AutoGen is a Microsoft framework for multi-agent AI applications, and its own repository now says it is in maintenance mode. Kortix is the open-source AI Operating System: your agents, skills, memory, and connectors live in one git repo you own, and each session runs on its own cloud computer.
AutoGen's cells state what its own repository documents, and the Kortix column quotes Kortix.
| Criteria | What it is | Open source | What you own | Runs the agents or defines them | Self-host | Model choice | Human review of changes | Project status |
|---|---|---|---|---|---|---|---|---|
| Kortix | The open-source AI Operating System: agents, skills, memory, connectors in one repo | Yes | Agents, skills, memory, and connectors in one git repo you own | Runs them: one isolated sandbox per session, on its own branch | Self-host is free; or Kortix Cloud, your VPC, on-prem | Any model provider with your own keys | Session work reaches main through a change request a person approves | Actively developed |
| Microsoft AutoGen | A framework for multi-agent AI applications | Yes | The application code you write, run in your own environment | Defines them: you write the orchestration in Python and run it | Yes, you run it in your own Python environment | OpenAI, Azure OpenAI, and other clients through Extensions | No built-in review gate; you wire it into your app | Maintenance mode: no new features, community-managed |
AutoGen's cells come from Microsoft's own repository, checked October 2026. Product status changes, so verify it before you commit.
# kortix.yamlagents: support-triage: file: agents/support-triage.md connectors: allAgents, what each may reach, and how the work is governed are entries in one file, versioned with the project.
AutoGen is a framework for creating multi-agent AI applications that can run on their own or work alongside people. Microsoft built it in its research group and gave it a layered design: a Core API for message passing and event-driven agents, an AgentChat API for common patterns such as two-agent chat and group chat, and Extensions that add model clients and code execution (the AutoGen repository). Its README now carries a maintenance notice: AutoGen receives no new features or enhancements and is community managed, and Microsoft steers new users to a different framework (the AutoGen README, checked October 2026). A team that chooses AutoGen today builds on a library whose feature work has stopped.
AutoGen is a library. You install it with pip, write Python, and run it in the environment you deploy. It defines agents and the conversations between them, and everything around that definition stays yours to run. Kortix is the leading open-source alternative to Claude Cowork and ChatGPT Work, and it sits one layer up: an open-source AI Operating System where the agents, their skills, the company memory, and every connector live in one git repo you own (Kortix on GitHub). Each session then runs on its own isolated cloud computer with the repo and the tools already on it.
On AutoGen, a team owns the code it writes and the environment it runs. The prompts, the tool wiring, the retries, the logs, the deployment pipeline, and the machine the process sits on are all the team's to build and to keep running. On Kortix, the company owns the configuration: the agents, the skills, the memory, and the connectors are files in one repo, and connector credentials are brokered server-side and never enter the machine. Session work reaches main through a change request a person approves.
A framework answers how agents talk to each other. A platform answers where the agents run, what they may reach, who approves their changes, and how the work reaches the codebase. Teams move to a platform when the library stops being the hard part, and they need a preview per change, scoped reach into 3,000+ apps, per-resource permissions, and an approval gate before anything merges. A framework still suits a team that wants to embed agents inside an application it operates itself. LangGraph is a framework still under active development, and the Kortix vs LangGraph page compares it with Kortix directly. The Kortix blog covers the same comparison.
Kortix replaces the runtime layer, not the idea of agents. AutoGen defines how agents converse inside your application; Kortix runs agents on isolated cloud computers, keeps the configuration in a repo you own, and lands changes through a review. A team that only needs a Python library for multi-agent conversations can stay on AutoGen.
AutoGen is a general Python library that embeds inside any application you build, and its job is to express multi-agent patterns in code. Kortix runs agents as finished work on its own machines rather than as a library inside your process. When your product is the application and the agents live inside it, AutoGen fits that shape.
Yes. Run it on Kortix Cloud, in your VPC, or on your own on-prem network, and self-host is free. The first run asks six things and no model key. Read the docs for the steps.
Yes. Kortix is open source: read it, fork it, and audit it at Kortix on GitHub. Your agents, skills, memory, and connectors are files you can read, and you can self-host the whole stack.
Build the company as a repo, then let the agents work.