RAG Without the Lag: Interactive Debugging for Retrieval-Augmented Generation Pipelines
Quentin Romero Lauro * 1 , Shreya Shankar * 2 , Sepanta Zeighami 2 , Aditya G. Parameswaran 2 · 2025
Read RAG Without the Lag: Interactive Debugging for Retrieval-Augmented Generation Pipelines by Quentin Romero Lauro * 1 , Shreya Shankar * 2 , Sepanta Zeighami 2 , Aditya G. Parameswaran 2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-13587.txt.
Description
Quentin Romero Lauro, Shreya Shankar, Sepanta Zeighami, and Aditya Parameswaran present RAGGY, a 2025 tool for debugging retrieval-augmented generation pipelines without the wait. Such pipelines fetch facts, then feed them to a language model. When an answer goes wrong, builders struggle to tell whether fetching or writing failed, and each tweak can cost hours of reprocessing.
RAGGY pairs a Python library of snap-together pipeline parts with a live interface for testing changes at once. The authors also watched twelve engineers debug with it, harvesting patterns of expert practice. Their findings feed design advice for the next wave of tools. Faster feedback loops could make these pipelines far easier to tune, a quiet bottleneck in applied AI work.
Catalog record
| CREATOR | Quentin Romero Lauro * 1 , Shreya Shankar * 2 , Sepanta Zeighami 2 , Aditya G. Parameswaran 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-13587