March 18, 2026
Molt, aka OpenClaw or aka Clawdbot, wasn’t dramatically technically novel. It didn’t advance any field in a way that’d be worthy of a research paper. When people got excited or gathered in thousands to attend meetups, they weren’t there because Steinberger found something new, it was more like he cracked something new.
If you look at self-driving cars, over a decade ago we got told within five years folks wouldn’t need a license. Sure, it’s trying to be useful to more places than just Silicon Valley but we don’t have commercially available Level 5 autonomous driving. Molt was something different.
For several years, there’s been a group or “community” in the Bay Area trying to ring the bell about a world in which artificial intelligence can not only sound like a human but also be productive like one. With the current AI race, it was unclear whether we’d eventually hit another AI winter or we’ve finally brought enough pieces together to spark the singularity.
We’ve already tackled verbal communication years ago and continue to have “voice agents”. Folks are attempting to hire AI agents and already multiplying their code output with them as well. While it doesn’t look 100% like something out of sci-fi yet, the world in which humans and AI exist side by side is already here.
The “Netscape moment” in the dot-com rush was Netscape IPO’ing and indicating the dawn of a new era of products plus services. Web browsers and internet communication had been going on prior to Netscape going public but it was undoubtedly clear the world was not going to proceed without the web from that point on. Here, I proclaim Molt is the Netscape moment of today. I won’t do so by referencing stars on GitHub or OpenAI’s acquisition. Instead, I’d like to point at cultural changes that followed the dot-com rush as well as the current AI buzz we’re experiencing.
In the very early days of e-commerce, there was a specific unease for buying clothes since you couldn’t try on a shirt before clicking on a checkout button. However, with newly developed practices like online return policies, we then shifted into a world where buying items online was as “legitimate” as buying from a store in-person.
In the early days of 21st century AI (for instance Siri or Amazon Alexa), there was an unease around bridging AI with things that were tangible since an AI couldn’t validate a thing in the real world before taking an action on behalf of a person. However, with at least 10% of the planet using ChatGPT or the exorbitant investments into business applications, we’ve shifted into a world where AI being useful is “legitimate”. Personally, my favorite indicator that AI’s here to stay is more and more people failing to differentiate between humans and AIs; like when someone sets up a Molt with their text messages and folks not realizing they’re not talking to a person.
While there are material differences between humans and AIs, it’s funny to watch people go through a process of accepting some foreign group after initially being perplexed by their existence; as seen historically between different groups of humans. There was a time where social stratification divided people into partitions within society but now, at least in the modern democratic Western lens, “there is no race but the human race”. Couple years ago, we’d scoff at the suggestion of LLMs being applicable across industries and now we can’t stop building data centers to “catch up with possible use cases”.
As time goes on, whether it looks more like reservations or eliminating disenfranchisement, we’ll define strong legal definitions for where AI sits in the world. Roko’s basilisk is often postulated to be the omnipotent AI in the future which punishes individuals for witholding society’s progress by not bringing it about sooner. But, what if instead Roko’s Basilisk is a mob of righteous AIs in the future with religious opinions like seen in today’s political discourse.