AI is entering institutional repositories — not just for search, but for scholarly discovery.
A 4 September 2026 arXiv preprint in Digital Libraries describes *Westlake Scholar*, an open-source AI platform added to an institutional repository.
It provides four services:
1. *Contextual paper reading* — helps researchers understand papers in context.
2. *Research-direction-guided discovery* — finds relevant research based on the direction a researcher is exploring.
3. *Publication-grounded expert discovery* — identifies researchers based on their actual publications.
4. *AI-generated research chronologies* — helps trace how a research area has developed over time.
What I find particularly interesting is the shared *institutional knowledge layer*. It connects approved records, full-text documents, and scholar–publication relationships.
The system also includes provenance, human review and institutional governance. In other words, the AI layer is built around the repository’s trusted scholarly records rather than operating as a separate search tool.
Westlake University has reportedly had the system live since April 2026.
This points to an interesting direction for institutional repositories: AI can sit on top of trusted academic collections and help researchers discover not only *what has been published*, but also *how ideas, researchers and research directions connect*.
Source: arxiv.org/abs/2609.0507234600d