Suna
The Company AI Command Center
Last verified:
What is Suna?
Kortix (Suna) is The Autonomous Company Operating System—a cloud computer where AI agents run your entire company. It is a fully open-source platform that lets you connect 3,000+ tools, configure autonomous AI agents as markdown files, set triggers (cron schedules and webhooks), and deploy them to operate 24/7 with persistent, shared memory across all agents. Unlike simple chatbots, Suna executes real work: research, data analysis, browser automation, file management, system operations, and complex multi-agent workflows.
Key features include: Agents (autonomous workers defined as markdown files with identity, permissions, tools, and triggers); Skills (60+ built-in reusable knowledge packs for coding, browser automation, deep research, legal writing, spreadsheets, data analysis); Autowork (autonomous execution loop where agents work until tasks are verified complete); Triggers (time-driven cron schedules and event-driven webhooks); Memory (persistent, filesystem-based, shared across all agents so context compounds over time); Orchestration (one agent delegates to many parallel sub-agents like departments); and Open standards (runs on OpenCode, an open foundation for agent skills and tools).
Kortix is for individuals, developers, small teams, and companies who want to automate their entire knowledge work—engineering, sales, finance, ops, legal, support. It serves developers who want to contribute to open-source AI, creators who need pre-configured agents, and businesses seeking enterprise-grade automation accessible to everyone. The platform runs inside a shared Linux Ubuntu machine with bash, real filesystem, package managers, databases, and full software ecosystem access.
The system connects every tool your company uses via OAuth apps, MCP servers, REST APIs, CLI tools, and environment variables. You define agents for every function (engineering, sales, finance, ops, legal, support), compose them into an autonomous workforce, and check in from the dashboard, mobile apps, Slack, MS Teams, Telegram, or Discord. Kortix is open-sourced so users can investigate, modify, and enhance AI capabilities.
Suna pricing
Pricing model: Freemium
Kortix is fully open-source and free to self-host. The platform is open-sourced so users can investigate, modify, and enhance AI capabilities at no cost. You run it on your own infrastructure using Docker or manual setup. API costs for LLM providers (Anthropic, OpenAI, Groq) and search APIs (Tavily, Firecrawl) are paid separately based on your usage with those providers.
Suna pros
- Fully open-source framework with transparent, modifiable code
- 3,000+ tool integrations via OAuth, MCP, REST, CLI, env vars
- Persistent shared memory across all agents—context compounds over time
- 60+ built-in Skills for coding, research, browser automation, legal writing
- Autowork autonomous loop self-verifies until task is proven correct
- Real Linux Ubuntu environment with bash, filesystem, package managers, databases
- Multi-agent orchestration with parallel sub-agents like company departments
- Time-driven and event-driven triggers (cron schedules and webhooks)
- Agents defined as simple markdown files with identity, permissions, tools
- 24/7 autonomous operation—agents run continuously without supervision
- Accessible via web dashboard, iOS/Android apps, Slack, Teams, Telegram, Discord
- Custom Skills are just writing code—easy to create new ones
- Git-versionable configuration for all agent settings and triggers
- OpenCode open standard foundation for agent skills and tools
- Single shared machine where every agent sees same filesystem and databases
- Built for all knowledge work: engineering, sales, finance, ops, legal, support
Suna cons
- Requires self-hosting setup with Python, Docker, and technical knowledge
- Needs API keys for LLM providers (Anthropic, OpenAI, Groq) and search APIs
- Full Linux environment may be overkill for simple chatbot use cases
- Setup wizard required—not a one-click SaaS deployment
- Active development—may have bugs or breaking changes in updates
- Requires 24/7 running machine with persistent resources
- Learning curve for defining agents as markdown files with frontmatter
- Best results require giving agents full context and all credentials—security consideration
Frequently asked questions about Suna
What is Suna?
Suna is Kortix's flagship open super AI worker—a generalist AI that handles research, data analysis, browser automation, file management, and complex workflows through natural conversation. It executes real work instead of just responding, acting like a genius colleague working 24/7.
What makes Kortix different from other AI assistants?
Kortix gives agents full context—every secret, integration, and piece of institutional knowledge—and lets them operate freely on a real Linux machine. All agents share the same filesystem, databases, and history, so context compounds over time instead of being siloed per session. It automates an entire company's knowledge work, not just single tasks.
How do agents work in Kortix?
Each agent is a markdown file with its own identity, permissions, tools, and triggers. Agents have bash access, a real filesystem, package managers, databases, and the entire software ecosystem. You define agents for functions like engineering, sales, finance, ops, legal, and support, then compose them into an autonomous workforce inside one shared machine.
What are Skills in Kortix?
Skills are reusable knowledge packs that teach agents how to do real work—coding, browser automation, deep research, legal writing, spreadsheets, data analysis, and more. There are 60+ built-in Skills, and creating new ones is just writing code. Each Skill teaches agents specific capabilities for their tasks.
What is Autowork?
Autowork is the autonomous execution loop where an agent works until the task is done, self-verifies the result, and only stops when it can prove the outcome is correct. It keeps agents going autonomously until job verification is complete, without needing constant human supervision.
How do triggers work?
Triggers are time-driven (cron schedules) or event-driven (webhooks) that fire agents automatically. They're defined in agent frontmatter as Git-versionable config. Use cases include morning briefings, recurring jobs, and real-time reactions to events. Actions can be prompts, commands, or HTTP requests.
How is memory handled in Kortix?
Memory is persistent, filesystem-based, and shared across all agents. Every decision, preference, and context is retained on the same machine. The longer the system runs, the smarter the whole company gets because context isn't siloed—it compounds over time across all agents.
What integrations does Kortix support?
Kortix connects to every tool your company uses: OAuth apps, MCP servers, REST APIs, CLI tools, and environment variables. There are 3,000+ integrations available including Gmail, Slack, GitHub, Stripe, Notion, HubSpot, and Drive. Custom integrations are trivial to add if something has an interface.
How do I get started with Kortix?
Clone the repository from GitHub, run python setup.py to go through the automated setup wizard that configures all services with progress saving, then use python start.py to manage the platform. You can add/update API keys for LLM providers and search APIs by running setup.py again. The platform runs with Kortix Super Worker ready to assist.
Who should use Kortix?
Kortix is for individuals, developers, small teams, and companies who want to automate their entire knowledge work. It's for developers who want to contribute to open-source AI, creators who need pre-configured agents, and businesses seeking enterprise-grade automation. It handles engineering, sales, finance, ops, legal, support, and everything in between.