# Claude Cowork for a team: what open-source Kortix adds

> Claude Cowork is Anthropic's agent for multi-step work on paid Claude plans. A company also needs one shared configuration, model choice, review of agent work, and access control. Kortix is open source, and puts all four in one git repo you own.

URL: https://claudecoworkalternative.com/claude-cowork-for-teams

## How an agent's work reaches main

`project tree`

```
project (git repo + kortix.yaml)
└─ session ──> cloud computer: an isolated sandbox on a branch named after the session
└─ the OpenCode agent works
└─ change request ──> you review & merge ──> main
```

Source: https://github.com/kortix-ai/suna

The project tree from the Kortix repository, quoted as written. A person reviews and merges what an agent proposes.

## What Claude Cowork is built for

Claude Cowork is Anthropic's agent for multi-step knowledge work. You describe an outcome, point it at folders and tools you choose, and it returns finished documents, spreadsheets, and reports for review. It runs on Anthropic's models only, the code is closed, and it is included with a paid Claude plan. The setup lives inside the product: projects, plugins, and skills sit in the Claude account, and a session in the cloud is saved with that account. For one person handing off occasional tasks, that is a complete package. Anthropic's own help center documents where it runs and how an administrator manages it, in [Cowork on Team and Enterprise plans](https://support.claude.com/en/articles/13455879-use-claude-cowork-on-team-and-enterprise-plans).

## What changes when the work belongs to a company

A team lead weighs four things a personal plan does not settle. Shared configuration asks whether everyone runs the same agents and instructions. Model choice asks whether the company can bring any model or is tied to one lab. Review asks whether a change lands only after a person reads it. Access control asks who may touch which tool and which data. Claude Cowork answers part of this: on Team and Enterprise plans an owner can turn Cowork on or off for the organization, set connector-tool approvals, and use roles, and Enterprise adds SSO, SCIM, and audit logs. The code and the model still belong to Anthropic, and the configuration still lives in individual Claude accounts rather than in version control the team owns.

## Where open-source Kortix fits

Kortix is the open-source AI Operating System, and it puts the company's shared setup in version control. A project is a git repo with kortix.yaml at its root, so your agents, their skills, your company memory, and every connector are files in it, and a config change is reviewed like any other change. [Kortix on GitHub](https://github.com/kortix-ai/suna) carries the whole system in the open.

Model choice stays the company's: any model provider with your own keys, or the ChatGPT plan you already pay for. Review is built into the path to production, because session work reaches main through a change request and merge is default-deny for agents, so a person reads the diff before it lands. Access control sits with the account: per-resource permissions for people and agents, with roles, groups, and an audit trail, plus SAML 2.0 single sign-on and SCIM 2.0 for the directory. Approval gates you set, off until you set them, so each tool call can be set to allow, ask, or block down to the arguments of each call.

The same system runs on Kortix Cloud, in your VPC, or on your own on-prem network, and self-host is free. Moving a product held in one account to a repo the team owns is mostly mechanical. The [migration guide](/migration) walks the order to do it in, the [comparison on this site](/comparison) shows the wider field, and [the Kortix blog](https://kortix.com/blog/kortix-vs-claude-cowork) takes the direct matchup apart.

_FAQ_

## Questions a team lead asks

Short answers, each sourced from the vendor's own page.

### Does Claude Cowork work for a whole company?

Yes, on Team and Enterprise plans, where an owner can enable or disable it and set connector-tool approvals. The code and models stay Anthropic's, and each member's configuration lives in that member's Claude account. See [the Team and Enterprise help page](https://support.claude.com/en/articles/13455879-use-claude-cowork-on-team-and-enterprise-plans).

### Can our team run one shared configuration?

In Cowork, projects, plugins, and skills live in individual Claude accounts. In Kortix they are files in one git repo you own, so the whole team runs the same versioned setup, and a change is a diff a person reviews.

### How does a team review an agent's work before it ships?

With Kortix, session work reaches main through a change request, and merge is default-deny for agents, so a person reads the diff before it lands. Cowork gates actions with in-product approvals instead.

### Can the company choose the model?

With Kortix, any model provider with your own keys, or the ChatGPT plan you already pay for. Claude Cowork runs Anthropic’s models only.

### Can we self-host and control access?

Kortix self-hosts free, and per-resource permissions for people and agents come with roles, groups, and an audit trail, plus SAML 2.0 SSO and SCIM 2.0. Cowork has no self-host option.

## Put the system in a repo your team owns.

Start free on Kortix Cloud, or self-host and keep the agents, skills, and memory under your control.

[Try Kortix](https://kortix.com) · [See how to migrate](https://claudecoworkalternative.com/migration)

Open source · Any model, your keys · Self-host, VPC, or on-prem