IBA - Innovative Business Practices
TR/EN
Copilot TrainingTraining

Copilot Training

These trainings answer two needs. The Microsoft 365 Copilot training helps every employee use Copilot in the apps they open every day. The Copilot Studio trainings are for teams moving from using a ready-made Copilot product to designing and governing their own agents. The shortest program runs one day, the longest six.

Copilot Training

Trainings in This Category

Select a training to see its agenda, project cases and technical requirements.

Office 365 & Microsoft Copilot

Saving Time — Everyday Productivity Training

01Training Overview

Duration
1 Day
Training Hours
09:30 – 16:30
Maximum Participants
20 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants will receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — based on the organisation's preference.
Prerequisites
None. No technical background is needed — suitable for all employees who use Microsoft 365.
Training Objective

This training aims to save every knowledge worker real time each week. The programme covers, hands-on, how Microsoft 365 Copilot works and where it applies, drafting and summarising emails in Outlook, producing meeting summaries and action lists in Teams, drafting and designing in Word and PowerPoint, generating analysis and insights in Excel, and prompting best practices for everyday productivity.

The training runs through short, independent scenarios across the five Microsoft 365 apps participants use every day — rather than a single project, it covers practical exercises spanning the whole daily workflow.

By the end of the training, participants use Copilot correctly inside the Outlook, Teams, Word, PowerPoint and Excel apps they use every day, saving real time each week.

Technical Content

Microsoft 365 Copilot Fundamentals

  • How Copilot works: generating responses based on email, calendar, file and chat data through Microsoft Graph
  • Organisation-wide Q&A and general-purpose chat with Copilot Chat
  • Basic awareness of what data Copilot can access and how permissions work
  • Creating skills and using them within Office 365

Outlook Copilot

  • Summarising and categorising unread emails by importance and urgency
  • Summarising long email threads with multiple replies in a single paragraph
  • Generating draft replies that reference previous correspondence and attached files
  • Rewriting emails in a chosen tone (formal, concise, persuasive, etc.) and adjusting length
  • Grouping the inbox by topic/project and suggesting a priority order
  • Getting Copilot coaching feedback on the tone and clarity of a draft email

Teams Copilot

  • Real-time conversation summary and tracking of key points during a live meeting
  • Generating an automatic post-meeting summary, decisions made and a list of action items
  • Quickly catching up on a missed meeting with "summarise what was discussed"
  • Summarising the discussion on a specific topic from chat history
  • Asking questions about earlier discussions during a meeting ("what was decided on this before?")
  • Generating a draft agenda for the next meeting from previous meeting notes

Word & PowerPoint Copilot

  • Word: generating a draft document/report from scratch from a prompt or source file
  • Word: rewriting, shortening/lengthening and changing the tone of existing text
  • PowerPoint: generating a complete presentation from a Word document or a prompt
  • PowerPoint: redesigning existing slides with Designer and getting image suggestions
  • PowerPoint: summarising speaker notes and getting presentation rehearsal feedback

Excel Copilot

  • Producing analysis by querying a data table in natural language, without writing formulas
  • Automatically detecting and summarising trends, outliers and patterns
  • Getting formula suggestions from natural language and explanations of existing formulas
  • Chart/visualisation suggestions suited to the data, with automatic creation
  • Quickly building models for simple "what-if" scenarios
  • Data cleansing suggestions: detecting inconsistencies, blank cells and duplicate records
  • A use case for a skill created within Excel

02Exercises and Project Scenarios

Scenario 1

Email Drafting and Summarising with Outlook

Covers summarising a long email thread and preparing a draft reply with Copilot.

Scenario 2

Meeting Summary with Teams

Covers producing an automatic summary and action list from a meeting recording.

Scenario 3

Drafting with Word & PowerPoint

Covers quickly drafting a document and turning it into a presentation.

Scenario 4

Analysis and Insights with Excel

Covers analysing a data table with Copilot and generating summary insights.

Together, the four scenarios show how Copilot fits into the whole daily workflow — correspondence, meetings, documents and data. The four scenarios below are independent of each other and all are covered in the training; unlike the "choose one case and grow it" model of the Power Platform trainings, this training is delivered separately across five Microsoft 365 apps because it needs to cover the participant's whole daily workflow (correspondence, meetings, documents, data). In each scenario, participants can bring their own real email/report/data samples; IBA provides sample content for those who cannot.

03Training Agenda (1 Day)

The agenda is completed end to end in a single day; participants start the morning with Copilot fundamentals and Outlook, and finish the day having practised across Teams, Word/PowerPoint and Excel.

  1. Day 1 — Everyday Productivity with M365 Copilot

    • Opening: how Microsoft 365 Copilot works — generating responses based on email, calendar, file and chat data through Microsoft Graph
    • General-purpose Q&A and access to organisational information with Copilot Chat
    • Writing effective prompts: techniques for giving context, role, format and examples
    • Exercise 1: Writing a first prompt with context/role/format and trying it in Copilot Chat
    • Copilot in Outlook: summarising and categorising unread emails by importance
    • Copilot in Outlook: summarising long email threads with multiple replies
    • Copilot in Outlook: generating draft replies that reference previous correspondence and attached files
    • Copilot in Outlook: rewriting emails in a chosen tone and prioritising the inbox
    • Exercise 2: Summarising a long email thread and preparing a draft reply that references previous correspondence
    • Exercise 3: Grouping the inbox by topic/project and producing a priority order
    • Copilot in Teams: real-time summary and tracking of key points during a live meeting
    • Copilot in Teams: generating an automatic post-meeting summary, decisions and a list of action items
    • Copilot in Teams: quickly catching up on a missed meeting and summarising a topic from chat history
    • Exercise 4: Producing a summary, decisions and an action list from a meeting recording
    • Exercise 5: Generating a draft agenda for the next meeting from previous meeting notes
    • Copilot in Word: generating a draft document/report from scratch from a prompt or source file, rewriting existing text
    • Copilot in PowerPoint: generating a complete presentation from a Word document, redesigning with Designer
    • Exercise 6: Drafting a report in Word, turning it into a PowerPoint presentation and improving its design with Designer
    • Copilot in Excel: producing analysis by querying a data table in natural language, without writing formulas
    • Copilot in Excel: automatically detecting trends, outliers and patterns; getting formula suggestions from natural language
    • Copilot in Excel: chart/visualisation suggestions suited to the data, with automatic creation
    • Exercise 7: Extracting summary insights from a data table and visualising them with a suitable chart
    • General review, Q&A

