Adaptive Applications: Separating App Needs from Environment Provisioning with Radius

A discussion with Hishi and host Aaron Stark about adaptive applications, an architecture pattern that separates what an application needs from how different environments provide it. It covers why teams may need to run applications beyond their original cloud environment and how capability-oriented abstractions can help address sovereignty, data residency, compliance, intermittent connectivity, performance, and special hardware needs.

■ Core concepts
- Motivation for adaptive applications and business continuity across environments
- Capability-oriented thinking for AI inference, database caching, and identity
- Platform-agnostic contracts versus environment-specific implementations

■ Radius and deployment
- Using Radius resource types and recipes to map capabilities to Azure, local, or other environments
- Examples involving ARM templates, Terraform templates, OpenAI-compatible endpoints, and local language models
- Single-command deployment workflow for the sample application

■ Demo and modernization
- A stock trading simulator with web front end, backend in a different language, messaging pipeline, and chat agent
- Identity provider choices such as Keycloak and federation with local Active Directory; mention of Entra ID
- AI-assisted analysis of existing codebases to identify gaps and generate artifacts, with community contributions and additional skills sought

■ Resources and adoption
- Pattern documentation published in the Azure Architecture Center and code, samples, and documents in a public repository
- Adoption is presented as a pattern that can be used without locking into a specific toolchain
- Next steps include reviewing the documentation and trying the streamlined sample experience

It is suited for developers, platform teams, and architects evaluating how to build or modernize applications that must run across Azure, on-premises, or edge environments. Viewers can gain an understanding of capability abstractions, environment-specific recipes, and tooling for modernization, along with where to find samples and guidance to experiment.