🧠 Tessen Aurix

Trader memory and improvement, not just trade history.

Most journals record what happened. Aurix looks at how you actually behave — when you intervene, when live results stop matching the backtest, and whether your own patterns are helping or hurting.

Post-mortems

Every failed grade gets a plain-language diagnosis of exactly which gate broke and what to try — not just a red X. Deterministic, computed from the report itself, no LLM guessing.

Grade-drift monitor

Live results checked against the backtest's promise, continuously. An alert fires the moment live expectancy meaningfully diverges from what the Grade said to expect.

Intervention tracking

Every manual kill-switch is logged and distinguished from an automatic risk halt. Redeploying the same agent shortly after stopping it gets flagged — a pattern worth noticing, not a judgment.

Time-of-day patterns

Win rate and average P&L broken down by hour, across every closed trade — so you can actually see if your own worst hours are dragging on results.

Agent journal

Free-text notes attached to any agent — a note to your future self about why you made a change and what you expected.

Honest by design

Rapid-redeploy flags and drift alerts are informational — Aurix never auto-blocks a decision or auto-resumes a halted agent. An auto-un-halting bot is worse than a halted one; the point is to show you the pattern, not to override your judgment.

Aurix is live in closed beta today.