AGI

AGI entered my life first as exactly the kind of thing I wanted proximity to: ambitious systems, agents, model infrastructure, interfaces between software and increasingly general capability. By the time I was working there I touched APIs, SDKs, partner-facing surfaces, iOS, on-device LLMs, model quantization, agents, control-plane problems, and owned api.agi.tech. [link/embed public work] This was real technical work in the domain I had spent years circling from the outside. There were moments where I felt extremely competent, where the distance between teenage me printing ML papers at McDonald's and adult me building actual AI infrastructure felt almost offensively short.

There were also organizational dynamics I either did not understand, did not respect, or did not model early enough. This is a recurring problem in my life: technical systems expose state; human organizations often expose narratives about state. Who thinks you're effective, whether priorities changed, whether somebody finds your communication expensive, whether the role they hired for is still the role they want—these variables can dominate the outcome while remaining weakly observable until they cash out as a decision.

On April 7, 2026, the decision cashed out. [verify exact chronology/wording] "You're not a good fit." That phrase hit me with way more force than the employment event alone explains. I had spent years trying to figure out where the fuck I fit: congregation, school, internet, startups, San Francisco, relationships, technical communities. And now here was another institution with a clean binary output: not here.

I was furious. I wanted to kill the guy who told me. I need to write that sentence exactly because sanitizing the anger into "I was upset" would be false, and I need to write the next sentence just as exactly: wanting that in the moment is not the same thing as forming an intention to do it, making a plan, preparing, or acting. I did none of those things. The internal state was violent rage and perceived injustice. The external behavior was not homicide. Those distinctions are not legalistic ass-covering; they are the actual structure of the experience.

It felt SO unfair. I had a model of my contribution, my capability, what I had built, what I could become there. Somebody else had a model with enough organizational authority to terminate mine. This was adult life violating the childhood prior again: you can try hard and be smart and care and still encounter another person's veto. There is no theorem guaranteeing they update toward your self-assessment. There is not even a theorem guaranteeing the better assessment wins.

I also felt humiliated by the phrase itself. "Not a good fit" is semantically soft and existentially brutal. It does not say "this function is wrong" or "this deliverable missed the deadline." It maps the failure onto compatibility between person and environment. Maybe that is why companies use it—it avoids litigating every local detail—but if you already have a lifetime of asking where you belong, the phrase has an almost comical amount of surface area.

[embed: public AGI work] [timeline: Jan–Apr 2026 role] [private artifact: termination notes, likely not public] [annotation: preserve what I knew then vs later interpretation]

Then the structure disappeared. No job. No daily proof that my technical ability had exchange value. No obvious reason I had moved to San Francisco beyond a future that had just been terminated. I had projects, ideas, people, research, but the reward circuitry did not care about the inventory. It cared that I felt useless.

I had been flirting with worthlessness as a philosophical problem for years. After AGI it became phenomenology.