Step through three AI systems and see how they make decisions
Sample data. No real agents. Nothing touches your code.
Start with AtomizeAtomize
How a search system finds the source text behind an answer.
You will Run one sample sentence through the first stages of Atomize: cut it into pieces and search them. See where Atomize re-ranks the matches and has a model write the answer, then change one setting and see whether the answer still traces back to its source.
Covers: splitting text into pieces, searching by meaning and words, re-ranking, writing an answer from sources
Agent Rally Point
How coding agents split work in one repository without overwriting each other.
You will Guess what happens at each step as two agents claim files, hit a refused claim and hand work to each other. Then see how Rally coordinates more agents, each in its own git worktree.
Covers: who is working on which file, a shared log, handing off work, separate worktrees
Build Loop
How Build Loop plans a new feature, splits the work, builds the parts in parallel, commits each batch and reviews the result before calling it done.
You will Decide how Build Loop should split a new feature, saved searches for a news app. Then watch it build the parts in parallel, commit them, catch a problem in review and fix it.
Covers: planning, splitting work by file, parallel builders, who commits, a seven-step review
These are interactive explanations. They do not run agents or change your projects. Simulation timings and sample outcomes are not production benchmarks.