Overview: Kortix is an open-source platform designed for developers and teams who want to create autonomous AI agents capable of performing complex tasks like web browsing, file manipulation, data analysis, and system operations. Unlike generic LLM interfaces, Kortix provides a full-stack infrastructure—backend API, frontend dashboard, isolated agent runtime, and persistent storage—to build, deploy, and monitor AI agents at scale. It’s built for users who need more than chat: they want agents that can act independently across digital environments, from automating customer service to conducting market research or managing DevOps workflows. The platform is optimized for self-hosting, with Docker and manual setup options, making it ideal for privacy-conscious teams or organizations requiring full control over their AI infrastructure.
What You Get
- Kortix Super Worker - A pre-built generalist AI agent that performs research, browser automation, file management, data analysis, and system operations via natural language commands
- Browser Automation - Agents can navigate websites, extract data, fill forms, and automate web workflows using integrated headless browser tools
- File & Document Management - Agents can create, edit, organize, and convert documents, spreadsheets, presentations, and code files within a secure sandbox
- Web Intelligence - Built-in web crawling, search integration (Tavily, Firecrawl), and data synthesis to gather and summarize information from multiple sources
- System Operations - Safe execution of command-line tasks, system administration, and DevOps automation through isolated runtime environments
- Agent Builder - Visual tools to configure custom agents with specific workflows, knowledge bases, and tool integrations
- API Integrations - Support for multiple LLM providers (OpenAI, Anthropic, Groq) via LiteLLM and external APIs for search, CRM, and data services
- Docker-based Agent Runtime - Each agent runs in an isolated container with restricted access to system resources for security and scalability
- Supabase-Powered Storage - Centralized database for user auth, agent configurations, conversation history, file storage, and real-time monitoring
Common Use Cases
- Building a customer service agent - Automate support ticket resolution by training an agent to parse FAQs, extract user intent from chat logs, and escalate complex issues while tracking satisfaction metrics
- Creating a market research agent - Configure an agent to crawl competitor websites, extract pricing and product features, compare them against your own offerings, and generate weekly competitive intelligence reports
- Automating data extraction from web forms - Use browser automation to fill out registration forms across 50+ platforms, extract confirmation emails, and save structured data into spreadsheets
- DevOps teams managing microservices - Deploy AI agents to monitor server logs, run health checks via CLI commands, trigger restarts on failures, and notify teams via Slack or email
- Content creators managing publishing schedules - Build an agent that generates social media posts from blog drafts, schedules them across platforms, and tracks engagement metrics
- Research teams analyzing academic papers - Train an agent to search PubMed, extract key findings from PDFs, summarize results, and generate annotated bibliographies
Under The Hood
Kortix AI Suna is a unified agent platform designed to support cross-platform development across web, desktop, and mobile environments. It integrates a modular architecture with shared backend services, enabling seamless agent workflows and tool execution across diverse environments.
Architecture
The system adopts a microservices-like structure with distinct apps and backend services, emphasizing clear separation of concerns.
- The architecture follows a multi-tiered design with well-defined boundaries between UI, business logic, and data layers.
- Component interactions are facilitated through standardized APIs and shared libraries such as @agentpress/shared.
- Design patterns like dependency injection and configuration-driven agent loading enhance modularity and extensibility.
Tech Stack
The project leverages a modern, multi-platform tech stack built with TypeScript and Python.
- Built primarily with TypeScript for frontend and React ecosystem, and Python for backend using FastAPI.
- Integrates Next.js, Electron, and Expo to support web, desktop, and mobile experiences respectively.
- Relies on Supabase for backend services and PostHog for analytics, with pnpm and Turbopack for development.
Code Quality
The codebase reflects a mixed quality with strengths in frontend structure and administrative tools.
- Testing is present but limited in scope, with a variety of test functions targeting different system components.
- Error handling follows common patterns with widespread use of try/catch blocks for robustness.
- Component structure and UI patterns show consistency, though some technical debt exists in configuration management.
What Makes It Unique
Kortix introduces innovative approaches to agent execution and tool integration in a unified platform.
- Implements MCP (Model Control Protocol) as a core integration layer for dynamic and extensible agent capabilities.
- Features a sandboxed execution model that isolates agent operations while enabling access to system resources.
- Combines desktop, web, and backend services into a single modular architecture with shared state management.