AI-First Process Automation for the Desktop Era
OpenAdapt.AI is an open-source Generative Process Automation platform that transforms desktop automation through machine learning. Record human demonstrations, train vision-language models, and deploy agents that adapt to any software environment.
- Learn by Demonstration: Automatically learns from user interactions to generate automation scripts
- Adaptive Intelligence: Adapts to software changes and complex environments using multimodal models
- Privacy-First: Built-in data protection with PII/PHI detection and redaction
- Cross-Platform: Works with browsers and desktop applications across macOS and Windows
- Multi-Model Support: Integrates with OpenAI, Anthropic, Google, Ollama, and vLLM
- Open Source: Community-driven development with MIT licensing
| Repository | Description | Status |
|---|---|---|
| OpenAdapt ⭐ | Main platform orchestrating GUI automation with LMMs. Meta-package integrating all ecosystem components | 🟢 Active |
Data Capture & Processing
| Repository | Description | Status |
|---|---|---|
| openadapt-capture | Production-ready event streams with time-aligned media for GUI interaction recording | 🟢 Active |
| openadapt-viewer | HTML viewer components for visualization dashboards and benchmark results | 🟢 Active |
| openadapt-web | Web interface components for browser-based interaction | 🟢 Active |
Machine Learning & Intelligence
| Repository | Description | Status |
|---|---|---|
| openadapt-ml | ML training toolkit for multimodal GUI-action models | 🟢 Active |
| openadapt-grounding | UI element detection and localization with OmniParser integration | 🟢 Active |
| openadapt-retrieval | Multimodal demo retrieval system for similarity search in GUI automation | 🟢 Active |
Evaluation & Quality
| Repository | Description | Status |
|---|---|---|
| openadapt-evals | Evaluation infrastructure and benchmarks for GUI agent performance testing | 🟢 Active |
Privacy & Security
| Repository | Description | Status |
|---|---|---|
| openadapt-privacy | PII/PHI detection and redaction for GUI automation data with Presidio integration | 🟢 Active |
Execution & Runtime
| Repository | Description | Status |
|---|---|---|
| openadapt-agent | Production execution engine with safety gates, audit logging, and human-in-the-loop confirmation | 🟢 Active |
Observability
| Repository | Description | Status |
|---|---|---|
| openadapt-telemetry | Error tracking and usage analytics with privacy filtering for platform monitoring | 🟢 Active |
| Repository | Description | Status |
|---|---|---|
| OpenAdapter | Effortless deployment platform for screenshot parsing and action models on AWS EC2 | 🟢 Active |
| OmniMCP | Model Context Protocol integration with OmniParser for AI UI interaction (68 stars) | 🟢 Active |
| OmniMCP.web | Web interface for OmniMCP server | 🟢 Active |
OpenAdapt has contributed deployment infrastructure to major open-source projects:
- PR #52: Add Dockerfile and client.py; deploy to EC2 on AWS via Github Actions
- Status: Open (most commented PR in the repository)
- Contribution: Production deployment infrastructure with Docker, client library, and AWS automated deployment
- Impact: Enables production deployment of OmniParser for screen parsing tasks
- PR #19: Add Dockerfile and client.py; deploy to EC2 on AWS via Github Actions
- Status: Merged (first PR to the repository)
- Contribution:
deploy.py, Gradio interface, client library, and AWS deployment automation - Impact: Provided the foundation for SoM's production deployment capabilities
| Repository | Description | Status |
|---|---|---|
| OpenCUA | Fork of Computer-Use Agents framework that cites OpenAdapt as foundational | 🟡 Fork |
| OmniParser | Fork of Microsoft's OmniParser with deployment enhancements | 🟡 Fork |
| SoM | Fork of Set-of-Mark visual prompting with deployment infrastructure | 🟡 Fork |
| Repository | Description |
|---|---|
| PydanticPrompt | Library for documenting Pydantic models to generate structured LLM outputs |
| Repository | Description |
|---|---|
| atomacos | macOS automation library (fork) - critical dependency for macOS support |
| pynput | Input control library (fork) - critical dependency for cross-platform input handling |
| Repository | Description |
|---|---|
| OmniParser | Fork of Microsoft's OmniParser for screen parsing with OpenAdapt-specific modifications |
| SoM | Set-of-Mark visual prompting technique for precise UI element targeting |
| Repository | Description |
|---|---|
| .github | Organization-wide GitHub configuration, templates, and this profile README |
# Install core platform
pip install openadapt
# Or use uv for faster installation
uv pip install openadapt
# Record a demonstration
openadapt record
# Train a model
openadapt train
# Evaluate performance
openadapt eval- Website: openadapt.ai
- Email: info@openadapt.ai
- Discord: Join our community (link on website)
- X/Twitter: @OpenAdaptAI
We welcome contributions across all repositories:
- Frontend Development: openadapt-web, openadapt-viewer
- Machine Learning: openadapt-ml, openadapt-grounding, openadapt-retrieval
- Data Capture: openadapt-capture
- Privacy & Security: openadapt-privacy
- Agent Runtime: openadapt-agent
- Evaluation: openadapt-evals
- Documentation: All repositories
See individual repository CONTRIBUTING.md files for specific guidelines.
OpenAdapt.AI is currently in alpha. While functional, some features are still under development. We're actively working towards production-ready releases across the ecosystem.
Professional implementation services and organizational consulting are available. Contact info@openadapt.ai for details.
Support OpenAdapt.AI development through GitHub Sponsors. Your contributions help maintain and expand this open-source ecosystem.
All OpenAdapt.AI repositories are licensed under the MIT License unless otherwise specified. See individual repository LICENSE files for details.
