🔒 Guided

Pre-launch preview. Authorised access only.

Incorrect code

Guided by A Guide to Cloud
Explore AB-900 AI-901 aws-aif-c01
Guided AB-620 Domain 2
Domain 2 — Module 2 of 12 17%
12 of 28 overall

AB-620 Study Guide

Domain 1: Plan and Configure Agent Solutions

  • Getting Started: Copilot Studio for Developers Free
  • Planning Enterprise Integration and Reusable Components Free
  • Identity Strategy for Agents Free
  • Channels, Deployment and Audience Design Free
  • Responsible AI and Security Governance Free
  • Agent Flows: Build, Monitor and Handle Errors Free
  • Human-in-the-Loop Agent Flows Free
  • Topics, Tools and Variables Free
  • Advanced Responses: Custom Prompts and Generative Answers Free
  • API Calls, HTTP Requests and Adaptive Cards Free

Domain 2: Integrate and Extend Agents in Copilot Studio

  • Enterprise Knowledge Sources: The Big Picture
  • Copilot Connectors and Power Platform Connectors
  • Azure AI Search as a Knowledge Source
  • Adding Tools: Custom Connectors and REST APIs
  • MCP Tools: Model Context Protocol in Action
  • Computer Use: Agent-Driven UI Automation
  • Multi-Agent Solutions: Design and Agent Reuse
  • Integrating Foundry Agents
  • Fabric Data Agents: Analytics Meets AI
  • A2A Protocol: Cross-Platform Agent Collaboration
  • Grounded Answers: Azure AI Search with Foundry
  • Foundry Model Catalog and Application Insights

Domain 3: Test and Manage Agents

  • Test Sets & Evaluation Methods
  • Reviewing Results & Tuning Performance
  • Solutions & Environment Variables
  • Power Platform Pipelines for Agent ALM
  • Agent Lifecycle: From Dev to Production
  • Exam Prep: Diagnostic Review

AB-620 Study Guide

Domain 1: Plan and Configure Agent Solutions

  • Getting Started: Copilot Studio for Developers Free
  • Planning Enterprise Integration and Reusable Components Free
  • Identity Strategy for Agents Free
  • Channels, Deployment and Audience Design Free
  • Responsible AI and Security Governance Free
  • Agent Flows: Build, Monitor and Handle Errors Free
  • Human-in-the-Loop Agent Flows Free
  • Topics, Tools and Variables Free
  • Advanced Responses: Custom Prompts and Generative Answers Free
  • API Calls, HTTP Requests and Adaptive Cards Free

Domain 2: Integrate and Extend Agents in Copilot Studio

  • Enterprise Knowledge Sources: The Big Picture
  • Copilot Connectors and Power Platform Connectors
  • Azure AI Search as a Knowledge Source
  • Adding Tools: Custom Connectors and REST APIs
  • MCP Tools: Model Context Protocol in Action
  • Computer Use: Agent-Driven UI Automation
  • Multi-Agent Solutions: Design and Agent Reuse
  • Integrating Foundry Agents
  • Fabric Data Agents: Analytics Meets AI
  • A2A Protocol: Cross-Platform Agent Collaboration
  • Grounded Answers: Azure AI Search with Foundry
  • Foundry Model Catalog and Application Insights

Domain 3: Test and Manage Agents

  • Test Sets & Evaluation Methods
  • Reviewing Results & Tuning Performance
  • Solutions & Environment Variables
  • Power Platform Pipelines for Agent ALM
  • Agent Lifecycle: From Dev to Production
  • Exam Prep: Diagnostic Review
Domain 2: Integrate and Extend Agents in Copilot Studio Premium ⏱ ~13 min read

Copilot Connectors and Power Platform Connectors

Graph-indexed search and action-based Power Platform integration.

Two connector families, two completely different jobs

☕ Simple explanation

Think of it like a library vs a phone call.

Copilot connectors work like a library card catalogue — they pre-index external content (ServiceNow articles, Confluence pages, file shares) into Microsoft 365 so your agent can search it. About 50 connectors are available, and the M365 admin sets them up.

