All Episodes
Displaying 21–40 of 60 in total
Key Phrase Extraction vs. Named Entity Recognition
Key phrases and named entities can come from the same document, but they answer different questions. This episode compares extracting central topics with identifying r...
Sentiment Analysis vs. Summarization
A long customer-support message may need both an attitude signal and a concise account of what happened. This episode explains why sentiment analysis and summarization...
Speech-to-Text and Text-to-Speech Fundamentals
A voice interaction often uses two opposite speech operations. Incoming audio must be recognized and converted into text before an application can work with the reques...
Computer Vision: Interpreting Existing Images
When an application receives an existing photograph and needs information about what the image contains, the workload is different from creating a new image or extract...
Image Generation: Creating New Visual Content
A written creative brief that asks for an image that does not already exist describes a generative visual workload. This episode explains how text-to-image generation ...
Content Understanding and Multimodal Information Extraction
Documents, photographs, audio recordings, and video can all contain information that a business needs in consistent fields. This episode introduces Azure Content Under...
Choosing the Right AI Capability Across Workloads
A project can require text summarization, speech transcription, visual analysis, image generation, and structured document extraction at the same time. This episode br...
Integrated AI Solution Design and Responsible AI Decisions
Real solutions often combine several AI workloads and several governance concerns. This episode uses a sensitive multimodal healthcare assistant to connect speech reco...
System Prompts vs. User Prompts
Stable application behavior and changing user requests belong in different parts of a prompt design. This episode explains how system prompts provide durable instructi...
Writing Effective Prompts and Resolving Conflicting Instructions
A model cannot reliably satisfy two instructions that directly contradict each other. This episode uses a support assistant with conflicting system rules to explain pr...
Deploying a Model from the Foundry Catalog
Finding a suitable model in the Microsoft Foundry catalog does not make that model callable by an application. This episode explains the difference between model selec...
Testing and Evaluating a Deployed Model
A deployment can be technically available and still fail the workload it was created to support. This episode explains how to test a deployed model in Microsoft Foundr...
Connecting Python Applications to Microsoft Foundry
A model that works in the Foundry portal can still fail from Python when the application uses the wrong endpoint, client, project path, or authentication. This episode...
Building a Lightweight Python Model Client
A simple application that sends one question to a deployed model does not require an agent or persistent conversation state. This episode walks through the conceptual ...
Managing Multi-Turn Chat and Conversation State
A chat interface can look continuous while the underlying requests remain unrelated. This episode explains how multi-turn conversation state preserves the context need...
Designing a Single-Agent Solution with Foundry Agent Service
An AI agent combines a model with instructions and, when needed, tools around a defined job. This episode introduces Microsoft Foundry Agent Service as a managed envir...
Configuring and Testing a Prompt Agent
Saving an agent configuration proves that the platform accepted the settings, not that the agent behaves correctly. This episode examines the three central parts of a ...
Selecting and Restricting Agent Tools
Giving an agent every available tool can create unnecessary access paths and make behavior harder to control. This episode explains how tools extend an agent beyond th...
Building a Python Client for an Existing Agent
A Python application that invokes an existing agent has a different job from the code or configuration that defines the agent itself. This episode explains how a clien...
Managing Agent Conversation State Across Turns
Multi-turn agent interactions require more than sending several messages to the same model. This episode explains how conversation state allows a later request to depe...