Where the knowledge is buried deepest.

Decades of experiments, results and hard-won lessons, findable again, so teams build on what was tried instead of repeating it.

Catalog rules, policy decisions and the reasoning behind them, in one index that reviewers and models can both use.

Architecture decisions and customer commitments, recovered from the threads, memos and tickets they were made in.

Audit evidence, runbooks and the people who know them, plus research on how AI agents fail near systems you cannot afford to lose.