imina is an open-source, all-in-one platform designed to simplify and accelerate the development of AI applications. It seamlessly integrates a visual workflow engine, RAG pipelines, and multi-agent systems, enabling developers to move quickly from prototype to production.
Inspired, and forked by Langflow, imina has been completely rewritten from the ground up to offer a more modern, powerful, and production-ready solution.
We are actively working on the following features and welcome community feedback:
- Formal Open-Source Release: Complete code cleanup and documentation to launch the community edition.
- Enhanced Multi-Agent Collaboration: Introduce more advanced coordination and self-planning capabilities for agents.
- Dynamic Workflow Control: Support for pausing, resuming, and modifying workflows at any node during execution.
- Knowledge Hub Editor: A more powerful rich-text editing and knowledge management experience.
- Component Marketplace: A community-driven platform for sharing and discovering custom components.
- Observability & Analytics: Detailed tracing, logging, and performance analysis for workflow executions.
-
Out-of-the-Box Applications:
- Intelligent Chat: A complete, multi-modal conversational AI with built-in RAG capabilities, web search, and persistent history.
- Knowledge Hub: A Notion-like system with a full document processing pipeline (upload, chunk, embed, summarize) to instantly create and manage knowledge bases for your AI.
-
Modern Workflow Engine
- Visually orchestrate complex AI workflows on an intuitive drag-and-drop canvas, supporting complex logic with branches, loops, and parallel execution.
- Access a rich library of over 200+ out-of-the-box components, covering data processing, major LLM/Embedding models, API calls, logic control, and more.
- Build both single agents and multi-agent systems with support for sequential, parallel, and hierarchical (manager-worker) collaboration patterns.
- Seamlessly write and execute custom logic with the Python Code component for infinite extensibility.
-
End-to-End RAG Pipeline
- Automated content extraction, chunking, and embedding for various document formats (PDF, DOCX, MD, etc.).
- An integrated knowledge hub to manage vectorized content, serving as a persistent knowledge source for your AI applications.
- Built-in text splitters (character, recursive, semantic) and rerankers to optimize retrieval quality.
-
Comprehensive Model & Ecosystem Support
- Seamless LLM Integration: Easily connect and switch between dozens of models from OpenAI, Anthropic, Google Gemini, DeepSeek, Ollama, Groq, and more.
- Diverse Vector Stores: Support for multiple vector databases, including Chroma, Pinecone, Qdrant, Weaviate, and Elasticsearch.
- Rich Tool Integrations: 50+ built-in tools for Google Search, Slack, Notion, Tavily, and other external services.
-
Built for Production
- Backend-as-a-Service: Expose all core functionalities via APIs to integrate Mina into your existing business systems.
- Triggers & Deployment: Trigger workflows via API, webhooks, CRON jobs, or email.
- Observability: Provide detailed logging, tracing, and performance analytics to monitor and continuously improve your applications.
- Enterprise-Grade Features (Upcoming): Support for SSO, multi-tenancy, and fine-grained access control.
Cloud
imina Cloud is in the works! You'll soon be able to try all features with zero setup. Stay tuned.
Self-Hosting
imina will be open-sourced soon. Once released, you can quickly get it running in your local environment using Docker.
# [To be provided upon open-sourcing]
git clone https://github.com/tempurai/imina.dev
cd imina/docker
docker-compose up -dAfter startup, access the imina console at
http://localhost:PORT.