Orcheo Paper Accepted to SIGIR 2026 Resource Track
Our Orcheo research paper has been accepted to the SIGIR 2026 Resource Track, highlighting modular workflow graphs, reproducible evaluation, and deployment-ready orchestration for conversational search systems.
We are pleased to share that our research on Orcheo, the open-source AI workflow orchestration platform, has been accepted to the SIGIR 2026 Resource Track.
The paper (authored by Shaojie Jiang, Svitlana Vakulenko, Maarten de Rijke) shows how complex conversational search systems can be built as modular, versioned workflows rather than brittle end-to-end scripts. This improves reproducibility, makes component-level experimentation easier, and creates a smoother path from local development to production deployment.
The work reflects a broader goal behind Orcheo: making AI automation, from quick vibe-coded workflows to production pipelines, more structured, reusable, and secure.
Key ideas highlighted in the paper:
- modular workflow graphs instead of brittle scripts,
- reproducible evaluation and component-level experimentation,
- smooth transition from local development to production deployment, and
- full-stack orchestration for agent-based systems.
Read the paper:
Explore the demos and documentation: