Building the smallest version that can teach you something

A practical way to separate an experiment from a disguised production system.

By Aki1 min readViews —

A useful first version does not prove that the final product will work. It reduces one important uncertainty with the least machinery possible.

Start with the question

Before choosing a framework, write down the decision the experiment needs to inform. If the result cannot change what you build next, the experiment is only a demonstration.

Small systems are valuable because their failures are legible. You can see what you assumed, what the user did, and where those two stories diverged.

Related paths

  1. Field Notes2026

    Architechural Masterclass in Kimi K3

    A 2.8T parameter model is impressive!! But what interested me more was, how do you even run this thing? This is my small attempt at unpacking the architecture choices that make Kimi K3 actually survive in production.

    Read note
  3. Research2026

    Cache Poisoning

    Yo!! A quick note before we start. I have yapped a lot about prompt caching in Part 1 , how it works, what cache hits and misses actually mean, and how to improve cache hit rates without making your API bill shoot the sky.

    Read note