IskayPet built an internal assistant using documents in Microsoft SharePoint
IskayPet built a knowledge assistant for employees in three countries using Amazon Bedrock Managed Knowledge Bases. According to the case study, it responds in four languages, cites sources and searches only documents accessible to the individual employee.
According to a case study on AWS Machine Learning Blog, IskayPet built an internal knowledge assistant using Amazon Bedrock Managed Knowledge Bases. It helps employees find answers in hundreds of documents in Microsoft SharePoint covering store operations, veterinary care, employee onboarding and purchasing. The solution serves operations in Spain, Portugal and France and works in Spanish, Portuguese, French and English.
The team used the native connector for Microsoft SharePoint and chose a separate knowledge base for each assistant, limited to the relevant sites, libraries or folders. It also enabled processing of visual content in documents to make images and scanned materials searchable. According to the article, permissions are retrieved from Microsoft SharePoint at every access check; the assistant searches only permitted documents and responds in the employee’s language with links to sources.
According to the case study, the team built the user interface, cloud architecture and deployment using Kiro and Claude Haiku 4.5 in Amazon Bedrock. The available description in the article does not name the specific model used to generate the assistant’s answers. The team is also adding Amazon Bedrock Guardrails to filter the content of responses. Details are available in the source article.
Why it matters
The case study shows a concrete approach to making company documentation accessible through natural-language questions while respecting existing access permissions. For teams preparing a similar solution, the key elements are the connector to the existing repository, the separation of knowledge bases by content and the separation of retrieval from the model that generates answers.
Two audiences, two different impacts
What this means
For individuals
Employees at IskayPet can ask operational questions in their own language and follow a link from the answer to the relevant source document. Their permissions continue to determine the scope of information available to them.
For a business
According to the case study, the managed infrastructure freed the team from having to operate a vector store and document processing separately. Expansion into another area or country is handled by adding an assistant connected to the relevant content.
ProcessesCheck the original
Event sources
only one source so far · 1 publisher, 0 independent. We count feeds from the same owner only once.