RESEARCH
Research what actually matters.
Evidence-based research on Microsoft Data & AI, turned into practical architecture decisions. Every topic is graded for confidence, kept fresh against official Microsoft sources, and feeds the technology explainers, use cases, labs and podcasts across TechExplained.
- Research topics
- 9
- Technologies
- 7
- Research lenses
- 8
From research to real-world decisions
The same three steps behind every topic. No opinions dressed up as facts.
Evidence
We start from primary Microsoft sources: Microsoft Learn, the Azure Architecture Center, the Well-Architected Framework and official documentation. Blogs and forums never replace an official source.
Validation
Every claim is checked, dated and graded for confidence. Where evidence is thin or the topic moves fast, we say so. Uncertainty is labelled, never hidden.
Decision
Findings become architecture guidance: what to choose, what to avoid, and which trade-offs to weigh for security, cost, governance and adoption.
One research topic. Multiple ways to learn.
Each research record is reusable. The same evidence shows up wherever you prefer to learn.
Read the research
The full record: evidence, technical context, implications and references you can verify.
Browse topicsSee it applied
Reference architectures and real Data & AI decisions that put the research to work.
Use casesPractice the decision
Interactive scenarios and simulations in the Architecture Lab, built on the same trade-offs.
Architecture LabHear the reasoning
Podcast episodes that talk through the architecture thinking behind the research.
PodcastsResearch is the foundation, not a footnote
It sits at the start of the learning flow and connects every other experience.
Explore the research
Filter by domain, technology or lens.
9 topics
Data & AI Cloud Security
How to secure an enterprise AI platform on Azure: identity, data isolation, prompt injection, excessive agency, and monitoring, using controls from Purview, Defender for Cloud, and Sentinel.
AI Agent Cost Management
Why AI agents quietly multiply your cost, which variables drive token consumption, and how to move from cost per token to cost per successful outcome.
Microsoft Fabric Capacity & Cost
How capacity, Capacity Units, OneLake storage, and transactions together determine the Fabric bill, and which architecture decisions drive cost.
Microsoft Foundry Agent Architecture
Foundry as a platform for the full lifecycle of AI apps and agents, not just model access: resources, projects, connected services, and the Agent Service.
Microsoft Foundry: The AI App & Agent Platform
Foundry as the enterprise platform to build, ground, observe, govern, and optimize AI apps and agents: resource and projects, Agent Service, model router, Foundry IQ for RAG, control plane, and observability.
Microsoft Purview AI Governance
Purview as the cross-cutting layer for data security, compliance, and AI governance, from sensitivity labels and DLP to DSPM for AI, spanning Copilot and third-party AI apps.
Enterprise AI Adoption
What holds organizations back from moving AI from experiment to production, with evidence that is deliberately limited: a small practitioner study rather than a universal statistic.
Microsoft Fabric vs Azure Databricks
An honest comparison on the axes that really differ: integration versus engineering control, capacity versus DBUs, and governance via Purview versus Unity Catalog.
Azure Databricks: Architecture & Cost
How account, workspace, Unity Catalog, control plane, and compute plane fit together, and why the Databricks bill is architecture-driven, not only DBU-driven.