04Technical Requirements

Computer

1 per participant (Windows or macOS), with an internet connection

Account

A work/school account with a Microsoft 365 Copilot licence assigned

Software

Outlook, Teams, Word, PowerPoint, Excel (desktop or web version)

Sample Content

Participants can bring their own email/report/data samples; IBA provides sample content for those who cannot

Licensing

This training requires participants to have a Microsoft 365 Copilot licence assigned in advance.

M365 Copilot is a paid add-on licence to a Microsoft 365 subscription (per-user add-on); the organisation must assign this licence to participants before the training.

For licence assignment and trial options, we recommend speaking to the organisation's Microsoft representative — this is outside the scope of the training.

05Learning Outcomes

Speed up correspondence

Summarise long email threads in Outlook and generate draft replies.

Produce meeting outputs

Extract summaries and action lists from Teams meeting recordings.

Prepare documents

Produce quick drafts and designs with Copilot in Word and PowerPoint.

Analyse data

Extract summary insights from a table in Excel.

Write effective prompts

Design prompts with context/role/format for their daily tasks.

Agent in a Day

Build an Agent in a Day Workshop

01Training Overview

Duration
1 Day
Training Hours
09:30 – 16:30
Maximum Participants
10 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants will receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — based on the organisation's preference.
Prerequisites
None. No prior experience with Microsoft Copilot Studio or AI agents is needed.
Training Objective

This training aims to teach participants how to build their first AI agent from scratch with Microsoft Copilot Studio and publish it, all in one day. The programme covers, hands-on, the core logic of agent design, setting up topics and conversation flows, connecting organisational documents as knowledge sources, and basic testing and publishing steps.

The training runs through a simple use case that each participant chooses from their own department; at the end of the day, participants leave with a working agent they can test.

By the end of the training, participants start from a blank page and own their first AI agent, suited to their own organisational scenario, tested and ready to publish.

Technical Content

Agent Design Fundamentals

  • Introduction to the Microsoft Copilot Studio interface
  • Topic and trigger logic
  • Conversation flow design
  • Connecting a knowledge source

Testing, Editing and Publishing

  • Trying out and debugging the agent with the test console
  • Improving responses
  • Basic publishing steps
  • Connecting to a channel (Teams)

02Exercises and Project Scenarios

Option A

FAQ/Support Agent

A simple support agent specialised in a single topic, answering based on frequently asked questions and core policy documents.

Option B

Survey/Information Gathering Agent

An agent that collects information from the user step by step, turning a simple form into a conversation flow.

Option C

Document-Based Q&A Agent

An agent that uses a single organisational document (e.g. an employee handbook) as its knowledge source and answers questions in natural language.

Whichever agent type is chosen, the same core skills are gained by the end of the day: design, connecting a knowledge source, testing and publishing. Participants can bring a scenario of similar scope from their own department; this does not affect the duration or price of the training. This training serves as an introduction to IBA's "Agent Development: Microsoft Copilot Studio — Advanced Agent Development Training" — the basic agent built here can later be taken to enterprise level in the advanced training.

03Training Agenda (1 Day)

The agenda is completed end to end in a single day; participants start the morning with a blank agent and finish the day with a tested agent.

  1. Day 1 — Building and Publishing an Agent from Scratch

    • Introduction to Microsoft Copilot Studio and a tour of the interface
    • Exercise 1: Creating a blank agent
    • Topic and trigger logic
    • Exercise 2: Designing the first topic and conversation flow
    • The knowledge source concept
    • Morning Deliverable: Connecting a document to the agent as a knowledge source
    • Debugging methods with the test console
    • Exercise 4: Live testing with the agent and improving its responses
    • Publishing steps and channel connection
    • Final Deliverable: Publishing the agent and trying it out in a channel
    • General review, Q&A and certificate presentation

04Technical Requirements

Computer

1 per participant (Windows or macOS), with an internet connection

Browser

Microsoft Edge or Google Chrome (latest version)

Account

Microsoft work/school account

Software Installation

None. All work is done in the browser.

Knowledge Source Document

Participants can bring a sample document (FAQ, handbook, etc.) to use for their own scenario; IBA provides a sample document for participants who cannot.

Licensing

No licence purchase is needed for this training.

The free Microsoft Copilot Studio trial environment is used throughout the training.

If the organisation puts the agent it develops into production, a separate licence may be needed under Copilot Studio's message-based licensing model — this is outside the scope of the training and a separate matter to assess.

05Learning Outcomes

Build their first agent

Create a Copilot Studio agent from scratch and design its basic conversation flow.

Connect a knowledge source

Add an organisational document to the agent as a knowledge source.

Test and improve

Try out the agent in the test console and correct its responses.

