🎙 Shift left Arch: Drawing the boundaries

We're not doing a surface-level "what is Kafka" episode. We're walking the full path from a raw event to a trustworthy real-time decision, one layer at a time:
→ Where does processing belong — the lakehouse, or shifted left into the stream?
→ What do you detect, and with which tool — CEP vs. ML vs. LLMs?
→ How does that signal reach the system that acts — MCP, REST, or Kafka?
→ How much autonomy do you hand to an AI agent, and how do you keep it trustworthy?