A short customer message can contain several kinds of information: a person’s name, product identifier, date, opinion and request for action. AI-901 tests recognition of text-analysis worklo
Microsoft AI-901: Create and Test One Foundry Prompt Agent
An agent is not simply a more complicated word for chatbot. A basic prompt agent in Microsoft Foundry combines a selected model, instructions, optional tools and a maintained identity in a s
Microsoft AI-901: Build a Lightweight Chat App with the Foundry SDK
A working chat client is a useful AI-901 exercise because it connects model concepts to an application boundary. The client must identify an approved model deployment , authenticate, su
Microsoft AI-901: Foundry Model Deployments and Prompt Testing
Microsoft Foundry can make a foundation model available for a lightweight application, but a successful prompt in a portal playground is not the same as a production-ready service. AI-9
Microsoft AI-901: AI Workloads, Models and Deployment Choices
A common AI fundamentals mistake is starting with a fashionable model instead of the problem. An organization may need a number predicted from historical data, a short message classifie
Microsoft AI-901: Responsible AI Decisions for Azure Solutions
Responsible AI is easier to understand when it becomes a series of design decisions rather than a slogan attached to a product. A recommendation system can be statistically accurate but
Microsoft AI-901: New Exam Blueprint and AI-900 Transition
The Microsoft Azure AI Fundamentals certification now uses Exam AI-901 , not the retired AI-900 exam. That change matters more than a new number on a booking page. AI-901 asks beginning AI d
Amazon AWS AIP-C01: Bedrock Tracing and GenAI Release Gates
Diagnose GenAI failures using Bedrock and AgentCore traces, retrieval provenance, model and prompt versioning, evaluation and controlled release gates.
Amazon AWS AIP-C01: Bedrock Streaming, Retries and Timeouts
Handle Amazon Bedrock ConverseStream errors, SDK retries, partial outputs, cancellations and side-effect safety with explicit completion and timeout states.
Amazon AWS AIP-C01: AgentCore MCP Identity and Safe Tool Actions
Design safe AgentCore MCP tool integrations using bounded schemas, caller and target identity, backend authorization, approval and idempotent actions.
Amazon AWS AIP-C01: Bedrock Prompt Versions and JSON Outputs
Use Amazon Bedrock Prompt Management, versioned prompt contracts and structured outputs without confusing valid JSON with verified business facts.
Amazon AWS AIP-C01: Bedrock Knowledge Base Freshness and Access
Keep Bedrock Knowledge Bases current and permission-aware through source versioning, synchronization, verified identity, ACL filtering and negative retrieval tests.
Amazon AWS AIP-C01: Bedrock Multimodal Data Pipelines
Build reliable multimodal Bedrock input pipelines by validating files, choosing visual versus textual processing, preserving provenance and diagnosing ingestion failures.
Amazon AWS AIP-C01: Bedrock Model Routing and Failover
How Amazon Bedrock intelligent routing, cross-Region inference and application-managed fallback differ—and how to keep model changes, data boundaries and output quality safe.
Microsoft AB-100: Business Agent Governance, DLP and Identity
Governing a business agent requires controls that remain effective when its prompt, retrieved evidence or tool output is misleading. A supplier document might contain a […]