Publish

Publish the agent and connect it to a channel.

Assess the scope

Assess which scenarios in their own department are suited to an agent.

AB-620 (Copilot Studio Agent Development)

Advanced Agent Development Training

01Training Overview

Duration
2 Days
Training Hours
09:30 – 16:30
Maximum Participants
10 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants will receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — based on the organisation's preference.
Prerequisites
Basic agent development experience with Microsoft Copilot Studio is required (attendance at IBA's "Agent in a Day" training or equivalent experience is recommended).
Training Objective

This training aims to give advanced Microsoft Copilot Studio skills to participants who want to move from simple, single-task bots to AI agents that are integrated with the organisation, governed and scalable. Throughout the programme, participants learn hands-on: advanced topic and entity design, connecting actions with Power Automate, external data/API integration, knowledge orchestration with generative answers, authentication and security, continuous improvement with analytics, and governance and lifecycle management.

The training is based on a basic agent the participant has built beforehand (or builds quickly at the start of the training); over two days, this agent is gradually taken to a governed, secure and scalable enterprise-level structure.

By the end of the training, participants do not have a bot that does a single task; they own an enterprise agent that meets the organisation's security and governance standards and can be monitored and maintained.

Technical Content

Advanced Agent Design and Integration

  • Advanced topic and entity design
  • Connecting actions to Power Automate flows
  • External data sources and API integration
  • Generative answers and knowledge orchestration

Security, Analytics and Governance

  • Authentication and security
  • Analytics and continuous improvement
  • Governance and lifecycle management for agents
  • Advanced deployment scenarios

02Exercises and Project Scenarios

Option A

Enterprise FAQ/Support Agent

A support agent that brings multiple knowledge sources (documents, FAQs, policies) together in a single agent and serves them with department-specific access rules.

Option B

Process Automation Agent

An agent that takes actions, managing a business process (request creation, approval tracking, information updates) end to end with Power Automate.

Whichever agent type is chosen, the technical scope and outcomes of the training stay the same — the same integration, security and governance topics are covered. Participants can bring an agent they built in IBA's "Agent in a Day" training or in their own organisation; for participants without an existing agent, a quick basic agent is built at the start of the training. The training is based on the participant's existing agent (or one built quickly at the start of the training); over two days, this agent is gradually taken to enterprise level: on Day 1, advanced topic/entity design, Power Automate actions and external data integration are added to the agent; on Day 2, a security, analytics and governance layer is added, finishing with advanced publishing scenarios. The organisation's own process can also be used as the case — provided it is of similar scope, this does not affect the duration or price of the training.

03Training Agenda (2 Days)

The agenda is delivered using the agent the participant brings to the training (or one built quickly at the start of the training).

  1. Day 1 — Advanced Agent Design and Integration

    • Advanced architectural concepts in Copilot Studio
    • Advanced topic design and using entities
    • Exercise 1: Adding an advanced topic/entity structure to the chosen agent
    • Connecting actions to Power Automate flows
    • Exercise 2: Enabling the agent to take an action that triggers a business process
    • External data sources and API integration
    • Exercise 3: Connecting an external data source to the agent
    • Generative answers and knowledge orchestration
    • Exercise 4: Configuring generative answers from knowledge sources
  2. Day 2 — Security, Analytics, Governance and Final Deliverable

    • Authentication methods
    • Exercise 5: Setting up user-specific authorisation
    • Analytics: usage data and measuring conversation quality
    • Exercise 6: Configuring the analytics dashboard and reviewing the first findings
    • Governance and lifecycle management for agents
    • Exercise 7: Defining governance rules (versioning, approval, publishing)
    • Advanced publishing scenarios (multi-channel)
    • Exercise 8: Deploying the agent to multiple channels
    • Final Deliverable: Testing the chosen agent end to end and publishing it
    • General review, Q&A and certificate presentation

04Technical Requirements

Computer

1 per participant (Windows or macOS), with an internet connection

Browser

Microsoft Edge or Google Chrome (latest version)

Account

Microsoft work/school account

Software Installation

None. All work is done in the browser.

External Data/API Access

Test access details for the external API or data source to be used in the integration scenario should be shared before the training.

Licensing

No licence purchase is needed for this training.

The free Microsoft Copilot Studio trial environment is used throughout the training.

If the organisation puts the agent it develops into production, a separate licence may be needed under Copilot Studio's message-based licensing model — this is outside the scope of the training and a separate matter to assess.

05Learning Outcomes

Develop enterprise agents

Combine multiple data sources and business processes in a single agent.

Set up secure access

Define user-specific authorisation with authentication.

Monitor performance

Use analytics data to identify where the agent falls short and improve it.

Establish governance

Manage the agent's publishing, versioning and lifecycle to enterprise standards.

Publish at scale

Deploy the agent securely to multiple channels (Teams, web, etc.).

Make investment decisions

Assess the difference between a simple bot and an enterprise agent and choose the right scope.

Microsoft Copilot Prompt Eng & M365 & Studio & Cowork

Prompt Engineering, M365, Copilot Studio and Cowork Training

01Training Overview

Duration
2 Days
Training Hours
09:30 – 16:30
Maximum Participants
10 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants will receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — based on the organisation's preference.
Prerequisites
None. No prior experience with any Microsoft Copilot product is needed.
Training Objective

This training aims to introduce participants to the full Microsoft Copilot toolset — from everyday productivity apps to custom agent development — and to build effective prompt writing skills. The programme covers, hands-on, an overview of the Copilot ecosystem, everyday productivity with Microsoft 365 Copilot, building custom agents with Copilot Studio, collaborative working scenarios, reusable prompt patterns, and responsible AI and governance.

