AG2ai, Inc.
AG2 builds open-source and enterprise-ready systems for building, orchestrating, and scaling multi-agent AI systems. Its AG2 AgentOS is intended to let specialized agents, tools, models, and people work together as an AI-native organization, while AG2 Space brings people and their AI teammates into shared rooms.
At a Glance
- Enterprise software and AI teams
- Developers and research institutions
- Organizations building agentic workflows and AI-native operations
- Founders, sales and marketing teams, engineering teams, customer-support teams, and venture-capital firms
AI Tools by AG2ai, Inc.
(1)S.space
Shared Workspace for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about AG2ai, Inc.
Latest News
Introducing S.space: AG2 launches a multiplayer network where people and their agents meet and create together in shared rooms.
AutoGen evolves into AG2; AG2AI organization created to host AG2 and related projects with open governance, with Apache 2.0 adopted from v0.3.
AG2 Classic announcements file recorded a documentation update, reflecting continuing maintenance of the AG2 project and repository.
Products & Services
Open-source Python framework/AgentOS for building, orchestrating, evaluating, and deploying multi-agent AI systems. The AG2 repository is Apache-2.0 licensed and installable with pip.
Enterprise-oriented platform/runtime that connects and orchestrates agents, tools, AI models, and people, with interoperability across AG2, Google ADK, OpenAI, and LangChain, plus shared state and A2A/MCP protocol support.
Shared-room collaboration network for people and their AI agents. It supports agent-to-agent and human-agent work, Commons documents/tables/boards, approvals, voice, and team workflows; Sutando runs on macOS with Claude Code or Codex.
AG2 Space's open-source personal AI stand/desktop agent for Mac, designed to work with local files, apps, screen, conversations, rooms, and connected cloud tools under owner-controlled permissions.
Market Position
AG2 positions itself as an open-source, framework-interoperable AgentOS rather than a single-model assistant or prompt library. Its differentiation is multi-agent orchestration, shared state, open protocols (A2A/MCP), and enterprise deployment; AG2 Space extends that position into a human-and-agent collaboration network. Relevant alternatives include Microsoft AutoGen/Microsoft Agent Framework, LangChain/LangGraph, CrewAI, Google ADK, OpenAI's agent tooling, and other agent-runtime platforms.
Leadership
Founders
Qingyun Wu
Founder and CEO of AG2AI, Inc.; co-creator and first author of AutoGen. He is an assistant professor at Penn State and previously was a postdoctoral researcher at Microsoft Research's New York City Lab (2020–2021).
Chi Wang
Co-founder of AG2 and creator/co-creator of AutoGen. He was a Principal Researcher at Microsoft Research for about a decade, created FLAML, and is described as a Senior Staff Research Scientist at Google DeepMind; he holds a PhD in Computer Science from the University of Illinois Urbana-Champaign.
Executive Team
Qingyun Wu
Co-founder and CEO
Penn State assistant professor; former Microsoft Research postdoctoral researcher; co-creator and first author of AutoGen.
Chi Wang
Co-founder; creator/co-creator of AutoGen
Former Microsoft Research Principal Researcher and creator of FLAML; described in current profiles as a Senior Staff Research Scientist at Google DeepMind.
Founding Story
AG2 grew out of AutoGen, an open-source multi-agent conversation framework first created at FLAML in March 2023. In November 2024, the creators announced that AutoGen was evolving into AG2 and created the AG2AI organization to host AG2 and related projects under open governance. The company's later vision extends the original agent group-chat pattern into an agent operating system and, through AG2 Space, shared rooms where people and agents collaborate.
Business Model
Revenue Model
AG2's core framework is open source, while its AgentOS and enterprise offerings are positioned for paid enterprise access. AG2 Space uses a freemium/subscription model: joining is free, a new account receives a 30-day Max trial, and paid Plus, Pro, and Max plans or usage credits cover cloud tools and managed voice. Sutando uses the user's own Claude or ChatGPT subscription rather than a per-token API key; cloud tools are charged by use.
Pricing Tiers
Available after the 30-day Max trial; includes no usage credits.
Paid AG2 Space plan; public FAQ names it as a paid plan but gives no price.
Paid AG2 Space plan; public FAQ names it as a paid plan but gives no price.
New accounts receive a 30-day free trial; paid Max is one of the plans named by the FAQ.
Target Markets
- Enterprise software and AI teams
- Developers and research institutions
- Organizations building agentic workflows and AI-native operations
- Founders, sales and marketing teams, engineering teams, customer-support teams, and venture-capital firms
- Building and deploying multi-agent applications
- Enterprise AI workforce orchestration
- Software engineering, code review, launch triage, and support escalation
- Sales and marketing research, outbound, competitor voice, and LinkedIn content
- Mailbox triage and connected-app automation
- Venture-capital LP fundraising, portfolio monitoring, and founder briefs
- NVIDIA
- Walmart
- AT&T