At Build 2026, Microsoft unveiled Azure Logic Apps Automation, a new SKU that ships as a managed SaaS experience at auto.azure.com. Sign in, and the compute, connectors, model endpoints, and knowledge services are already there. No assembly required. The pitch: let business teams build production-grade automations without needing a dedicated integration developer, and do it without giving up the security and governance controls that got Azure into the enterprise in the first place. It's in public preview now.
Every team has an AI agent demo. Few have one in production. Logic Apps Automation closes that gap with the security, identity, networking, and observability you need to ship, built in, not bolted on.
The gap they're targeting is real. Building production-grade automations on Azure today typically requires assembling compute, connectors, identity, networking, model endpoints, and knowledge-retrieval infrastructure separately, and then governing the result. Logic Apps Automation collapses that into a single managed environment. Every project gets an isolated compute boundary. Workflows, data, and agents never share a runtime with other tenants. VNET integration and private endpoints are available for reaching internal systems without exposing them to the internet. Identity, RBAC, audit logging, and policy controls are on by default.
Agents are integrated through three distinct patterns. Agent-loop orchestration carries forward the existing model where Logic Apps actions serve as callable tools inside an agent loop, the same pattern covered in InfoQ's recent piece on Logic Apps code interpreters. Foundry agent integration lets workflows invoke Microsoft Foundry Hosted or Prompt Agents directly from the canvas: the platform handles the wiring, the workflow calls the agent, gets results, and continues. The third option is a managed sandbox for agent harnesses: teams can bring a well-known harness like GitHub Copilot and run it in an isolated, managed sandbox.
The most architecturally significant addition is Knowledge as a Service (KBaaS). On the Automation SKU, the platform provisions and manages the vector store and AI models entirely. There is no Cosmos DB to deploy, no embeddings model to configure, and no connections to set up. Teams upload documents, attach the knowledge base to their agent, and the platform handles ingestion, chunking, embedding, and retrieval. This is a fully managed RAG pipeline with zero infrastructure to operate. On the Standard SKU, the same capability works with customer-owned Cosmos DB and AI model resources for teams that need full control over their data and models.
Sonny Gillissen, Azure MVP, and an early adopter, noted:
With the power of AI, automations just got on steroids! Simply tell it what you need, explain the intent, et voilà!
In a detailed blog post, Gillissen walked through the experience of building automations using natural language, noting that the built-in AI assistant is the primary interface change compared to the existing Logic Apps designer.
Alongside Logic Apps Automation, Microsoft announced that the Logic Apps MCP Server has reached general availability. The MCP Server exposes existing Logic Apps workflows as MCP-compatible tools that agents can discover and invoke directly. This means organizations can reuse years of existing automation investments without building custom APIs. Workflows that already connect enterprise systems and business processes become AI-callable capabilities in minutes.
The Build 2026 announcements also include Codeful Workflows, a code-first development experience built on the Logic Apps Standard SDK. Teams that prefer writing .NET code over visual design can now create workflows directly in code, giving organizations the choice among a low-code canvas, a natural-language AI assistant, and a pro-code SDK within the same platform.
The positioning against competitors is clearer now. Power Automate targets individual productivity automation within Microsoft 365. Logic Apps Standard targets integration developers building complex enterprise workflows. Logic Apps Automation targets the gap between them: teams that need enterprise-grade infrastructure but want a SaaS-like experience with AI-assisted creation. The three agent integration patterns (agent loop, Foundry, managed sandbox) and Knowledge as a Service make it the most complete agent-ready integration platform in Azure's portfolio.
Logic Apps Automation is available in public preview at auto.azure.com. The Build 2026 announcement session provides a walkthrough of the full experience.