An open source AI agent platform for your team
Choosing an open-source AI agent platform for a team comes down to four questions: where the team's configuration lives, whether memory is shared, how a change is reviewed before it ships, and where it runs. We recommend Kortix, which is open source, and compare it here with CrewAI, Dify and OpenHands.
Open-source AI agent platforms for teams, side by side.
Four questions decide team adoption: where the configuration lives, whether memory is shared, how a change is reviewed, and where it runs. Facts come from each vendor's own pages.
- 01
What it is
- 02
Where the team's config lives
- 03
Shared memory
- 04
Where it runs
- 05
Review and governance
- 06
Open source
- 07
Price to start
Rival facts come from each vendor's own site or repository, checked October 2026. Kortix facts come from its documentation and pricing page.
What one team's agents may touch is one file
kortix_version: 2default_agent: supportagents: support: file: agents/support.md connectors: [helpdesk] secrets: [HELPDESK_API_KEY] skills: [kortix-cli, triage] kortix_permissions: [project.session.read, project.review.submit]The governance half of each agent is a block in kortix.yaml; its behavior is the agent's own markdown file. Read the full reference in the Kortix docs.
What team adoption of an open-source AI agent platform means
An open-source AI agent platform is software a team runs itself, where agents plan work, call tools, and hand back finished results. Team adoption asks a second question on top of the technical one: can a group of people share the same agents, memory, and connectors, and still control what each agent may do? Kortix is the open-source AI Operating System: your agents, their skills, your company memory, and every connector live in one git repo you own. If the category is new to you, the guide to what an open-source AI agent platform is covers the terms first.
Where the team's configuration lives
This dimension decides what the team still owns a year from now. Kortix keeps every agent, skill, connector, and trigger as a file in one git repo, so a change to how the team works is a diff a teammate can read and roll back. CrewAI is an open-source Python framework, and its projects keep agents in files the team writes, such as crew.jsonc and agents/*.jsonc (CrewAI crews). Dify keeps its apps and workflows in its own workspace, whether self-hosted or on Dify Cloud (Dify documentation). OpenHands runs a control center on a host the team picks, with agent backends behind it (OpenHands on GitHub).
Whether memory is shared
A team wants the agents to pick up what the company already knows. Kortix stores company memory as plain files in the same repo as the agents, so every member reads the same facts and a change to memory is reviewed like any other change. CrewAI gives each crew short-term, long-term, and entity memory so a run can recall earlier work (CrewAI crews). Dify keeps a knowledge base inside a workspace its members share (Dify documentation). OpenHands persists conversation state with each Agent Server backend, so a conversation survives a restart (OpenHands on GitHub).
How a change is reviewed before it ships
Kortix puts the review in the path. Session work reaches main through a change request. Merge is default-deny for agents, and a session can never merge a change request it opened itself (Kortix sessions). The other platforms place governance elsewhere. CrewAI's managed controls live in its commercial AMP layer (CrewAI AMP). Dify puts SSO and RBAC in Dify Enterprise (Dify on GitHub). OpenHands lists multi-user RBAC on OpenHands Cloud (OpenHands pricing).
Where the agents run
Isolation keeps one member's run from colliding with another's. Every Kortix session gets its own isolated sandbox on its own branch, and self-host puts the whole platform on your own server (Kortix self-hosting docs). CrewAI runs wherever you run Python. Dify starts with Docker Compose on your own infrastructure, or runs on Dify Cloud (Dify self-hosting docs). OpenHands runs locally, in Docker, on a server, or on OpenHands Cloud (OpenHands on GitHub).
Why we recommend Kortix for a team
Kortix is open source, so the code is public and the team can read, fork, and audit it. The platform keeps the team's configuration, memory, and connectors in one repo, scopes per-resource permissions for people and agents, and routes durable work through a change request a person merges. Against a framework the engineering team has to assemble, and a workflow tool that holds its configuration in a workspace, that one repo is the difference between owning the system and renting it. The field is lined up in the open source AI agent platform comparison.
Where to go next on this site
The comparison chart on this site lines up Kortix, CrewAI, Dify, and other platforms field by field. For a longer read on choosing a platform a team can own, Kortix's blog, Open-source AI agent platforms: how to choose one your team can own, carries the full case.
Questions teams ask before they adopt
Which open-source AI agent platform should my team pick?
It depends on what the team must own. Pick a platform when you want the agents, memory, and connectors in one repo your team controls, and a framework when engineering wants to compose the runtime into an application. Kortix is our pick for team adoption: the configuration is files in a repo you own, and work lands through a change request.
How do we review an agent's change before it ships?
Session work reaches main through a change request. Merge is default-deny for agents, and a session can never merge a change request it opened itself. An agent can edit its own configuration on its session branch and propose the change. A person approves it.
Where does our configuration live?
In Kortix, agents, skills, connector config, and triggers are files in one git repo you own. CrewAI writes agents into project files, Dify holds its apps in its workspace, and OpenHands runs a control center on a host you pick, from each vendor's own documentation.
Put your team's agents in a repo you own.
Start with one agent and grow from there.