How IskayPet unified corporate knowledge with Amazon Bedrock Managed Knowledge Bases
aws ·
IskayPet, a leading pet care company in Iberia, implemented a corporate knowledge assistant using Amazon Bedrock Managed Knowledge Bases to solve challenges with scattered operational information. The solution integrates directly with existing Microsoft SharePoint repositories, eliminating the need for data migration or self-managed RAG infrastructure like vector databases. By leveraging hybrid search and managed reranking, the system accurately retrieves relevant passages from documents across four languages and three countries while enforcing document-level access controls. The entire application, including the frontend and AWS architecture, was developed rapidly using Kiro and Claude Haiku 4.5 on Amazon Bedrock. This approach allows each business area to have isolated knowledge bases, ensuring that employees only see information they are authorized to access. Production results indicate a significant reduction in search time, dropping from several minutes to approximately seven seconds per query. Additionally, the company reports a 95% decrease in time spent searching for procedures, enhancing overall employee productivity. The architecture supports scaling to handle roughly 100,000 queries monthly without requiring re-architecting as new areas or countries are added.
aws