Power Platform connectors work like making a phone call — your agent dials an external service (Salesforce, SAP, Jira) in real time, asks a specific question, and gets a live answer back. Over 1,000 connectors are available, and you as the developer configure them inside Copilot Studio.

One gives your agent a searchable brain. The other gives your agent hands to reach out and grab live data.

Copilot connectors (formerly Microsoft Graph connectors) are an admin-configured pipeline that crawls external data sources and indexes content into the Microsoft 365 Search index. At runtime, the agent’s generative answers capability queries this index — the agent does not call the external system directly. Approximately 50 connectors are available covering sources like ServiceNow, Salesforce Knowledge, Confluence, Oracle, file shares, and databases. Configuration happens in the M365 admin centre and requires Search Administrator or Global Administrator role.

Power Platform connectors are developer-configured integrations that make live API calls at runtime. The agent invokes a connector action (e.g., “Get record”, “Create item”) and receives structured JSON. Over 1,000 connectors exist, covering virtually every SaaS platform. Developers add these directly in Copilot Studio as agent tools. Authentication is configured per connection (OAuth 2.0, API key, or service principal).

The fundamental distinction: Copilot connectors provide searchable knowledge (the agent finds information). Power Platform connectors provide actionable tools (the agent does something).

Copilot connectors vs Power Platform connectors

This table is critical for the exam. When a scenario describes how data reaches the agent, you need to instantly identify which connector family is in play.

Copilot connectors vs Power Platform connectors — know the difference
FeatureHow data flowsNumber availableWho sets it upAuth modelBest for
Copilot connectorsCrawl external source → index into M365 Graph → agent searches index via generative answers~50 (ServiceNow, Confluence, Salesforce KB, Oracle, file shares, databases, etc.)M365 admin in the admin centre (Search Admin or Global Admin role)Tenant-level — Graph permissions and connector-specific credentialsSurfacing external documents and KB articles as searchable knowledge alongside M365 content
Power Platform connectorsAgent calls connector action at runtime → live API request → structured JSON response1,000+ (Salesforce, SAP, Jira, ServiceNow, custom APIs, SQL, HTTP, etc.)Developer inside Copilot Studio (adds as an agent tool)Connection-level — OAuth 2.0, API key, or service principal per connectionReal-time lookups, write-back operations, and calling external business logic

Copilot connectors deep dive

Copilot connectors bring external content into the Microsoft 365 search ecosystem. Once indexed, the content is searchable by M365 Copilot, SharePoint search, and Copilot Studio agents.

How the pipeline works:

  1. Admin configures the connector in the M365 admin centre — specifies the external data source, credentials, crawl schedule, and schema mapping.
  2. Connector crawls the external system on schedule (frequency depends on the connector — typically every few hours).
  3. Content is indexed into the Microsoft 365 Search index as external items with properties, content, and access control lists (ACLs).
  4. Agent queries the index via generative answers — the user’s question is matched against indexed content, and relevant excerpts are used to generate a grounded response.
💡 Key detail: item capacity and schema

Each Copilot connector connection can index up to 5 million items (depending on your M365 licence tier). The schema defines which properties are searchable, queryable, and retrievable. Getting the schema right is critical — if a field is not marked as searchable, the agent cannot find content based on that field. The exam may test whether you know that schema configuration affects search behaviour.

Power Platform connectors deep dive

Power Platform connectors are the action layer. They do not store or index anything — they call external APIs in real time and return structured data.

How they work in Copilot Studio:

  1. Developer adds a connector as a tool in the agent’s configuration. You select the connector and choose which actions to expose (e.g., “Get a record”, “List items”, “Create entry”).
  2. Developer configures inputs — map the agent’s conversation context (variables, slot values) to the connector action’s input parameters.
  3. At runtime, when the orchestrator determines the tool is needed, it calls the connector action, passes inputs, and receives a JSON response.
  4. Agent uses the response in the conversation — either directly in a message or as input to further logic.
💡 Connection vs connector — know the difference

A connector is the definition (the API shape, actions, triggers). A connection is an authenticated instance of that connector for a specific user or service account. One connector can have many connections. When the exam asks about “configuring a connection,” they mean setting up authentication — not building the connector itself.

