology and public story helped it “meme it into success.” Dario represents a different model: Masad says he is not playing the social-media game in the same way, instead publishing substantial essays while the product largely speaks for itself.

Claude’s billboard campaign offered another way to state a company proposition publicly. The ads read “Keep thinking” and “If there’s no solution / Code one,” followed by “Claude / The AI for problem solvers.” The message is carried by the campaign rather than by a founder’s feed.

The decision, Masad says, turns on the company’s circumstances and the leader’s disposition: whether the CEO enjoys public communication and can become good at it. A founder may improve with time, but Masad does not treat a founder-led social account as inherently advantageous.

Platform choice should follow the audience a company needs to reach.

Masad says X can also start stories that travel further. On the day of the discussion, he says, journalists had contacted him about a post concerning Anthropic. But he considers X a risky dependency: algorithm changes can alter reach with limited transparency, and he sees the environment as less transparent than “Twitter 1.0.” Even so, he considers the platform difficult to avoid or replace despite its name change and damage to its brand.

He says he began emphasizing Instagram only recently and found that clipped video generated substantial views and rapid follower growth. Interviews can be useful raw material because they can be turned into multiple formats and used to grow more than one account. He would still begin with X, he says, because it can provide outsized influence even if it does not directly reach the eventual mass audience.

Amjad Masad says the hard part of going direct is not producing posts. It is understanding the “meta”: the larger debate already commanding attention.

He praises Satya’s longer essays as arguments that were well reasoned, connected to a moment when people were focused on token costs, and advanced Microsoft’s point of view within that debate. Posting without that connection can result in no attention at all, Masad says.

Masad made the point more tersely in an X post shown during the discussion: “the way to gp viral is to dress up your argument, your worldview, in the current conversation.” Another displayed post quoted his claim that people should no longer learn to code, but should instead learn “how to think,” break down problems, and communicate clearly. Both place a Replit-adjacent position inside a broader argument already legible to an online audience.

Masad says he learned from a pandemic-era “going direct” group associated with Balaji and Lulu, where participants discussed and critiqued media strategies. But he differs from an antagonistic posture toward reporters. Journalists are people responding to incentives, he says; founders can build relationships with them and help them without giving up the ability to communicate directly.