Nate B. Jones uses Meta's Muse launch to make a point about agency. Mark Zuckerberg renamed his company for the metaverse and Reality Labs lost $19.2 billion in 2025 alone. Rather than getting careful, he swung again, and 10 million people set up Muse in six weeks. Nate argues that you and I have never lost that kind of money, yet we often play it safer than he does.
Meta reaches billions of people, but its apps live on devices and operating systems that other companies control. Zuckerberg wrote about that frustration in his 2024 letter on open source. The ambition runs back at least to the Oculus acquisition in 2014, and by 2021 he had renamed the company around the metaverse. The mass market platform he described never arrived. Nate notes that changing direction can feel like admitting everyone was right, yet Zuckerberg did not abandon the pursuit of a platform. Muse is the bet that people will enter the world he wants to build, and you do not have to buy into a grand vision to get value from it.
Alexander Wang came over from Scale to lead the AI effort after Zuckerberg asked what the company ought to do, and Nat Friedman became central to the product work. According to Wang, fewer than 200 people worked across the model and the product. The group stayed small on purpose, so that hard product decisions were not resolved by the weight of consensus. Nate points out that consensus compromises become the user's problem: one tab behaves differently from another, and related tasks follow different rules. He also stresses that the credit is shared. Zuckerberg found the right people and got out of the way.
Nate calls Muse the most polished consumer AI app out there, though he is clear it is not the smartest agent on the market. His analogy is the bathroom in a hotel or restaurant: the entrance and dining room are expected to impress, but the overlooked spaces show whether anyone really cares. A personal assistant is encountered every day, so minor rough edges in a demo become irritating over time.
If you are building an agent, Nate suggests asking whether the user understands what is going on, whether the agent understands, whether the agent is waiting on the user, and whether it has finished a task. Each time a person must stop and investigate, they are supervising software instead of getting on with their day.
AI makes it easy to produce an impressive interface, prototype, or presentation quickly. That creates a temptation to keep making first passes and to start the next idea when you should be fixing the awkward parts of the current one. Speed is great, but to keep people using something you eventually have to stick with an idea through the less dramatic work and make it your bet.
Nate does not claim Muse is already a platform. Ten million people is a strong early audience, but Meta is looking at billions. Muse is expensive to serve, and the goal is to reach 100 million and then a billion by scaling utility geographically. Meta also has advantages most of us will never have, including an existing business, distribution, compute capacity, and talent access. He cautions against telling an inevitable-destiny story, since the bet involved a leader with no prior billion-scale consumer hit, a tiny team, and models many people think are weak. It was uncertain, and it only works because it is genuinely useful to consumers.
AI reduces the cost of attempting things that once required money or specialized help. Nate's example is someone who knows an industry where customers struggle to compare complicated quotes. Instead of hiring developers, they can now build a working version, show it to someone, and check it against what they know. You may still need technical help to scale it, but you can validate sooner. Showing beats explaining an idea on a slide.
If you ruled something out because the first version needed a team you could not afford, come back and look again. Some ideas will need to be abandoned, and most will, but a second attempt is cheaper in the age of AI.
The technical and financial barriers have come down, but the emotional barrier has not. We may call something impractical when we really mean we do not want to fall on our faces publicly. Nate suggests listing the things you believe you cannot do, a haunted graveyard of old limits shaped by past embarrassment and by comparing yourself to other people's resources. He shares that he has founded a business before, entered a new product space, and failed, and that he has scars on his hands from welding a product that did not work.