Kortix

Kortix is an open-source AI command center for companies that want to build, run, and govern AI agents in one place. It supports tool connectivity, agent permissions, and multiple deployment options for teams moving from experiments to production.

AI代理開發 AI開發者工具

訪問網站

Overview

Kortix is an open-source AI command center for companies that want to build, run, and govern AI agents in one place. The product is positioned around agent operations rather than chat, with a focus on connecting agents to tools, managing access, and keeping policy visible to the team.

The pricing and enterprise pages show a platform designed for both small experiments and production use. Free includes sandbox compute credits, Team adds pooled usage credits and managed frontier models, and Enterprise adds SAML SSO, SCIM directory sync, advanced RBAC, audit logs, and Cloud, VPC, or on-prem deployment options.

Core capabilities

Central agent control plane

Build, run, and govern AI agents in one platform rather than splitting setup, execution, and oversight across different systems.

Tool connectivity

Connect agents to a large tool ecosystem; the site positions the platform as able to connect to 3,000+ tools.

Credential proxying

Use a single Kortix token so connectors, models, and integrations are proxied through one credential rather than many exposed secrets.

Permission controls

Set allow, ask, and block rules per action, including network-level patterns, so agents only access what they need.

Git-based governance

Track policy, agents, and configuration in one `kortix.toml` inside a Git repo, with changes handled as diffs.

Model access options

Choose between managed models, BYOK, or a ChatGPT subscription depending on the plan and workflow.

Common use cases

Team agent operations
Set up agents that need access to multiple tools and credentials while keeping the environment governed through scoped tokens and permission rules.
Early-stage experimentation
Run pilots or demos on the Free plan using sandbox credits and one project before moving to a paid seat-based plan.
Department-level deployment
Use Team for production work where pooled credits, managed frontier models, and more projects or seats are needed.
Enterprise governance
Adopt Enterprise when you need SAML SSO, SCIM directory sync, audit logs, and deployment control across Cloud, VPC, or on-prem.
Git-centered administration
Keep policies, agents, and config in Git so changes can be reviewed, committed, and reverted like code.

Pros and Cons

Pros

Combines agent building, execution, and governance in one platform.
Supports a wide tool ecosystem, with the site claiming 3,000+ tool connections.
Offers multiple model access paths, including managed models, BYOK, and ChatGPT subscription support.
Provides governance features such as scoped tokens, action-level permissions, audit logs, and Git-based configuration.
Includes deployment options for Cloud, VPC, and on-prem on Enterprise.

Cons

The source does not provide a detailed public feature list beyond high-level platform capabilities.
Some integration details are broad, so readers may need to check product documentation for specific connectors and supported workflows.
Pricing separates seat fees, usage credits, and model/provider costs, which may take a moment to understand for first-time buyers.

FAQ

What does the Free plan include?

The pricing page shows a Free plan with 500 credits per month for sandbox compute only and 1 project. It also says free LLM models are included, while premium models can be used through your own API key or ChatGPT subscription.

What is included in the Team plan?

The Team plan is shown at $40 per seat per month and includes pooled usage credits, managed frontier models, and support for up to 200 projects and up to 100 seats.

How are agents and access governed?

The enterprise page says agent infrastructure can live in one Git repo, with policy, agents, and config in a single `kortix.toml`. It also says connectors, models, and integrations are proxied through a single Kortix token.

What deployment and admin options are available?

The enterprise page says deployment options include Cloud, VPC, or on-prem. It also mentions SAML SSO, SCIM directory sync, advanced RBAC, audit logs, an SLA, and a DPA on Enterprise.

How does usage-based billing work?

The pricing page says compute is billed by the second and Agent Computers auto-stop when idle. It also states Agent Computer runtime is about $0.10/hour and that credits cover sandbox runtime and managed model usage differently.

Quick Facts

Category: AI command center
Platform: Web-based SaaS with enterprise deployment options
Primary users: Companies and teams building AI agents
Source domain: suna.so
Open source: Yes, described as open-source
Deployment options: Cloud, VPC, or on-prem on Enterprise