Reproduce “Opus: A Large Work Model for Complex Workflow Generation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Théo Fagnoni, Bellinda Mesbah, Mahsun Altin, Phillip Kingston
The Opus framework develops a Large Work Model (LWM) that generates and optimizes complex workflows for Business Process Outsourcing (BPO) by integrating a Work Knowledge Graph (WKG) and an explicit Intention layer. In a medical coding use case, Opus Alpha 1 models surpassed leading LLMs by up to 38% in workflow generation performance.
Key insights
- Integrating a structured Work Knowledge Graph with a Large Work Model significantly enhances the generation of context-aware, semantically aligned, and auditable workflows for proprietary, complex domains.
- Explicitly modeling and capturing user intention through a dedicated layer provides a more precise and reliable foundation for workflow generation compared to implicit prompt engineering.
- A two-phase methodology, separating workflow generation from a subsequent, cost-function-driven optimization phase, is effective for refining workflows to meet specific operational constraints.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.