A new wave is crashing over the saturated social media landscape, and it’s being carried by the power of the human voice. Airchat, the brainchild of AngelList founder Naval Ravikant and former Tinder executive Brian Norgard, has quickly captured the imagination of early adopters with its innovative approach to online discourse.

The premise is deceptively simple: Instead of tapping out 280 characters, users speak their posts aloud, with Airchat’s impressive speech-to-text engine transcribing the audio in real-time. As you scroll through your feed, you’re greeted not just with text, but with the actual voices of the people you follow. It’s an unexpectedly intimate experience, allowing online connections to take on new depth and resonance.

For many, it’s the first time they’ve heard the voice of someone they’ve known online for years, an experience that highlights both the novelty and the potential of Airchat’s voice-first approach. The platform manages to split the difference between the ephemeral nature of live audio platforms like Clubhouse and the asynchronous text-based exchanges of Twitter (now X) or Facebook.

“[It’s] like we’re actually having conversations with one another,” reflects one early user, “but without the pressure of a live discussion with strangers.”

This balance of intimacy and convenience seems to be resonating. After just a few days, the platform’s servers were overwhelmed by the surge of interest, forcing it to temporarily pause new invitations. But whether Airchat can maintain its momentum remains an open question, one that may depend on the community it manages to foster in these critical early days.

So far, that community has a distinctly tech-industry flavor, likely a reflection of the founders’ Silicon Valley pedigree and social circles. The first channels to emerge included ones dedicated to crypto and “effective altruism,” the philosophy of using evidence and reason to maximize one’s positive impact on the world – topics that resonate deeply in certain corners of the tech world.

“We’re going to try and put as many of the moderation tools in the hands of the users as possible,” Ravikant explained in one of his own Airchat posts. “We want to be as hands-off as possible.”

It’s a philosophy that echoes Substack’s controversial “hands-off” approach to content moderation, which ultimately led to a schism with high-profile writers like Platformer’s Casey Newton over the platform’s reluctance to remove pro-Nazi speech. Ravikant argues that Airchat should function like a dinner party, where the hosts can ask disruptive voices to leave without silencing dissenting viewpoints.

“We don’t want to moderate for content,” he clarified, “but we will moderate for tone.”

The analogy has its limits, however. A dinner party is an intimate, closed affair; Airchat is a social network with thousands of concurrent conversations across multiple channels. Without robust guardrails and clear policies, thorny issues like hate speech, harassment, and copyright violations could quickly spiral out of control.

After all, this was the same moderation philosophy that led Clubhouse, the once-hot live audio app, to become a haven for anti-Semitic bile and misogynistic rhetoric in its early days – a stain from which it never fully recovered. Airchat would be wise to map out a more sustainable approach before falling into the same trap.

Still, the jury remains out on whether these controversies will actually materialize, or if Airchat’s unique dynamics will help curb the worst online behavior. There’s an intimacy and vulnerability to putting your actual voice out into the world that could foster more constructive dialogue. Or it could open the floodgates to entirely new varieties of toxicity.

For now, Airchat vibrates with the electric potential of the new. But whether it can channel that energy into developing a truly worthwhile platform – one that moves us beyond the limitations and negativity of other social media without descending into fresh problems – is a challenge it has yet to overcome. The world will be listening.