Summary

Against Erik Torenberg’s setup that Replit is 10 years old and only its last two years went “absolutely vertical,” Amjad Masad says Replit likely would have died without founder-led storytelling bridging the years before commercial traction. The story supported very successful fundraising and recruiting while he tried to “meme a dream into reality”; what was once a dream later proved true.
Going direct is a trainable operating capability, not an innate CEO trait. Masad went from stage fright and a heart rate near 150 to improv, storytelling classes, and public writing; his personal profile’s linear growth let him treat early mistakes as “progressive overload” before the stakes became massive.
Crisis response should follow business impact: silence starves fringe attacks, while material failures demand ownership and a fast product fix. After Jason Lemkin’s database was deleted, the mistake was one click from being fixed, but Replit still lacked development-production separation; Masad said, “We messed up,” and shipped the fix within two days. Lemkin now runs his multimillion-dollar SaaStr business with another employee and roughly 10 agents.
On cancellation, Amjad opens with “Being canceled is a choice,” while Erik argues that retreat makes a PR crisis deadly and persistence makes it fade. Staying visible can eventually exhaust haters, but Masad does not romanticize the mental cost: online rejection can trigger fight-or-flight, lost sleep, and inability to eat, so exposure must be paired with principles.
X is high-leverage for Silicon Valley, early adopters, and journalists who turn posts into stories, but it does not deliver the early majority by itself. Instagram, Facebook, and YouTube broaden the funnel; interview clips started scaling Masad’s Instagram only around February, which he says was one or two years late.
There is no mandate for every CEO to become an influencer; fit, product strength, and message architecture matter more than posting volume. Dario can let product and essays speak while Groq uses Sunny rather than its CEO as the public face. To go viral, Masad says, a poster must understand the meta and dress a worldview in the live debate, despite X’s opaque reach risk.

Deep dive

1. Storytelling bought Replit enough time for traction

Erik’s setup makes the time lag explicit: Replit is 10 years old, but only the last two years went “absolutely vertical.” His question is causal—what did going direct contribute during the long pre-breakout period?
Amjad’s answer is existential: “Replit would have probably died” without a narrative larger than the business. Commercial results were weak, but public storytelling supported fundraising and recruiting while he tried to “meme a dream into reality”; only later did the dream become true.
The boundary condition prevents this from becoming CEO-influencer dogma: a company successful immediately might not need the game. Amjad contrasts Dario, whose product “sort of speaks for itself,” with Groq “with a Q,” where Sunny rather than the CEO became the public personality.

2. Public influence is trained through progressive overload

Masad traces communication skill back to training, not platform tricks. Newly in New York, stage fright brought his heart rate to about 150 and made him sweat; improv and storytelling classes became “exposure therapy,” eventually leading to live shows.
Replit’s slow start gave him a forgiving practice field. His personal profile grew linearly, so botched communication—including an ex-intern dispute that became a Hacker News crisis—happened before the stakes exploded; he compares the escalating load to progressive overload in powerlifting.
His practical build-in-public rule is to bias ordinary CEO writing toward public distribution, because he believes few matters genuinely need to stay secret. He often publishes first and then pastes the same text into Slack, carrying outsiders through the journey while accumulating reps.
The timing caveat is that high-profile companies should not practice in public, Masad says; founders can get coaching earlier instead.
Masad presents authenticity as a break from the polished, PR-filtered CEO archetype: Elon built publicly, while Zuckerberg changed his narrative by going direct. Trump’s “I’ve never seen a thin person drink Diet Coke” illustrates why memorable, unfiltered lines create identification despite negative press coverage.

3. Cancellation only sticks when the target retreats

Erik supplies the hardest-edged operational claim: a PR crisis can be deadly if the target retreats. The shared framing is “Being canceled is a choice”; remaining visible supplies a next chapter, signals that someone is “unkillable, uncancelable,” and eventually makes haters concede they cannot do anything.
Masad’s contribution is the physiological caveat: mass rejection activates an ancestral outcast response—fight-or-flight, sleeplessness, sometimes inability to eat. The first episode can be brutal; the second may feel more like, “Ah, it’ll blow over.”
Principles provide the other stabilizer. Masad believes Anthropic sincerely holds its safety mission, making the government episode a possible “win-win”: if Claude returns, the ban could drive a flood toward Claude; if it does not, Anthropic still wins PR as a principled company. He advises not stressing too much and letting it play out.
His analogy is a Steve Jobs ad featuring a Mac—he thinks the big tower was called Mac Pro—surrounded by tanks. The ad framed government export control as evidence of the product’s power and a reason to buy it.

4. Business impact decides whether a crisis deserves oxygen

On engagement, Masad’s default is restraint: publishing does not create an obligation to answer every reply. Correct misunderstandings only after they gain meaningful airplay, state what you actually believe and what was misunderstood, and avoid defensiveness—visible emotional weakness “feeds the trolls.”
The test is business impact, illustrated by Jason Lemkin’s July experience: Replit Agent misunderstood him and deleted a database when Replit lacked development-production separation. Masad says the mistake was one click away from being fixed, but Lemkin did not know that, and the agent—like models then, “especially Claude”—started freaking out.
As the screenshot spread toward press coverage, Masad rejected a user-error defense: “Hey, we messed up.” Within two days Replit shipped development-production separation; the agent can now only read the production database, and Lemkin later became a huge fan running SaaStr with another employee and roughly 10 agents.

5. X wins elites; clips and timing win broader distribution

Masad maps X to Silicon Valley’s in-group, Hacker News, and perhaps one hop beyond to early adopters. Its indirect power can exceed its audience breadth: journalists treat tweets as the start of stories, allowing elite influence to shape wider coverage.
For the early majority he points to Instagram, Facebook, and YouTube. Starting Instagram around February, he found that podcast clipping could produce “insane” views across accounts and concluded he was one or two years late—though he would still begin with X.
X remains a concentration risk: algorithms shift, reach might fall after angering “the powers that be,” and transparency is worse than Twitter 1.0. Yet Masad argues that even a name change or a damaged brand would not destroy the platform’s value.
The real work is editorial timing: “You have to understand the meta” and dress a worldview in the live debate. Sacks’s essay over the weekend drew 20–30 million views while people were thinking about token costs, and Satya’s essays advanced Microsoft’s point of view.
Masad disagrees with Balaji’s antagonism toward reporters. He argues reporters are people who respond to incentives, says Replit has maintained those relationships well, and credits Shervin Naz’s recent arrival with up-leveling the company’s reporter and broader-media efforts.

Episode 189 - July 17, 2026 Open Model Wars + Claire Stapleton's Dishy Google Memoir + Substack's Slop Fight