Rather than being built around a single project, the training runs through independent hands-on scenarios that each show a different side of the Copilot toolset — so participants see the whole toolset.

By the end of the training, participants can use Copilot not just as a chat tool but as a complete toolset ranging from everyday productivity to custom agent development, and can write prompts correctly.

Technical Content

The Microsoft Copilot Ecosystem and Everyday Use

  • Overview of the Microsoft Copilot toolset
  • Everyday productivity with Microsoft 365 Copilot (Word, Excel, Teams, Outlook)
  • Prompt engineering fundamentals and reusable patterns

Copilot Studio, Collaborative Scenarios and Governance

  • Fundamentals of building a custom agent with Copilot Studio
  • Using Copilot in collaborative working scenarios
  • Responsible AI and governance
  • End-to-end hands-on Copilot scenarios

02Exercises and Project Scenarios

Scenario 1

Everyday Productivity Task with M365 Copilot

Covers drafting a report, summarising data and producing an action list from meeting notes with Copilot in Word, Excel and Teams.

Scenario 2

Prompt Engineering Exercise

Covers designing reusable prompt patterns and turning them into a prompt library the team can share.

Scenario 3

Building a Simple Agent with Copilot Studio

Covers building a simple FAQ/support agent based on a single knowledge source from scratch in Copilot Studio and testing it.

Scenario 4

Collaborative and Responsible Use Scenario

Covers carrying out a team task end to end with Copilot and applying a responsible AI/governance checklist.

Together, the four scenarios show the full Copilot toolset, from everyday productivity to custom agent development and responsible use. The four scenarios below are the training's standard set of exercises and are covered in order throughout the training — they are not alternatives, all of them are covered; Copilot's strength does not come from a single chat window but from being able to adapt to different tools, from everyday productivity apps to custom agent development. If the organisation wants a scenario of similar complexity from its own processes (report preparation, meeting summaries, FAQ answering, etc.), it can replace one of the existing scenarios. Changing a scenario does not affect the duration or price of the training.

03Training Agenda (2 Days)

The agenda is delivered day by day through four independent scenarios; the first day focuses on M365 Copilot and prompt engineering, the second on Copilot Studio and responsible use.

  1. Day 1 — M365 Copilot and Prompt Engineering

    • Overview of the Microsoft Copilot toolset
    • Everyday productivity with Microsoft 365 Copilot
    • Exercise 1: Preparing a report draft and data summary with Copilot in Word/Excel
    • Exercise 2: Producing a summary and action list from meeting notes in Teams
    • Prompt engineering fundamentals: context, role, format
    • Exercise 3: Designing a reusable prompt pattern
    • End-of-Day Deliverable: Creating a prompt library the team can share
  2. Day 2 — Copilot Studio, Collaborative Scenarios and Responsible Use

    • Introduction to Copilot Studio and a tour of the interface
    • Exercise 4: Creating a blank agent and connecting a knowledge source
    • Exercise 5: Testing and improving the FAQ/support agent
    • Using Copilot in collaborative working scenarios
    • Exercise 6: Carrying out a team task end to end with Copilot
    • Responsible AI and governance principles
    • Final Deliverable: Applying a responsible AI checklist to a Copilot output
    • General review, Q&A and certificate presentation

04Technical Requirements

Computer

1 per participant (Windows or macOS), with an internet connection

Browser

Microsoft Edge or Google Chrome (latest version)

Account

A work/school account with a Microsoft 365 Copilot licence assigned

Software Installation

None. All work is done in the browser and Microsoft 365 apps.

Copilot Studio Access

Access to Copilot Studio for the agent-building exercise (the free trial environment is sufficient)

Licensing

The M365 Copilot part of this training requires participants to have a Microsoft 365 Copilot licence assigned in advance.

M365 Copilot is a paid add-on licence to a Microsoft 365 subscription (per-user add-on); the organisation must assign this licence to participants before the training.

The custom agent exercise with Copilot Studio can be done in the free trial environment; no separate purchase is needed.

If the organisation puts an agent developed in Copilot Studio into production, a separate licence may be needed under the message-based licensing model — this is outside the scope of the training.

05Learning Outcomes

Speed up daily work

Produce reports, summaries and content with Copilot in Word, Excel and Teams.

Write effective prompts

Design reusable prompt patterns and share them with their team.

Build a simple agent

Create a support agent based on a single knowledge source in Copilot Studio.

Apply it to teamwork

Carry out a team task end to end with Copilot.

Use it responsibly

Evaluate Copilot outputs for accuracy, bias and privacy.

Power Automate & Copilot Studio Process Automation

Process Automation

01Training Overview

Duration
6 Days
Training Hours
09:30 – 16:30
Maximum Participants
10 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants will receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — based on the organisation's preference.
Prerequisites
Before the training, each participant is expected to set up a Microsoft 365 developer tenant and confirm that Microsoft Entra access and premium connector/AI Builder usage are ready. A short "run your first flow" warm-up task should be completed before the training.
Training Objective

This training aims to give IT teams end-to-end skills in the process automation and AI capabilities of Microsoft Power Platform. The programme covers, hands-on, building resilient Power Automate flows, connecting systems through custom connectors and API integrations, applying security and governance rules, and developing AI-powered agents with the current Copilot Studio experience.

The training is hands-on from start to finish. Throughout, a single realistic IT scenario ("IT Operations Hub") is built together step by step, with a new technical layer added to it each day. As a result, participants finish the training not with scattered exercises but with a working, reusable, integrated automation solution. On the final day, all the skills learned are put to the test end to end, in teams, on an independent IT scenario.

