Amazon just rebranded and updated its AI Agent framework Here's everything you need to know about Agent Squad... Recently, the AWS team rebranded their Multi-agent orchestrator framework into Agent Squad and introduced a few notable updates. Let's first take an overview of how the framework works, and then we will dive into the updates. 📌 Here's what the framework workflow looks like: 1. Input - User input triggers the pipeline. The Orchestrator captures intent and session metadata, starting the routing process. 2. Classification (Global Context) - A Classifier LLM evaluates the full conversation history and agent profiles to determine the best match for the user’s request. 3. Fetch Agent Characteristics - Classifier pulls agent details—LLM models, tools, and properties—to ensure smart, context-aware routing beyond keywords. 4. Agent Selection and Last Conversation Data - The agent is selected, and a specific conversation history is retrieved, keeping memory scoped and isolated for clarity and relevance. 5. Agent Processing (Local Memory + Tools + LLM) - Agent uses short-term memory, optional long-term retrieval, its assigned LLM, and tools to generate a comprehensive response. 6. Output - Final output is sent back to the user via the Orchestrator, supporting both streaming and synchronous delivery for smooth interaction. 7. Conversation Storage - Responses are stored in memory (Short-term) or persistent memory (Long-term), with a framework allowing you to choose between any database you like. 8. Global History Update for Follow-Ups - Conversation updates feed back into the global context, enabling coherent follow-ups and maintaining continuity across turns. 📌 New Updates: 1. New Supervisor agent enables dynamic coordination of agents using an "agent-as-tools" approach, ensuring context-aware and coherent responses. 2. Improved intent for query routing, ensuring the most relevant agent handles each request. 3. Full implementation in both Python and TypeScript. 4. Availability of ready-to-use agents and multiple classifier implementations. 5. Structured execution logs, sub-step tracking, and Amazon CloudWatch integration for improved monitoring and troubleshooting. 6. Increased limits for agent collaborators and execution steps, supporting larger and more complex workflows. Learn more about the framework from the link in the comments. If you are a business leader, we've developed frameworks that cut through the hype, including our five-level Agentic AI Progression Framework to evaluate any agent's capabilities in my latest book. 🔗 Book info: https://amzn.to/4irx6nI Save 💾 ➞ React 👍 ➞ Share ♻️ & follow for everything related to AI Agents | 35 comments on LinkedIn