The plan is not the work
Plans are useful when they create movement. They become dangerous when they turn into theater: too much control on paper, too little contact with reality.
v2 starts as a public laboratory because I want decisions to stay close to the place where they hurt. AI, quant, and automation look simple on the board. Learning happens when data, models, markets, costs, and users enter the same room.
The best plan does not try to predict everything. It leaves enough memory behind to change without losing context.
What goes into the laboratory
- Experiments with agents, LLMs, and research automation.
- Quantitative studies with data, hypotheses, backtests, and clear limits.
- Small interfaces, often in Flutter, when an idea needs to become a product.
- Honest reviews of theses that looked better before meeting cost, latency, market, or user.
How progress is measured
Progress here is not post volume. It is accumulated clarity. A good record should explain the hypothesis, the data, the constraint, the decision, and what changed after the system met the world.
input: hypothesis + data + constraint
process: model + automation + test + friction
output: better system and public memory
Next steps
The first commitment of v2 is simple: publish what is being built, preserve context, and let the brand become clearer through the work.