Shared Source Initiative
Our initiative is focused on continuously fine-tuning engines with high-quality, domain-specific data — across both frontier/SOTA LLMs and models optimized for the edge. We crowdsource expertise and data to build shared-source models that support specific business domains of knowledge work, making sovereign AI accessible to every organization.
Want to contribute to building state-of-the-art open-source domain models?
All contributors receive free access to the model they contributed towards. Join us in building the next generation of specialized AI.
Reach out at vc@tuningengines.com
Application Computer Usage Model
What We're Building
A state-of-the-art Application Computer Usage Model built by taking the largest open-source dataset of computer-use recordings relating to application usage and embedding it into Qwen OSS multi-modal embeddings.
This model will understand how humans interact with software applications — clicks, navigation flows, form filling, data entry — enabling AI agents that can truly operate desktop and web applications the way people do.
Technical Approach
- Largest open-source computer-use recording dataset
- Qwen multi-modal embedding architecture
- Application-specific interaction pattern recognition
- Cross-platform UI understanding (desktop & web)
- Action sequence prediction and replay
- Crowdsourced data from real-world application usage
Planned Domain-Specific Models
Our continuous fine-tuning initiative extends across critical business domains — building and refining specialized models on both frontier and edge architectures as the community grows.
Clinical Medical Reasoning
Domain-specific LLM trained on medical literature, clinical notes, and diagnostic reasoning patterns.
Legal Document Analysis
Fine-tuned model for contract review, case law analysis, and regulatory compliance workflows.
Financial Analysis & Reporting
Specialized model for earnings analysis, risk assessment, and financial document understanding.
Scientific Research Assistant
Model trained on research papers, lab protocols, and scientific methodology across disciplines.