🔖 Features
🧠 Intelligent Intent Classification — Dynamically routes queries to the most suitable Agent based on context and content.🔤 Bilingual Support — Fully supports both Python and TypeScript.🌊 Flexible Agent Responses — Supports both streaming and non-streaming responses from different Agents.📚 Context Management — Maintains and utilizes session context across multiple Agents for coherent interactions.🔧 Scalable Architecture — Easily integrate new Agents or customize existing Agents to fit specific needs.🌐 Universal Deployment — Can run anywhere — from AWS Lambda to local environments or any cloud platform.📦 Pre-built Agents and Classifiers — Provides a variety of ready-to-use Agents and multiple classifier implementations.
❓What is Multi-Agent Orchestrator?
Multi-Agent Orchestrator is a flexible framework for managing multiple AI Agents and handling complex conversations. It intelligently routes queries and maintains context during interactions.
The system provides pre-built components for quick deployment while allowing easy integration of custom Agents and session message storage solutions.
This adaptability makes it suitable for a wide range of applications, from simple chatbots to complex AI systems, meeting diverse needs and efficiently scaling.
🏗️ Advanced Architecture Flowchart
-
The process starts with user input, which is analyzed by the classifier.
-
The classifier uses the features of the Agents and the conversation history of the Agent to select the most suitable Agent for the task.
-
Once an Agent is chosen, it processes the user input.
-
Then, it saves the conversation, updates the Agent‘s conversation history, and returns the response to the user.
SupervisorAgent: Agent Coordination
Multi-Agent Orchestrator now includes a powerful new component —SupervisorAgent, which enables complex team coordination among multiple specialized Agents. This new component implements an “Agent as Tools” architecture, allowing a primary Agent to coordinate a group of specialized Agents working in parallel, maintaining context consistency and providing coherent responses.
Main Features
🤝 Team Coordination – Coordinates multiple specialized Agents to collaboratively handle complex tasks.⚡ Parallel Processing – Executes multiple Agent queries simultaneously.🧠 Intelligent Context Management – Maintains conversation history across all team members.🔄 Dynamic Task Delegation – Smartly assigns sub-tasks to the appropriate team members.🤖 Agent Compatibility – Supports all types of Agents (e.g., Bedrock, Anthropic, Lex, etc.).
How to Use SupervisorAgent
SupervisorAgent can be used in two powerful ways:
-
Direct Use – Directly call it when specialized team coordination is needed for a specific task.
-
Classifier Integration – Add it as an Agent to the classifier, building a complex hierarchical system with multiple specialized teams.
Example Use Cases
-
Customer Support Teams, with specialized sub-teams.
-
AI Film Production Studios
-
Travel Planning Services
-
Product Development Teams
-
Healthcare Coordination Systems
Demo APP
The following video showcases an extended version of a demo application using 6 specialized Agents:
-
Travel Agent: Powered by Amazon Lex Bot.
-
Weather Agent: Utilizes Bedrock LLM Agent and queries the open-meteo API using tools.
-
Restaurant Agent: Implemented as an Amazon Bedrock Agent.
-
Math Agent: Utilizes Bedrock LLM Agent, equipped with two tools for performing mathematical operations.
-
Technical Agent: A Bedrock LLM Agent designed to answer technical questions.
-
Health Agent: A Bedrock LLM Agent focused on handling health-related queries.
Watch how the system seamlessly switches between different topics, from booking flights to checking the weather, solving math problems, and providing health information. Note that each query can select the appropriate Agent, ensuring consistency even with brief follow-up inputs.
This demo highlights the system’s ability to handle complex, multi-turn conversations while maintaining context consistency and leveraging specialized Agents across multiple domains.
GitHub:
https://github.com/awslabs/multi-agent-orchestrator
