CertifyDeckCargando contenido…
The official Microsoft Certified: Multi-Agent AI Solutions Expert syllabus, organized by domain. Check off each point as you review it.
Inicia sesión para guardar tu progreso entre sesiones.
Design logical architecture for multi-agent solutions
Before creating agents, pick the least complex level: a direct model call when a prompt is enough, a single agent with tools as the usual default, and multiple agents only for security or compliance boundaries, multiple teams, planned growth, or when testing shows one agent is not enough. Fixed steps are modelled as deterministic workflows.
AI agent orchestration patterns ↗Single agent or multiple agents ↗
A multi-agent workflow combines patterns (sequential, concurrent, group chat, handoff and magentic) per stage. Control loops need acceptance criteria, an iteration cap and a fallback such as escalating to a person. Human-in-the-loop steps are defined as approval or feedback, optional or mandatory, and state should be persisted at those points so work resumes without being replayed.
AI agent orchestration patterns ↗Microsoft Agent Framework Workflows - Human-in-the-loop (HITL) ↗
Document each agent’s responsibilities, non-overlapping roles and prohibited actions in an agent charter. Instructions are treated as versioned code with identity and tone, scope and boundaries, tool mandates and citation rules. Autonomy grows gradually from no permitted actions, and boundaries are enforced with deterministic controls, not only with the system message.
Process to build agents across your organization ↗Secure autonomous agentic AI systems ↗
Each agent should have its own identity with minimal RBAC roles and explicitly allowed tools (allowed_tools, approvals for high-risk operations). In Foundry, the agent identity covers unattended access or different access levels per agent; the project managed identity gives all agents the same access, and OAuth identity passthrough preserves each user’s permissions.
Set up authentication for Model Context Protocol (MCP) tools ↗Agent identity concepts in Microsoft Foundry ↗
MCP connects each agent to its tools and data: the host creates one client per server, the server exposes tools, resources and prompts, and the transport is stdio locally or Streamable HTTP remotely. A2A connects autonomous agents to each other: they discover one another through the Agent Card at /.well-known/agent-card.json and receive updates by polling, streaming or push notifications to a webhook.
A2A and MCP: Detailed Comparison ↗MCP Architecture overview ↗
Responsible AI principles (fairness, reliability and safety, privacy and security, inclusiveness, transparency and accountability) are applied through the Govern, Map, Measure and Manage functions. To avoid overreliance, HAX-based design builds realistic mental models, signals when to verify and facilitates verification with sources. Guardrails apply to inputs, tools and final output.
Responsible AI for agent design ↗Fostering appropriate reliance on GenAI ↗
Short-term memory is the current session context and is managed by the agent framework; Foundry long-term memory extracts, consolidates and retrieves user profile, chat summary and procedural memories across sessions. Agents share only the context the next agent needs, compacted when useful, with per-user isolation, explicitly shared segments and retention policies.
Memory in Microsoft Foundry Agent Service ↗AI agent orchestration patterns ↗
Give each agent the model its task demands: small models for classification or extraction, non-reasoning models for fast replies, and reasoning models with higher effort for planning or research. Reasoning tokens occupy the context window, are billed as output and count toward max_output_tokens. Model router picks a model per request, and choices should be validated with evaluations.
Azure OpenAI reasoning models ↗GPT-5 vs GPT-4.1: choosing the right model for your use case ↗
Specify technology components to include in a multi-agent solution
In production, A2A connects agents that cross process, team or organizational boundaries and are discovered through their Agent Card; agent-as-tool is enough inside one process. Foundry toolboxes expose tools and MCP servers through one endpoint with centralized authentication and versioning, and Foundry IQ provides shared knowledge bases that respect user permissions.
Agent-to-Agent (A2A) ↗What is Microsoft Foundry Agent Service? - Microsoft Foundry ↗
Zero Trust for multi-agent systems: each published agent gets its own Entra identity with least-privilege RBAC roles, the token audience matches the target resource, and Conditional Access blocks by default and on risk. Private networking avoids public paths, and Purview and Foundry Control Plane apply DLP policies, auditing and compliance to agent interactions.
Agent identity concepts in Microsoft Foundry - Microsoft Foundry ↗Configure Microsoft Entra agent identities for increased security - Microsoft Entra ↗
Tiered persistence separates session state (AgentSession and history providers, such as Cosmos DB to resume on another instance), shared team state (workflow state with shared scopes and durable checkpoints), and long-term semantic memory (Foundry memory stores isolated per user scope, with TTL and item deletion).
Create and use memory in Foundry Agent Service - Microsoft Foundry ↗Azure Cosmos DB ↗
Prompt agents have no compute to manage; hosted agents run your container in per-session sandboxes with managed scaling, identity and endpoint, and their cost depends on per-session CPU and memory. The Durable Extension on Azure Functions adds serverless durable orchestration, and self-hosting (for example, on AKS) gives full control of networking and runtime.
Hosted agents in Foundry Agent Service - Microsoft Foundry ↗Durable Extension ↗
Multi-agent observability builds on OpenTelemetry with the GenAI conventions: nested invoke_agent spans, execute_tool spans with arguments and results, and W3C trace context propagated to MCP servers. Agent Framework exports traces to Application Insights; sensitive data is captured only in development, and checkpoints make complete workflow runs replayable.
Agent monitoring combines continuous evaluation on sampled traffic, scheduled evaluations to detect drift, red team scans and each agent's dashboard. Foundry Control Plane aggregates fleet health and cost, Insights finds recurring patterns in traces, and Azure Monitor alerts with action groups start automated remediation through Functions or Logic Apps.
Monitor agents with the Agent Monitoring Dashboard - Microsoft Foundry ↗Observability in Generative AI - Microsoft Foundry ↗
The development lifecycle relies on dev containers with the azd feature and its extensions, Foundry Toolkit for Visual Studio Code (Agent Inspector, tool catalog and deployment), the azd ai agent CLI to run, invoke and deploy, and AI instruction files (copilot-instructions.md, AGENTS.md, .instructions.md) together with the Microsoft Foundry Skill for coding agents.
Microsoft Foundry Toolkit for Visual Studio Code overview - Microsoft Foundry ↗Install the Azure Developer CLI ↗