Microsoft has announced the public preview of built-in Knowledge for Azure Logic Apps at Build 2026. The new capability turns your documents into a ready-to-use knowledge base that agents and workflows can query directly, without you building and operating a custom retrieval pipeline.
For teams that have stayed away from agent projects because of the ingestion, chunking, vector store, and search work involved, this lowers a real barrier. You upload files in common formats, including Word, PDF, PowerPoint, Excel, HTML, Markdown, and plain text, and the service handles parsing, summarising, vectorising, and semantic retrieval for you.
How we are responding
We are testing the preview in our Azure environment so we can recommend it where it fits our customers’ integrations and automations. The uses we are tracking first are the document-heavy internal workflows that still rely on someone reading the right file: support runbooks, HR or finance policies, product specifications, contracts, and procedure manuals.
We are watching the two SKU options closely. On Logic Apps Standard you bring your own Cosmos DB vector store and AI models, which suits organisations that already have those resources and want full control. On the Automation SKU the platform manages the store and models for you, so the path from documents to a working knowledge base is shorter. Before we propose either for production work, we are documenting the differences in pricing, data residency, model choice, and operational responsibility.
What it means for you
If your integrations or automations run on Azure Logic Apps, this gives you a way to build agentic workflows that answer questions grounded in your own documents rather than generic model training data. Authentication runs through Microsoft Entra ID, and managed identity is the recommended option so you are not manually rotating keys.
Because this is a public preview, we would not put it on a critical path yet. It is a good time to pick one contained internal workflow, such as helping staff find answers in policy documents, and trial it. Ask us for a quick assessment and we will look at whether your document set and Azure setup are a fit.
