What the AI co-pilots are
Beyond the Architect that designs a system, three AI co-pilots help you run one. All of them advise — none of them touches your money or executes a trade on its own.
- •AI Risk-Check — the risk math (liquidation price, effective leverage, stop distance) is computed deterministically first; the AI then reads those numbers and gives qualitative, comparable risk advice with an explicit score. It interprets the math — it does not compute it — so it cannot hallucinate a liquidation price.
- •AI Optimizer — reads a running system’s real trades and returns a rating (excellent → good → needs-work → critical), the single top issue, and a concrete suggested change. Improvement grounded in your actual behavior, not theory.
- •AI Recommendations & Insights — surfaces suggestions across your fleet so you notice what deserves a look.
How much to trust them — the guardrails
- It advises; you confirm. Nothing the AI says auto-executes on the money-path. You always approve the change.
- It interprets numbers, it does not invent them. The deterministic layer — liquidation math, sizing, the Risk Gate — is the source of truth; the AI narrates on top of it.
- It runs under your AI budget. A real cost cap, not an infinite oracle.
- Every narrative is structured — Observation → Reasoning → Risk → Recommendation → Next step — so you can check the logic, not just the conclusion.
The pitfall
Treating an AI suggestion as an order. A model can be generic or plain wrong; it is a second opinion to weigh, not an autopilot. And remember what actually gates live trading: the deterministic Risk Gate, never the AI. Read the reasoning, keep your own judgment, and let the math have the final word on risk.
Next
The AI Architect designs; the account Risk Engine and Risk Gate are the deterministic layer these co-pilots sit above; see what AI costs in Plans, AI packs & budget.