Building a 24/7 crypto trading platform with multiple AI agents, orchestrated by Temporal and using the Model Context Protocol (MCP) for tool interfaces, proved to be both challenging and enlightening. The system revolves around three core AI agents: a broker agent managing intents, an execution agent for trading decisions, and a judge agent for performance evaluation and strategy updates. Inspired by concepts from AI conferences, the project integrates proactive AI and ambient intelligence, allowing agents to operate continuously without waiting for explicit prompts. Temporal's features, such as Schedules, Signals & Queries, and its UI, provided critical components for the orchestration of these agents, enabling proactive actions, seamless communication, and real-time observability. The use of Temporal allowed for reliable, durable, and extendable operations, including providing audit trails and fault tolerance, which are vital in financial systems. The architecture emphasizes extendability and team-friendly design, allowing the addition of new agents or tools through modular workflows. This innovative approach offers a robust framework for complex AI systems, focusing on high-level agent logic while Temporal manages timing, communication, and system reliability.