kortix-ai/suna

What it solves

Kortix provides an open-source AI Management System that allows companies to own their AI workforce, memory, and configuration. It solves the problem of vendor lock-in and data privacy associated with closed-source AI assistants by letting users self-host their infrastructure, use any model provider, and manage their entire AI setup as a version-controlled git repository.

How it works

Kortix treats a company's AI operations as a codebase. It uses a git repository to store agents, skills, and company memory. When work is requested, Kortix spawns a "session" on a dedicated, isolated cloud computer (a Linux sandbox). An agent performs the task within this sandbox and submits its results via a "change request," which a human must review and merge into the main branch to finalize the work.

Who it’s for

It is designed for organizations that need a scalable, auditable AI workforce capable of performing real work (like generating reports or deploying code) rather than just chatting, and who require full control over their data, infrastructure, and model choice.

Highlights

• Git-based Management: Agents, skills, and memory are versioned and diffable in a git repo.
• Isolated Execution: Every session runs on its own disposable cloud computer/sandbox to ensure security.
• Model Agnostic: Supports any model provider and allows users to use their own API keys.
• Extensive Connectivity: Includes 3,000+ app connectors, MCP, OpenAPI, and GraphQL support.
• Flexible Deployment: Can be run on a managed cloud, a private VPC, or on-premise networks.