By the end of the training, participants are not people who can build a single flow; they are IT professionals who can deliver a secure, versionable and maintainable automation solution at enterprise scale.

Technical Content

Flow Architecture, Data Processing and Integration

  • Cloud flow architecture, trigger types and how connectors work
  • Expressions, data processing with Power Fx and JSON parsing
  • Conditions, loops, pagination and error/retry handling
  • The HTTP action and creating custom connectors with OpenAPI
  • Multi-stage approval processes and conditional routing
  • Security: DLP policies, environment strategy, running as a service principal

AI, Governance and Enterprise Delivery

  • The current Copilot Studio experience: structure, instruction-based agent behaviour, knowledge sources
  • AI Builder and Copilot in Power Automate: document processing, categorisation, text generation
  • Solution architecture, reusable child flows and refactoring
  • Cross-environment deployment (pipelines), source control and lifecycle management
  • Governance and monitoring: the Center of Excellence approach, analytics
  • Developing an end-to-end case in teams, presentation and architecture/security review

02Exercises and Project Scenarios

Main Scenario

IT Operations Hub

A working, end-to-end IT automation hub, grown step by step over 5 days, from new-user onboarding notifications to IT ticket/request management, and from custom connector integration to an AI-powered support assistant.

Day 6 — Case Workshop · Independent Team Scenarios

User onboarding/offboarding automation, incident notification and automatic escalation, access/request self-service flow — teams choose one of these three scenarios, implement it end to end using the skills from the previous 5 days, and present it.

The same IT Operations Hub scenario is followed for five days; participants leave the training not with scattered exercises but with a solution that works end to end. This training uses a single fixed scenario; there is no alternative project option — each day adds a new technical layer to the previous one. The organisation's own IT process (e.g. ticket/request management, onboarding/offboarding automation) can replace this scenario provided it is of similar complexity; this does not affect the duration or price of the training. On the final day, these skills are tested in teams on an independent scenario: teams choose one of the user onboarding/offboarding automation, incident notification/escalation or access self-service flow scenarios and implement it end to end.

03Training Agenda (6 Days)

The agenda is delivered step by step, day by day, through the IT Operations Hub scenario; each day adds a new technical layer to the previous one.

  1. Day 1 — Fundamentals and First Flows

    • The Power Platform ecosystem, environments and how connectors work
    • Anatomy of a cloud flow: trigger, action and connector concepts
    • Trigger types: instant, automated, scheduled; the difference between webhook and polling
    • Exercise 1: Building a welcome/notification flow that runs when a new user is created in Entra
    • Concurrency and debugging from run history
    • Exercise 2: Developing scheduled reminder and manually triggered flows
  2. Day 2 — Flow Logic and Data

    • Data processing with expressions and Power Fx
    • Conditions, loops and filter/select/join data operations
    • JSON parsing, schema generation and pagination
    • Exercise 3: Retrieving paginated data, parsing the JSON output and producing a summary report
    • Error handling: scopes and retry policies
    • Exercise 4: Building a resilient, self-recovering flow for deliberate error/limit scenarios
  3. Day 3 — Integration and Security

    • The HTTP action and creating custom connectors from OpenAPI
    • Exercise 5: Connecting an external ticket/request system with a custom connector
    • Approval processes and conditional routing
    • Exercise 6: Setting up multi-stage approval and escalation based on priority
    • Security: DLP policies, environment strategy, connection reference management
    • Service principals, run-only users and storing secrets securely
  4. Day 4 — Copilot Studio and AI Builder

    • The current Copilot Studio experience: structure, preview, evaluation, monitoring
    • Instruction-based agent behaviour, knowledge sources, skills and memory
    • Exercise 7: Defining an IT support assistant and connecting it to the knowledge base and ticketing tool
    • AI Builder and Copilot in Power Automate: document processing, categorisation, text generation
    • Exercise 8: An AI step that reads an incoming request and suggests a category/priority/owner, with the flow escalating accordingly
  5. Day 5 — ALM, Governance and Advanced Topics

    • Solution architecture, reusable child flows and refactoring
    • Exercise 9: Bringing the pieces from previous days into a single solution and turning shared steps into child flows
    • Cross-environment deployment (pipelines) and source control
    • Exercise 10: Deploying the solution from the development environment to test/production environments
    • Positioning: comparing workflows, agent flows, cloud flows and Logic Apps; licensing and cost
    • Governance and monitoring: the Center of Excellence approach, analytics
  6. Day 6 — Case Workshop

    • Teamwork and architecture planning (design first, then build)
    • Final Deliverable: Developing end to end the realistic IT scenario assigned to each team (onboarding/offboarding automation, incident notification/escalation or access self-service flow)
    • Presentation, peer review and architecture/security review by the trainer
    • General review, scaling and governance recommendations, certificate presentation

04Technical Requirements

Computer

1 per participant (Windows or macOS), with an internet connection

Browser

Microsoft Edge or Google Chrome (latest version)

Microsoft 365 Developer Tenant

Each participant should have set up their own tenant and tested Microsoft Entra access before the training

Premium Connector / AI Builder Access

To be confirmed by the organisation before the training — see Licensing

Preparation

Completing a short "run your first flow" warm-up task before the training is recommended; it avoids time lost to setup issues

Licensing

This training goes beyond standard free licences — some steps require premium connector and AI Builder access.

Participants can set up a free Microsoft 365 developer tenant before the training; this step itself requires no purchase.

To use premium connectors and AI Builder, the organisation needs a suitable Power Automate/Power Platform licence (premium plan or AI Builder credits) or must activate a trial licence — this is a separate matter that needs to be clarified before the training.

