Text decision tree
Choose between existing Copilot features, local inference, prebuilt AI services, Azure Machine Learning and Microsoft Foundry.
Adapted from Chapter 20 of Decision Trees: Frameworks for Technology Choices. Sources reviewed 2026-11-05.
START
Q0 Build confirmed after checking existing apps?
YES -> Build is justified. Continue to workload selection. Then Q1.
NO -> Otherwise return to the application and agent choices in chapters nineteen and twenty-one. Then STOP.
Q1 Inference must run locally or disconnected?
YES -> Foundry Local. Evaluate Azure Local for site deployment. Then STOP.
NO -> Otherwise continue to the perception question. Then Q2.
Q2 A prebuilt perception API meets the need?
YES -> Foundry Tools. OCR, document, speech, vision or translation. Then STOP.
NO -> Otherwise check whether you need to train or own a model. Then Q3.
Q3 Need custom model training and an ML lifecycle?
YES -> Azure Machine Learning. Also compare the existing Fabric or Databricks plane. Then STOP.
NO -> Otherwise proceed to foundation-model work. Then Q4.
Q4 Otherwise: foundation-model work
DEFAULT -> Microsoft Foundry. Then apply the deployment decision tree. Then END.
END: This selects the platform. The Foundry deployment tree covers geography, capacity, model source, tuning and migration.
AI-generated chapter illustration
BookWriter-generated chapter 20 illustration
This AI-generated illustration gives a visual overview. It can simplify labels or imply connections that do not preserve every decision. For exact branches, prerequisites and stop conditions, follow the video graph and the text tree above.
Reference links
- September Foundry update — 2026-10-16.
- Microsoft Copilot harness in Foundry Agent Service — 2026-10-28.
- Hosted-agent network egress controls — 2026-09-24.