{"slug": "open-source-python-framework-for-ai-agents-workflows-and-automations", "title": "Open-source Python framework for AI agents, workflows, and automations", "summary": "BindAI released v0.1, the first public version of its open-source, modular Python framework for building AI applications, installable via `python -m pip install bindai`. The release ships AI agents and agent execution, tool calling, workflow orchestration, memory, knowledge and RAG, multiple model providers, multi-agent execution, MCP HTTP tool integration, background automation workers, a REST Service API with authentication, streaming responses, and Docker/Docker Compose deployment. BindAI said advanced distributed infrastructure and enterprise capabilities remain post-v0.1 roadmap work.", "body_md": "Build AI software.\n\nBindAI is an open-source, modular Python framework for building AI applications with reusable components for agents, tools, workflows, memory, knowledge and RAG, model providers, integrations, automation, and runtime infrastructure.\n\nWhether you're building an AI assistant, document-processing application, workflow automation, or multi-agent system, BindAI provides building blocks that can grow with your application.\n\nBindAI v0.1 establishes the first public release foundation for building and deploying AI applications.\n\nThe v0.1 release includes:\n\n- AI agents and agent execution\n- Tools and tool calling\n- Workflows and workflow orchestration\n- Memory\n- Knowledge and RAG\n- Multiple model providers\n- Multi-agent execution\n- External service connections\n- MCP HTTP tool integration\n- Automation definitions and event triggers\n- Background automation workers\n- REST Service API\n- API authentication\n- Project, agent, workflow, and run execution APIs\n- Streaming API responses\n- Docker and Docker Compose deployment\n- Runtime execution events and observability foundation\n\nSome advanced distributed infrastructure and enterprise capabilities remain post-v0.1 roadmap work.\n\nInstall the main BindAI package:\n\n```\npython -m pip install bindai\n```\n\nVerify the installation:\n\n```\nbindai version\n```\n\nNote: use `bindai version` to display the installed CLI version.\n\nClone the repository:\n\n```\ngit clone https://github.com/BindBrain/BindAI.git\n\ncd BindAI\n```\n\nBindAI is organized as a multi-package **uv workspace**.\n\nInstall the development workspace with:\n\n```\npython -m pip install uv\n\nuv sync\n```\n\nActivate the environment.\n\n```\n.\\.venv\\Scripts\\Activate.ps1\nsource .venv/bin/activate\n```\n\nVerify the development installation:\n\n``` python\npython -c \"import bindai; print('BindAI import OK')\"\n\nbindai version\n```\n\nThe repository root is a workspace containing multiple BindAI packages. Do not use `pip install -e .` from the repository root.\n\nIndividual packages are located under:\n\n```\npackages/\n```\n\nand are managed together through the workspace configuration.\n\nThe recommended agent construction API is `Agent.builder()`:\n\n``` python\nfrom bindai import Agent\n\nagent = (\n    Agent.builder()\n    .name(\"assistant\")\n    .instructions(\"You are a helpful AI assistant.\")\n    .build()\n)\n\nresponse = agent.run(\"Explain what BindAI is.\")\n\nprint(response.output)\n```\n\nYou can incrementally add capabilities such as:\n\n- Tools\n- Memory\n- Knowledge and RAG\n- Workflows\n- Human approval\n- Scheduling\n- Multiple AI providers\n- Multi-agent execution\n- Automation\n- External integrations\n- MCP tools\n\nBindAI agents provide the foundation for AI application execution.\n\nCurrent capabilities include:\n\n- Agent construction and configuration\n- Conversation management\n- Prompt management\n- Tool calling\n- Structured outputs\n- Streaming responses\n- Memory integration\n- Knowledge and retrieval integration\n- Agent delegation\n- Multi-agent team execution\n\nTools allow agents and workflows to interact with application functionality and external systems.\n\nCurrent capabilities include:\n\n- Automatic tool registration\n- Function schema generation\n- Tool metadata\n- Context-aware execution\n- Tool results\n- MCP-discovered tools\n- Tool execution through the runtime\n\nBindAI workflows provide orchestration around agents and other workflow nodes.\n\nSupported workflow patterns include:\n\n- Sequential execution\n- Conditional branching\n- Loops\n- Parallel execution\n- Retry policies\n- Timeouts\n- Human tasks\n- Scheduling\n\nBindAI provides pluggable memory infrastructure for conversation and long-term application context.