The current Copilot Studio experience can be used in the free trial environment throughout the training.

05Learning Outcomes

Build resilient flows

Develop self-recovering automation flows that handle errors and edge cases professionally.

Integrate systems

Connect external systems to Power Automate through custom connectors and API integration.

Work securely and with governance

Build flows that follow enterprise security rules using DLP policies, environment strategy and service principals.

Add an AI layer

Give automation flows decision/classification capabilities with Copilot Studio and AI Builder.

Deliver at enterprise scale

Version the solution, split it into child flows and deploy it across environments (ALM).

Run an independent project

Design, develop and present a new IT scenario end to end as a team.

AIF-101 Microsoft Foundry (Azure AI Foundry)

AI Agents and Business Integration Training

01Training Overview

Duration
2 Days
Training Hours
09:30 – 16:30
Maximum Participants
10 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — set according to the organisation's preference.
Prerequisites
No software development knowledge is required. Basic experience with Power Apps or Power Automate is recommended.
Training Objective

This training teaches participants to choose a model, design prompts and build an AI agent that answers from corporate documents on Microsoft Foundry (formerly Azure AI Foundry). In the second half of the training the agent is connected to Copilot Studio, Power Automate and Microsoft Teams, so participants make the agent available inside the tools employees use every day.

All work is done in the Foundry portal and the Power Platform interfaces; no code is written. Before the training, a case is selected together with the organisation, and this case grows over the two days. On the first day the models are compared and the case's prompt is prepared; on the second day the same case is turned into an agent that uses tools and is connected to business systems.

The training overlaps with the platform, model and agent topics of Microsoft's new AI-103 certification. As it does not cover the code side in detail, it is not a substitute for exam preparation; it is a starter programme for teams that will use Foundry together with Power Platform.

Participants leave the training with a working agent that answers from their own documents and can be used from Teams.

Technical Content

The Microsoft Foundry Platform

  • The move from Azure AI Studio to Microsoft Foundry and the platform's current components
  • Setting up the Foundry resource and project structure
  • Using the models, agents, tools, evaluation and monitoring areas of the portal
  • Deciding which of Foundry, Copilot Studio and the Power Platform AI features fits which need
  • Defining project permissions with role-based access control (RBAC)
  • How quota, cost and region choices affect the project plan

Models and Prompt Engineering

  • Comparing large language models (LLMs), small language models (SLMs) and multimodal models in the model catalog
  • Model deployment options and deploying a model to the project
  • Writing a system prompt in the playground and tuning output with parameters (temperature, max tokens)
  • Getting structured (JSON) output so answers can be used in business workflows
  • Comparing two models on the same scenario for quality and cost
  • Preventing harmful or out-of-scope answers with content filters and guardrail settings

Agents with Foundry Agent Service

  • Defining the agent's role, goal and behaviour rules with instructions
  • Uploading corporate documents and making the agent answer from them (grounding)
  • The logic of connecting knowledge sources with Foundry IQ and Azure AI Search
  • Extending the agent's abilities with built-in tools (file search, web search, code interpreter)
  • Testing the agent in the playground and reviewing the conversation history (thread)
  • Measuring the groundedness and relevance of answers with evaluators

Power Platform and Microsoft 365 Integration

  • Connecting the Foundry agent to a Copilot Studio agent for a single chat experience
  • Defining a Power Automate flow as a tool (action) for the agent to open a request record
  • Publishing the agent to Microsoft Teams and the Microsoft 365 Copilot channel
  • Using Dataverse or SharePoint data as a knowledge source for the agent
  • Adding a human approval (human-in-the-loop) step for critical operations
  • A pre-production checklist for security, data loss prevention (DLP) and cost

02Exercises and Project Scenarios

Option A

Insurance Policy and Claims Assistant

An agent that answers customer representatives' questions about policy terms, coverage and the claims process from policy documents. When it cannot answer, it opens a claim request through Power Automate and notifies the relevant team in Teams.

Option B

Retail Customer Service Agent

An agent that answers questions about returns, exchanges, shipping and campaign terms from company policies. It creates a complaint record for a customer who reports an order problem and sends return requests above a set amount for manager approval.

Option C

HR Policy and Leave Assistant

An internal assistant used through Teams that answers employees' questions about leave, benefits, expense and travel policies from HR documents. It forwards leave requests to the manager with a Power Automate flow.

The scenarios below are examples and are not all covered together. The organisation chooses one of them, and that case is worked on over the two days. A different case from the organisation's own processes can also be proposed. As long as it is of similar scope, changing the scenario does not affect the duration or price of the training.

Whichever case is chosen, the technical scope and outcomes stay the same. On Day 1 a Foundry project is set up for the selected case, two models are compared on the same questions, and the case's system prompt and safety settings are prepared. On Day 2 this prompt is turned into an agent that answers from the case's documents. The agent gains the ability to open a request record with Power Automate, is published to Teams through Copilot Studio, and its answer quality is measured with evaluators at the end of the day. Participants finish the training with a single working agent that can be piloted inside the organisation.

03Training Agenda (2 Days)

The agenda is worked through on the project case selected before the training. Every step below that mentions "the selected case" is applied to the scenario the organisation prefers (Insurance Policy Assistant / Retail Customer Service / HR Policy Assistant, or a process specific to the organisation).

