About

I build AI systems, and I allocate capital. They have never felt like separate kinds of work. Both come down to the same question: who is allowed to decide, and who answers for the result.

Now

I am building and testing a personal AI system around durable private context, actions that stay inside explicit authority, and work that can move between a person's own machines and shared compute. It is not public yet. I publish the parts that hold up. The question underneath them is what happens when an AI system moves from producing information to exercising delegated authority: what it may see, what it may decide, what it may do, and how that authority can be withdrawn.

What I believe

• Policy should not be probabilistic. A model can recommend an action. The rule that permits it should be code someone can read, version, and test.
• Autonomy should scale with reversibility. The useful question is not whether an agent is autonomous, but how much autonomy a particular action can bear.
• An audit log is not a control. A record written after an action is evidence. It cannot stop the action.
• Novelty belongs to models. Repetition belongs to software. Use a model to find a path the first time. Once the path repeats, call it.
• A person's thinking should live near the person. Private context is the asset. Where it is allowed to travel is the design.
• Analysis is becoming cheap. Commitment is not. The person who authorizes a decision still carries the consequence after the analysis is done.

Where this comes from

Before I worked in technology, I ran my family's commercial real estate business in Kazakhstan. My father and brother run it today, and I still serve on its board. That is where I learned what a decision costs after the analysis is finished, and it is why I have since invested in dozens of real estate projects across Texas.

For several years I also ran a small quantitative fund, with a team and outside investors. I enjoyed the research and the engineering more than the administration and compliance, so I closed it to outside money; it now manages only my family's capital. I was an early investor in MetaVi Labs, a life sciences company, and remain one.

I mention this because it explains the writing. When a system acts, someone still owns the outcome. I have been that person, and I have also been on the receiving end of governance. Both make me want it to be something you can read and run, not something you wait on.

How the writing fits together

Four threads run through the essays and shorter notes. Delegated authority: what changes when a system moves from producing information to taking action. Personal AI: what should remain near the person. Distributed governance: where policy attaches when computation crosses machines nobody fully controls. Ownership and incentives: who decides, who carries the risk, and who remains responsible. They are one question seen from four sides.

I write here when I have something definite enough to defend. I live in Texas.

Email is the way to reach me:dastan.aitzhanov@gmail.com. Code I have published lives on GitHub, and selected public work is collected on the Work page.