\n\nCurrent memory implementations include:\n\n- In-memory memory\n- SQLite\n- PostgreSQL\n- Vector memory\n- Pinecone\n- Chroma\n\nBindAI provides a Knowledge and Retrieval architecture for document-based AI applications.\n\nCurrent capabilities include:\n\n- Document loading\n- Document chunking\n- Embeddings\n- Vector retrieval\n- BM25 retrieval\n- Hybrid retrieval\n- Reranking\n- Conversational retrieval\n- Knowledge ingestion pipelines\n- RAG integration\n\nBindAI supports a modular provider architecture for connecting AI applications to different model providers.\n\nCurrent providers include:\n\n- OpenAI\n- Anthropic\n- Google Gemini\n- Groq\n- Ollama\n- OpenRouter\n\nThe provider architecture is modular so additional providers can be added independently.\n\nBindAI provides a connection layer for external services.\n\nCurrent v0.1 integrations include:\n\n- Webhooks\n- GitHub\n- Slack\n- Notion\n- Jira\n- Discord\n- Resend\n- Vercel\n- Netlify\n- Google Sheets\n- Google Docs\n- Gmail\n- Google Drive\n\nConnections can be used as reusable integration components for applications, agents, workflows, and automation.\n\nBindAI v0.1 includes a lightweight MCP HTTP integration for discovering and calling tools exposed through an MCP-compatible HTTP service.\n\nThe MCP package provides:\n\n- MCP tool discovery\n- Remote tool definitions\n- Tool schema propagation\n- HTTP-based tool execution\n- Integration with BindAI's tool system\n\nThe current implementation is intentionally lightweight. It is an HTTP bridge for MCP-style tool discovery and execution rather than a complete MCP server or full MCP protocol implementation.\n\nBindAI provides an automation layer for defining, executing, tracking, and running automations in the background.\n\nCurrent automation capabilities include:\n\n- Automation definitions\n- Event trigger framework\n- Unified trigger management\n- Automation execution runs\n- Persistent automation state abstractions\n- Automation run history\n- In-memory state storage\n- In-memory run history\n- Background automation workers\n\nThe current background worker uses a process-local thread pool. Distributed queues and horizontally scalable worker infrastructure are planned for later releases.\n\nAutomation execution is built on top of BindAI's existing executable and runtime architecture rather than introducing a separate execution model.\n\nBindAI v0.1 includes a REST API package for exposing BindAI applications and execution capabilities as a service.\n\nThe API currently provides:\n\n- Health checks\n- API key authentication\n- Agent execution\n- Workflow execution\n- Project execution\n- Run tracking\n- Streaming responses\n- Background execution\n\nThe API uses FastAPI and can be run with Uvicorn.\n\nExample:\n\n```\nuv run uvicorn bindai_api.app:app --host 0.0.0.0 --port 8000\n```\n\nThe health endpoint is publicly accessible:\n\n```\nGET /health\n```\n\nProtected API routes use:\n\n```\nAuthorization: Bearer <BINDAI_API_KEY>\n```\n\nThe Service API is intended to provide the foundation for deploying BindAI applications as services.\n\nBindAI v0.1 includes Docker and Docker Compose support for running the Service API and its supporting runtime.\n\nBuild the Docker image:\n\n```\ndocker build -t bindai .\n```\n\nRun the API container:\n\n```\ndocker run --rm -p 8000:8000 bindai\n```\n\nThe repository also includes Docker Compose configuration for local multi-service development and deployment.\n\nThe v0.1 deployment model is intentionally straightforward. Distributed queues, Kubernetes deployment, and horizontally scalable worker infrastructure are planned for later releases.\n\nBindAI v0.1 includes an event-driven observability foundation across runtime and agent execution.\n\nCurrent capabilities include:\n\n- Runtime execution events\n- Event bus infrastructure\n- Execution context\n- Agent lifecycle events\n- Model request and response events\n- Tool execution events\n- In-memory event recording\n- Event retrieval by execution\n\nThe current observability layer provides the foundation for future tracing, metrics, dashboards, and external observability integrations.\n\nIt should not yet be considered a complete production monitoring platform.\n\nBindAI is built as a modular package ecosystem.