  1. Day 1 — Foundry Fundamentals, Model Selection and Prompt Design

    • Introduction to Microsoft Foundry: how the platform evolved and its current components
    • Foundry, Copilot Studio and Power Platform AI compared: which one, when?
    • Foundry resource, project structure and permissions
    • Exercise 1: Creating the Foundry project for the selected case
    • The model catalog: LLMs, SLMs and multimodal models; deployment options
    • Exercise 2: Deploying two models and comparing quality and cost on the case's questions
    • Prompt engineering: system prompt, few-shot prompting, parameters
    • Exercise 3: Writing the selected case's system prompt and getting structured (JSON) output
    • Exercise 4: Configuring content filters and guardrail settings
    • Interim Deliverable: The case's prompt, the selected model and the decision matrix are ready
  2. Day 2 — Agent Development and Integration with Power Platform

    • Foundry Agent Service: instructions, tools and knowledge sources
    • Exercise 5: Building the agent that answers from the selected case's documents
    • Exercise 6: Testing the agent in the playground and improving its instructions
    • The architecture for using a Foundry agent together with Copilot Studio
    • Exercise 7: Adding a Power Automate flow that opens a request record as a tool for the agent
    • Exercise 8: Publishing the agent to Microsoft Teams through Copilot Studio
    • Human approval for critical operations and responsible AI principles
    • Exercise 9: Measuring answer quality with evaluators
    • Final Deliverable: Testing the selected case end to end in Teams and drawing up the pilot checklist
    • Overall review, Q&A and certificate presentation

04Technical Requirements

Computer

1 per participant (Windows or macOS), with internet access

Browser

Microsoft Edge or Google Chrome (current version)

Account

The organisation's Microsoft Entra ID work account; permission to create a Foundry project in the organisation's Azure subscription

Power Platform Access

Access to Copilot Studio and Power Automate (a trial licence is sufficient)

Software Installation

Not required. All work is done in the browser.

Licensing

The exercises run on the organisation's own Azure subscription.

The organisation's subscription must have Microsoft Foundry access and model quota. Model usage in Foundry is billed pay-as-you-go, and this cost goes to the organisation's subscription. The subscription, participant permissions and quota check are completed with the organisation's IT team at least one week before the training.

A trial licence is sufficient for the Copilot Studio and Power Automate work. Using the agent in production after the training may require a separate licence, which is outside the scope of this training.

05Learning Outcomes

Choose the right tool

Decide, with reasons, which of Foundry, Copilot Studio or the Power Platform AI features fits an AI request from inside the organisation.

Choose and deploy a model

Compare two models for a business scenario on quality and cost and deploy the right one to a Foundry project.

Design prompts

Turn the rules of a business process into a system prompt and get structured, consistent answers.

Build an agent that answers from documents

Upload the organisation's policy or procedure documents and build an agent that gives source-based answers.

Connect the agent to business systems

Make the agent take actions with Power Automate flows and publish it to Teams through Copilot Studio.

Measure quality and safety

Set up content filters, measure the agent's answer quality with evaluators and draw up a pre-pilot checklist.

Copilot Studio Governance

Enterprise Agent Management and Monitoring Training

01Training Overview

Duration
2 Days
Training Hours
09:30 – 16:30
Maximum Participants
10 Participants
Hands-on Ratio
80% Hands-on / 20% Theory
Certificate
Participants receive an IBA Consultancy Certificate of Participation at the end of the training.
Format
Online, On-site or Hybrid — set according to the organisation's preference.
Prerequisites
No agent development knowledge is required. Basic familiarity with the Microsoft 365 and Power Platform admin centers is enough.
Training Objective

This training is designed for the IT, information security, compliance and CoE teams that manage the Copilot Studio agents used in the organisation. Participants do not build agents. Throughout the training they take over an advanced fleet of agents in a test environment prepared by IBA and bring that fleet in line with corporate rules.

Every agent in the fleet carries a governance problem that is common in real organisations: a web agent with authentication turned off, an agent that uses public websites as a knowledge source, an orphaned agent left behind by an employee who has left the company, and so on. Over two days participants inventory these agents, classify them by risk, bring them under control with environment, permission and DLP policies, monitor them with audit logs and limit their credit consumption.

The admin centers used are the Power Platform admin center, the Microsoft 365 admin center, Microsoft Entra and Microsoft Purview. Topics are covered through hands-on exercises carried out directly in these portals.

Participants leave the training with a management routine they can apply in their own tenant the next day and a draft Governance Playbook tailored to their organisation.

Technical Content

Governance Approach and Environment Management

  • Placing agents at a risk level according to the three-zone (Citizen / Partnered / Professional) governance model
  • Mapping the control planes by tracing the same agent across the Power Platform admin center, the Microsoft 365 admin center, Entra and Purview
  • Listing, filtering and exporting all agents across the tenant with the Power Platform inventory
  • Managing, at tenant level, the publishing of agents that use generative AI, data movement across geographies, and self-service trials
  • Building a secure environment strategy with Dev / Test / Prod separation and default environment routing
  • Applying policies to many environments at once with environment groups and rules; enabling Managed Environments

Access Control and Data Protection

  • Restricting environment access and Copilot Studio author permissions with Entra security groups
  • Assigning Dataverse security roles on the principle of least privilege and narrowing access to conversation transcripts
  • Setting up a read-only auditor account with the AI Administrator and AI Reader roles
  • Blocking unauthenticated chat, knowledge sources, HTTP requests, channels and event triggers with Copilot Studio-specific DLP connectors
  • Restricting the processing of sensitive content in agents with Purview sensitivity labels and DLP policies
  • Mapping transcript retention, data masking and data residency settings to KVKK (Turkish data protection law) and internal policies

Lifecycle, Sharing and Publishing Control (from an administrator's perspective; no agent development)

