Submit a Tool

AutoGen

by Microsoft · 2023
No reviews yet
Freemium · AutoGen is free and open source under the MIT licence. Additional costs may apply for external AI model APIs, cloud hosting, infrastructure and other connected services. Free plan
Visit Website ↗ Docs
AutoGen

AutoGen is an open-source programming framework developed by Microsoft for building AI agents and multi-agent applications. It enables developers to create systems in which multiple AI agents can communicate, collaborate, use tools, exchange information and work together to complete tasks. Rather than relying on a single AI assistant to handle every stage of a problem, developers can assign different responsibilities to specialised agents and coordinate their activities through defined workflows.

The framework was designed to support both autonomous agent operations and human-in-the-loop collaboration. Developers can build applications where agents work independently, consult one another, request human input or pass tasks between specialised components. This makes AutoGen suitable for experimenting with complex AI workflows, research systems, business-process automation and intelligent applications that require several steps to reach a result.

AutoGen 0.4 introduced a redesigned architecture built around asynchronous messaging and event-driven communication. This allows agents to exchange messages without depending entirely on a simple, sequential conversation. The architecture supports both request-and-response interactions and event-based workflows, giving developers more flexibility when designing systems that need to respond to changing conditions or coordinate several agents at once.

AutoGen is organised into three major framework components: Core, AgentChat and Extensions. AutoGen Core provides the underlying infrastructure for developing event-driven and distributed agent systems. It uses an actor-based approach to support message passing, agent runtimes and communication between components. Developers can use Core to build more customised systems, including applications that need distributed execution or coordination across different environments.

AgentChat provides a higher-level programming interface for creating conversational AI applications. It supports single-agent and multi-agent systems, including common collaboration patterns such as group chats. Developers can use prebuilt agents and team configurations to create applications without implementing every communication mechanism from the ground up. AgentChat is particularly useful for prototyping and building task-oriented applications in which agents collaborate through conversations.

Extensions connects AutoGen’s core components with external services, libraries and tools. These extensions provide model clients, code-execution capabilities, agent implementations and other features that developers can use to expand their applications. The framework includes integrations for services such as OpenAI and Azure OpenAI, along with components for connecting to MCP servers and running code through Docker-based execution environments.

AutoGen also includes developer tools designed to make experimentation and testing easier. AutoGen Studio provides a web-based graphical interface for prototyping agent workflows without writing all the application code manually. It allows users to configure agents, experiment with teams and test multi-agent interactions. AutoGen Bench provides benchmarking capabilities for evaluating agent performance and comparing different approaches.

The framework can support applications such as research assistants, coding agents, document-processing workflows, business-process automation and systems that combine web browsing, code execution and file handling. Developers can also customise agents with their own tools, memory components, model clients and communication patterns.

An important consideration for anyone evaluating AutoGen today is its project status. Microsoft’s AutoGen repository has been archived and is no longer actively maintained. Microsoft states that it is not providing new features, bug fixes or security updates and recommends Microsoft Agent Framework as the successor for new projects. AutoGen remains available as an open-source framework for learning, experimentation and reference, but its archived status makes it a less suitable choice for new production applications, particularly those handling sensitive information.

Pricing

AutoGen is free and open source. The framework does not require a paid AutoGen subscription or licence fee to download, install or use its core components. Its code is released under the MIT licence, allowing developers to use and modify it within the terms of that licence.

The framework’s lack of a subscription fee does not mean every application built with it is free to operate. Developers may need to pay for the external AI models, API services, cloud infrastructure, hosting, databases or other resources used by their applications.

For example, connecting an AutoGen application to a commercial model provider may involve separate API charges based on the provider’s pricing and usage. Running distributed agents or deploying applications on cloud infrastructure can also introduce additional costs.

AutoGen Studio and AutoGen Bench are part of the open-source ecosystem and do not have separate AutoGen subscription prices listed in the official documentation. The total cost of an application therefore depends on the external services and infrastructure selected by its developer.

Review

AutoGen is a capable framework for developers interested in building AI agents that collaborate to complete complex tasks. Its strongest feature is its support for multi-agent communication, allowing developers to divide work among specialised agents instead of placing every responsibility on one assistant. This approach can be useful for research, coding experiments, workflow automation and applications that require several coordinated stages.

The separation into Core, AgentChat and Extensions also gives the framework a flexible structure. Beginners experimenting with conversational agents can start with AgentChat, while developers who need greater control over messaging, runtimes and distributed systems can work with Core. Extensions make it possible to connect the framework to external model providers, execution environments and other services.

AutoGen Studio is another useful component because it provides a graphical environment for prototyping agent workflows. AutoGen Bench also gives developers a way to evaluate and compare agent behaviour. Together, these tools make the framework valuable for experimentation and understanding how multi-agent systems operate.

However, AutoGen has limitations. It requires programming knowledge for many development tasks, and designing reliable multi-agent systems can involve considerable configuration, testing and debugging. Applications that rely on several agents may also use more model calls and computational resources than simpler AI applications. Costs from external model providers and hosting services can therefore increase as usage grows.

The most significant drawback is that Microsoft’s AutoGen repository is now archived and no longer actively maintained. Developers should not expect new features, regular fixes or ongoing security updates from Microsoft. AutoGen Studio is also intended for prototyping rather than production deployment.

AutoGen remains useful as an educational, research and reference framework, particularly for developers who want to understand agent orchestration and multi-agent architecture. However, its archived status should be taken seriously when choosing a framework for a new application. Microsoft recommends Microsoft Agent Framework for new projects that require actively supported development and production-oriented capabilities.

Key Features

  • Open-source AI agent development framework
  • Multi-agent collaboration and orchestration
  • Single-agent and multi-agent applications
  • Asynchronous agent messaging
  • Event-driven architecture
  • AgentChat programming interface
  • AutoGen Core runtime
  • Extensions for external services and tools
  • Group chat and conversational agent patterns
  • Custom agent development
  • Human-in-the-loop workflows
  • Local and distributed agent runtimes
  • Model-client integrations
  • Tool calling and code execution
  • Docker-based code execution
  • MCP server connectivity
  • AutoGen Studio visual prototyping interface
  • AutoGen Bench evaluation and benchmarking
  • Agent tracing, observability and debugging
  • Python and .NET interoperability

Pros & Cons

Pros

  • Free and open source under the MIT licence
  • Supports multi-agent collaboration
  • Flexible architecture for different agent workflows
  • Provides both high-level and low-level development interfaces
  • Includes a visual prototyping tool
  • Supports external model providers and tools
  • Offers asynchronous and event-driven communication
  • Supports Python and .NET interoperability
  • Useful for AI research, learning and experimentation

Cons

  • Project archived and no longer actively maintained by Microsoft
  • No ongoing Microsoft security or bug-fix updates
  • Not recommended by Microsoft for new production applications
  • Requires programming knowledge for advanced development
  • Multi-agent workflows can be complex to configure and debug
  • External model and infrastructure costs may apply
  • AutoGen Studio is not designed as a production-ready application
  • Community support and responses may be limited

Reviews

No reviews yet

No reviews yet. Be the first to review this tool.

Sign in to write a review

Related Tools

Daloopa

Daloopa, Inc.

Daloopa is an AI-powered financial data and research platform designed to help investment professionals access, extract, organise, and analyse company financial information.…

API Freemium
No reviews yet

Rogo AI

Rogo Technologies, Inc.

Rogo is an enterprise artificial intelligence platform purpose-built for the financial services industry. It is designed to function as an AI analyst…

API Subscription
No reviews yet

AlphaSense

AlphaSense, Inc.

AlphaSense is an AI-powered market intelligence and business research platform that helps organisations discover, analyse, and interpret information from financial, corporate, industry,…

API Subscription
No reviews yet