\n\n| Package | Purpose | \n|---|---|\n| `bindai` | Main framework package | \n| `bindai-agent` | AI agent framework | \n| `bindai-application` | Application layer | \n| `bindai-automation` | Automation definitions, triggers, execution state, history, and workers | \n| `bindai-cli` | Command-line interface | \n| `bindai-config` | Configuration | \n| `bindai-connections` | External connections and integrations | \n| `bindai-core` | Core framework abstractions | \n| `bindai-embeddings` | Embedding providers | \n| `bindai-group` | Agent groups and multi-agent execution | \n| `bindai-host` | Hosting infrastructure | \n| `bindai-knowledge` | Knowledge and RAG | \n| `bindai-mcp` | MCP HTTP tool integration | \n| `bindai-memory` | Memory providers | \n| `bindai-model` | Model abstractions | \n| `bindai-project` | Project management | \n| `bindai-prompt-builder` | Prompt construction | \n| `bindai-prompts` | Prompt system | \n| `bindai-providers` | Provider abstractions and registry | \n| `bindai-retrieval` | Retrieval implementations | \n| `bindai-runtime` | Runtime infrastructure and execution events | \n| `bindai-task` | Tasks and human tasks | \n| `bindai-tool` | Tool system | \n| `bindai-workflow` | Workflow engine | \n\nThe repository is structured as a workspace so individual packages can evolve independently while remaining part of the BindAI ecosystem.\n\n```\n                         AI Application\n                               |\n              +----------------+----------------+\n              |                |                |\n            Agents          Workflows         Tools\n              |                |                |\n              +----------------+----------------+\n                               |\n                 +-------------+-------------+\n                 |                           |\n              Memory                    Knowledge\n                 |                           |\n                 +-------------+-------------+\n                               |\n                         Retrieval / RAG\n                               |\n              +----------------+----------------+\n              |                |                |\n          Providers       Connections          MCP\n              |                |                |\n              +----------------+----------------+\n                               |\n                          Automation\n                               |\n                 +-------------+-------------+\n                 |             |             |\n              Triggers       Runs         Workers\n                 |             |             |\n                 +-------------+-------------+\n                               |\n                    Service API / Deployment\n                               |\n                    Runtime / Core\n                               |\n                         Observability\n```\n\nThe architecture is intentionally modular. Applications can use individual components or combine them into larger AI systems.\n\nAutomation builds on the same executable and runtime abstractions used by the rest of BindAI, while keeping automation state and historical run records in the automation layer.\n\nThe Service API provides an application-facing HTTP layer over the underlying framework and runtime.\n\n```\nBindAI/\n|\n+-- packages/\n|   +-- bindai/\n|   +-- bindai-agent/\n|   +-- bindai-application/\n|   +-- bindai-automation/\n|   +-- bindai-cli/\n|   +-- bindai-config/\n|   +-- bindai-connections/\n|   +-- bindai-core/\n|   +-- bindai-embeddings/\n|   +-- bindai-group/\n|   +-- bindai-host/\n|   +-- bindai-knowledge/\n|   +-- bindai-mcp/\n|   +-- bindai-memory/\n|   +-- bindai-model/\n|   +-- bindai-project/\n|   +-- bindai-prompt-builder/\n|   +-- bindai-prompts/\n|   +-- bindai-providers/\n|   +-- bindai-retrieval/\n|   +-- bindai-runtime/\n|   +-- bindai-task/\n|   +-- bindai-tool/\n|   +-- bindai-workflow/\n|\n+-- docs/\n+-- examples/\n+-- scripts/\n|\n+-- CHANGELOG.md\n+-- CONTRIBUTING.md\n+-- CODE_OF_CONDUCT.md\n+-- SECURITY.md\n+-- LICENSE\n+-- README.md\n+-- RELEASE.md\n+-- docs.json\n+-- Dockerfile\n+-- docker-compose.yml\n+-- pyproject.toml\n```\n\nFull documentation is available at:\n\nThe documentation currently covers:\n\n- Getting Started\n- Core Concepts\n- Agents\n- Tools\n- Memory\n- Knowledge and RAG\n- Workflows\n- Projects\n- Connections\n- Automation\n- Service API\n- API Authentication\n- Streaming\n- Background Runs\n- Deployment\n- MCP\n- Roadmap\n\nThe `examples/` directory also contains runnable examples covering many current BindAI capabilities.\n\nBindAI is being developed incrementally.