The developer’s workflow for each type

StepCopilot connectorsPower Platform connectors
1. PlanIdentify external content to surface as knowledgeIdentify external actions the agent needs to perform
2. ConfigureM365 admin centre → Connectors → set up source, schema, crawlCopilot Studio → Tools → add connector → select actions
3. AuthenticateAdmin provides source credentials (service account)Developer creates a connection (OAuth, API key)
4. MapDefine schema: searchable, queryable, retrievable fieldsMap input/output parameters to agent variables
5. TestSearch for indexed content in M365 Search portalTest the tool in the Copilot Studio test pane
6. MonitorCheck crawl status and error logs in admin centreCheck connector run history in Power Platform admin centre
Scenario: Kai wires up ServiceNow and Salesforce for Pacific Mutual

Kai has two integration tasks for Pacific Mutual’s agent:

Task 1 — ServiceNow Knowledge Base (Copilot connector): Pacific Mutual’s IT team maintains 5,000 knowledge articles in ServiceNow. Kai works with the M365 admin to set up the ServiceNow Copilot connector. The admin configures the connector in the M365 admin centre, provides ServiceNow service account credentials, maps the article schema (title, body, category as searchable; article ID as retrievable), and sets a 4-hour crawl schedule. After the first crawl completes, Kai tests: “How do I reset my VPN password?” — the agent returns a grounded answer citing the ServiceNow KB article.

Task 2 — Salesforce CRM (Power Platform connector): Insurance agents need live policy lookups. Kai adds the Salesforce connector as a tool in Copilot Studio, selects the “Get Record” action, configures OAuth 2.0 authentication with a Salesforce service account, and maps the policy number from the conversation to the record ID input. At runtime, when a user asks “What is the status of policy PM-2024-7891?”, the agent calls Salesforce live and returns the current policy status, premium amount, and renewal date.

Two sources. Two connector types. Same agent.

Question

How many Copilot connectors are available vs Power Platform connectors?

Click or press Enter to reveal answer

Answer

Approximately 50 Copilot connectors vs over 1,000 Power Platform connectors. Copilot connectors are curated for search indexing. Power Platform connectors cover a much broader range of API-based integrations.

Click to flip back

Question

What role is required to configure a Copilot connector in the M365 admin centre?

Click or press Enter to reveal answer

Answer

Search Administrator or Global Administrator. Copilot connectors are admin-configured at the tenant level, not developer-configured in Copilot Studio.

Click to flip back

Question

What is the difference between a connector and a connection?

Click or press Enter to reveal answer

Answer

A connector is the API definition (shape, actions, triggers). A connection is an authenticated instance of that connector. One connector can have multiple connections with different credentials.

Click to flip back

Question

How does a Copilot connector make external content searchable by the agent?

Click or press Enter to reveal answer

Answer

It crawls the external source on a schedule and indexes content into the M365 Search index. At runtime, the agent's generative answers capability queries this index — the agent never calls the external system directly.

Click to flip back

Knowledge Check

Kai wants Pacific Mutual's agent to search ServiceNow knowledge articles. The articles rarely change (updated weekly). Which approach is correct?

Knowledge Check

A developer adds a Salesforce Power Platform connector to their agent. What happens at runtime when a user asks for policy details?

Knowledge Check

Which statement correctly describes the authentication difference between the two connector types?

🎬 Video coming soon

Copilot Connectors and Power Platform Connectors

← Previous

Enterprise Knowledge Sources: The Big Picture

Next →

Azure AI Search as a Knowledge Source

Guided

I learn, I simplify, I share.

A Guide to Cloud YouTube Feedback

© 2026 Sutheesh. All rights reserved.

Guided is an independent study resource and is not affiliated with, endorsed by, or officially connected to Microsoft. Microsoft, Azure, and related trademarks are property of Microsoft Corporation. Always verify information against Microsoft Learn.