AI crypto trading,
organised like a trading company.

Fifteen roles, one trading company

Each agent has one job, like a desk in a real trading firm. The CIO sets direction, analysts and scouts bring ideas, the planner turns them into trade plans, risk and compliance check them, and the execution trader places the orders you allow.

• CIO, News, Surveillance, Scout and Planner
• Risk, Execution, Positions and Performance
• Compliance, Strategy R&D and Backtests
• DevOps, Finance and Operations

Autonomy behind hard caps

AI agents can be wrong with great confidence. That is why autonomous live mode only arms behind hard caps that the agents cannot change: a budget, a maximum position size, maximum leverage, a daily loss limit and a drawdown kill switch. Until you arm it, the agents research and plan without trading.

Apex Intelligence and the Personal Copilot

Not every AI task needs a trading team. Apex Intelligence is private AI research with evidence, charts and data tools for the exchanges you trade on, included in every plan. The Personal Copilot answers questions about your own Apex activity with linked evidence and proposes actions for you to review.

Neural Grid: AI decision models on your own GPU

Neural Grid runs 25 neural strategies and pairs decision models on your own local GPU with account-private paper trading and gated live execution. AI nodes can also use a local GPU or your own model key inside strategies. On the Starter plan, Forecast Lab adds GPU-trained volatility and direction forecasts, measured against a simple EWMA baseline so you can see whether they actually add skill.

AI that shows its work

Every AI feature in Apex OS is built to be checked: research answers come with evidence, agent companies are ranked on profit and loss per Grok dollar, and models trade on paper before they get live capital. You see what the AI did and what it cost, not just what it claims.