\n\nThe v0.1 release establishes the foundation for:\n\n- AI agents\n- Workflows\n- Tools\n- Memory\n- Knowledge and RAG\n- Model providers\n- Multi-agent execution\n- External connections\n- MCP HTTP integration\n- Automation\n- Background workers\n- REST Service API\n- Docker and Compose deployment\n- Runtime observability foundation\n\nFuture development areas include:\n\n- Distributed queues\n- Horizontally scalable workers\n- Kubernetes deployment\n- Advanced tracing and metrics\n- Advanced agent capabilities\n- Expanded automation infrastructure\n- Additional integrations\n- Visual workflow tooling\n- Enterprise capabilities\n- Voice AI\n- Templates and business solutions\n\nSee the current roadmap in the documentation for implementation status and priorities.\n\nThe repository contains examples demonstrating current framework capabilities.\n\n```\nexamples/\n```\n\nExamples cover areas such as:\n\n- Basic agents\n- Tools\n- Streaming\n- Memory\n- Knowledge and RAG\n- Workflows\n- Multi-agent execution\n- Human approval\n- Retry and timeout handling\n- FastAPI integration\n- CLI usage\n- MCP\n- Custom providers\n- Custom embeddings\n- Custom memory\n- Custom retrievers\n- Configuration\n- Automation\n- Production-oriented application structure\n\nThe examples are intended to demonstrate framework usage and should be evaluated according to the maturity of the underlying APIs.\n\nContributions are welcome.\n\nBefore contributing, please read:\n\n- `CONTRIBUTING.md`\n- `CODE_OF_CONDUCT.md`\n- `SECURITY.md`\n\nTypical development setup:\n\n```\nuv sync\n```\n\nRun the test suite with:\n\n```\nuv run pytest\n```\n\nRun linting with:\n\n```\nuv run ruff check .\n```\n\nCheck formatting with:\n\n```\nuv run ruff format --check .\n```\n\nRun type checking with:\n\n```\nuv run mypy packages examples\n```\n\nPlease add or update tests when changing framework behavior.\n\n- Documentation: [https://docs.bindai.dev](https://docs.bindai.dev)\n- GitHub: [https://github.com/BindBrain/BindAI](https://github.com/BindBrain/BindAI)\n- Website: [https://bindai.dev](https://bindai.dev)\n\nBindAI is released under the **MIT License**.\n\nSee [`LICENSE`](https://github.com/BindBrain/BindAI/blob/refactor/packages/LICENSE) for the complete license text.\n\nBindAI aims to become a complete open-source ecosystem for building AI software.\n\nThe long-term vision includes:\n\n- Production-oriented AI applications\n- Modular AI infrastructure\n- Advanced agent and multi-agent systems\n- AI workflow automation\n- Enterprise integrations\n- Visual workflow tooling\n- Developer-focused AI application infrastructure\n\nBindAI is under active development.\n\n**BindAI v0.1 establishes the first public release foundation** for building, integrating, automating, and deploying AI applications.\n\nThe current release provides a substantial foundation for:\n\n- Agents\n- Workflows\n- Tools\n- Memory\n- Knowledge and RAG\n- Model providers\n- Multi-agent execution\n- External connections\n- MCP tool integration\n- Automation\n- Background workers\n- REST APIs\n- Docker deployment\n- Runtime observability\n\nThe v0.1 automation system includes automation definitions, event triggers, execution state, run history, and background workers.\n\nThe v0.1 Service API provides authenticated REST endpoints for application execution, projects, workflows, agents, runs, streaming, and background execution.\n\nSome roadmap areas remain under development and should not yet be considered complete distributed or enterprise platform capabilities.\n\n**Build AI Software. Scale Everywhere.**\n\nMade with ♥ by **BindBrain**", "url": "https://wpnews.pro/news/open-source-python-framework-for-ai-agents-workflows-and-automations", "canonical_source": "https://github.com/BindBrain/BindAI", "published_at": "2026-09-14 23:51:55+00:00", "updated_at": "2026-09-15 00:26:54.687057+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "developer-tools", "ai-products", "artificial-intelligence"], "entities": ["BindAI", "BindBrain", "Python", "GitHub", "Docker", "Docker Compose", "MCP", "REST Service API"], "alternates": {"html": "https://wpnews.pro/news/open-source-python-framework-for-ai-agents-workflows-and-automations", "markdown": "https://wpnews.pro/news/open-source-python-framework-for-ai-agents-workflows-and-automations.md", "text": "https://wpnews.pro/news/open-source-python-framework-for-ai-agents-workflows-and-automations.txt", "jsonld": "https://wpnews.pro/news/open-source-python-framework-for-ai-agents-workflows-and-automations.jsonld"}}