Azure teams were seeing strong interest in Azure OpenAI, but most developers struggled with where to start.
Standing up an AI workload required stitching together AKS, Azure OpenAI, App Services, data stores, and queues — too many manual steps across services. Documentation lived outside the product, so developers couldn't learn by doing.
And the AI-specific primitives — models, deployments, quotas — were unfamiliar to most Azure developers, with no in-product onboarding.
I led design for a new App Creation Flow that bundled recommended services with a one-click sample app.
We redesigned the Recommended Services feature so suggestions dynamically reflected project context — surfacing Key Vault, App Configuration, and SQL Database at the right time during setup.
Recommendations shifted from static lists to context-aware suggestions, making them feel native rather than bolted on.
Shipped two paths: "Try a sample application" — a quick-start that bundles Azure OpenAI with supporting services and deploys a working AI app with a single click — and "Deploy your application" — a custom setup path for developers who already have code and need to configure their own service stack.
Reduced AKS app deployment from hours to minutes, roughly 80–95% faster. Shipped to the Azure Portal in 2025 and became the reference pattern for guided experiences across the portal.
