Letter

2

Agentic Humans

By

Navid Nathoo

The technology industry is spending billions to make AI agentic, meaning systems that can plan, act on their own, and figure things out without being told what to do at each step. Meanwhile the education system spends sixteen years training humans to do the opposite. Follow the syllabus. Complete the assignment as specified. Choose from the four answers provided. Wait to be told what comes next.




This mattered less when the economy ran on instruction-following. If most jobs consisted of executing well-specified tasks, training people to execute well-specified tasks was reasonable. But AI just made executing well-specified tasks nearly free. So the economic value of a human is migrating to what happens before a specification exists. Someone has to notice the problem, decide it's worth solving, and find a path when nobody has marked one. That's agency, and school suppresses exactly the trait the economy now rewards most.




Can you train agency? I think so. It may be partly inborn, but there are clearly ways to magnify it, and there are clearly ways to suppress it, because we've built an entire system that does. Put a person in an environment where every problem is pre-defined and every answer is pre-determined, and their agency atrophies. Put the same person in front of real problems with real stakes and no answer key, and it grows. People aren't born passive. The system trains them into it, one multiple choice test at a time.




Who you're surrounded by matters as much as what you work on. If you spend your days among people optimizing for grades, you'll absorb the assumption that work means clearing checkpoints someone else set. If you're surrounded by people building things, you'll absorb the opposite assumption, and it will change what you attempt.




No institution exists whose explicit job is developing agency. We have institutions for transferring information, institutions for credentialing, and institutions for sorting people. So we built one. At Zero, people work on real problems with real ambiguity, where they decide what matters, act without permission, and live with the results. Agency comes first, because everything else can be learned by someone who has it, and very little can be learned in any way that matters by someone who doesn't.




The world is building agentic AI. We're building agentic humans. Both are coming. Only one currently has an industry behind it.




Navid Nathoo

Founder, Zero