老師Roushi

Use cases

Three operators.

Roushi is built for one shape of customer, with two natural neighbors.

01

If you run a portfolio

Operators with five or more products and one to three operators total.

Decisions made in product A get forgotten when you work on product B. The Drizzle gotcha that bit you in March doesn't surface when the same trap appears in repo C in May. Marketing copy drifts the moment code ships. Roushi is the brain you wished you had at product number two — every decision, lesson, and pattern gets a slug, a citation, and a place in the typed graph. When you open a new product workspace, the relevant rules and lessons already greet you in the SessionStart context.

The marketing site IS Roushi practising what it preaches.
02

If you're a solo founder

One operator, one or two products, agent-paired work.

The business context lives in your head. Slack is where decisions go to die. Notion is a passive store — you have to remember to query it. Roushi captures the durable bits (the why, the lessons learned the hard way, the trade-offs you made on purpose) and re-surfaces them in the agent you're already pair-programming with. The brain isn't a separate destination; it's a layer below Claude Code that keeps your context across sessions.

Notion is a passive store. Roushi acts.
03

If you ship code with AI agents

Heavy Claude Code / Cursor / Windsurf users; teams piloting agent-first workflows.

Every agent session starts cold. The model has the codebase but not the why. Roushi is the persistent memory layer: PreToolUse hooks inject relevant brain context before each Edit/Write, the SessionStart hook loads applicable rules, and remote agents can query the same brain over HTTPS MCP. The result: a continuity across sessions and across agents that no single agent provides on its own.

Memory across agents and sessions.

Recognize yourself in one of these? Get in touch.