Our Conversational Search Paper Is Now on arXiv

Our new paper, “Orcheo: A Modular Full-Stack Platform for Conversational Search,” is now available on arXiv:

• https://arxiv.org/abs/2602.14710

The paper shows how Orcheo helps teams build and evolve conversational search systems as modular workflow graphs instead of brittle end-to-end scripts. It focuses on three practical outcomes:

• cleaner experiment iteration through component-level swaps,
• stronger reproducibility through explicit workflow structure, and
• a smoother path from local prototypes to production deployments.

Authors and affiliations:

• Shaojie Jiang - AI Colleagues (AIC), University of Amsterdam
• Svitlana Vakulenko - WU Vienna University of Economics and Business
• Maarten de Rijke - University of Amsterdam

If you want a quick walkthrough before reading the full paper, see our companion post:

• Applying Orcheo to Conversational Search: Key Insights from Our New Paper

We hope this release is useful to both researchers and product teams building reliable LLM-powered search experiences.