  • Controlling releases to production with pipeline approval steps and a managed solution rule
  • Auditing settings that sit outside the solution (channels, authentication, sharing) after deployment
  • Finding orphaned agents left behind by departed employees, transferring ownership, quarantining and retiring them
  • Enforcing sharing rules with Editor / Viewer permissions, the viewer limit and security group restrictions
  • Approving, rejecting or blocking publishing requests in the Microsoft 365 admin center based on their data source and action details
  • Interpreting the pre-publish security scan and the runtime protection status

Monitoring, Reporting and Cost Management

  • Investigating incidents by querying agent changes and user interactions in Purview Audit
  • Reviewing risky interactions and conversation transcripts with DSPM for AI
  • Monitoring usage and quality with Copilot Studio analytics and the Copilot hub; evaluating a Sentinel alert scenario
  • Allocating Copilot Credit capacity per environment and setting overage behaviour
  • Defining a monthly consumption limit per agent and spotting abnormal usage in consumption reports
  • Designing the enterprise management model with CoE roles, an agent request process and a monthly audit routine

02Exercises and Project Scenarios

Option A

Holding Company Agent Fleet

Made up of agents belonging to the HR, finance and procurement departments. The key topics are special categories of personal data such as employee data, who can access conversation transcripts, and oversharing across departments.

Option B

Finance / Insurance Agent Fleet

Made up of customer-facing web agents and internal operations agents. Mandatory authentication, channel restrictions, sector regulatory requirements and keeping a complete audit trail come to the fore.

Option C

Manufacturing / Retail Agent Fleet

Made up of IT helpdesk, supply and field operations agents. Connectors to external systems, agents that run on autonomous triggers and high credit consumption come to the fore.

The cases below are alternatives to each other and are not all covered together. The organisation chooses the case closest to its sector. Each case has a fleet of seven agents loaded by IBA into the test environment. In the fleet, one agent has authentication turned off, one uses public websites as a knowledge source, one sends data to an external service over HTTP, one runs on an autonomous trigger, one is shared with the whole company, one has been left orphaned, and one consumes far more credits than expected. Participants do not build agents; they manage this fleet.

Whichever case is chosen, the technical scope and outcomes stay the same. On Day 1 participants take over the selected case's fleet. They inventory the agents and place them in risk zones, set up the tenant settings and the environment structure, define the role model and make the violations in the fleet visible with DLP policies. By the end of Day 1 the organisation's secure baseline configuration is complete. On Day 2 the agent that uses sensitive data is protected with Purview and retention settings, a request to go to production is taken through the approval process, the orphaned agent is handed over and the overshared agent is brought in line with the rules. Then a security incident is investigated with audit logs, and the agent consuming too many credits is found in the consumption report and limited. The training closes with a presentation of the Governance Playbook that gathers the decisions made over the two days.

03Training Agenda (2 Days)

The agenda is worked through on the case selected before the training. Every step below that mentions "the selected case" is applied to the agent fleet of the sector scenario the organisation prefers (Holding Company / Finance-Insurance / Manufacturing-Retail).

  1. Day 1 — Setting Up Visibility, Environments and Access Control

    • The Copilot Studio governance approach: the three-zone model and the control planes
    • Agent visibility in the Power Platform inventory and the Microsoft 365 admin center
    • Exercise 1: Inventorying the selected case's agent fleet and assigning the agents to risk zones
    • Tenant settings: controlling generative AI publishing, data movement and self-service trials
    • Environment strategy: Dev / Test / Prod, routing, environment groups and rules, Managed Environments
    • Exercise 2: Setting up the environment structure and applying rules in bulk with an environment group
    • Permissions: Entra groups, Dataverse roles, the Copilot Author setting, AI Administrator / AI Reader
    • Exercise 3: Building the role matrix (RACI) and verifying permissions with the auditor account
    • Exercise 4: Rolling out Copilot Studio DLP policies step by step and reporting the violations in the fleet
    • Interim Deliverable: The secure baseline configuration for the selected case is completed
  2. Day 2 — Data Protection, Lifecycle, Monitoring and the Management Model

    • Data security and compliance: sensitivity labels, Purview DLP, transcript retention, masking, data residency
    • Exercise 5: Protecting the agent that uses sensitive data with Purview and retention settings; mapping to a KVKK checklist
    • Lifecycle management: approved deployment, managed solutions, orphaned agents and retirement
    • Exercise 6: Handling a request to go to production through the approval process; handing over or quarantining the orphaned agent
    • Exercise 7: Bringing the overshared agent in line with the sharing rules and reviewing and approving / rejecting a publishing request
    • Monitoring and audit: Purview Audit, DSPM for AI, analytics, Copilot hub, Sentinel
    • Exercise 8: Investigating an incident with audit logs: "Who changed this agent's knowledge source, and when?"
    • Exercise 9: Credit allocation, a monthly limit per agent, and finding the agent with abnormal consumption in the consumption report
    • Final Deliverable: Preparing and presenting the draft Governance Playbook for the selected case
    • Overall review, Q&A and certificate presentation

04Technical Requirements

Computer

1 per participant, with internet access

Browser

Current Edge or Chrome (two separate profiles); no installation required

Account

A lab account provided by IBA or the organisation's test tenant

Licensing

No licence purchase is required on the participant side for this training.

The exercises run in a test environment prepared by IBA; the required Copilot Studio, Managed Environments and Purview features are ready in this environment. Agent 365 and Defender real-time protection are shown as demos.

05Learning Outcomes

Agent inventory

Find all agents and classify them by risk zone.

Environment management

Bring the tenant and the Dev/Test/Prod environments up to the corporate standard.

Access management

Distribute permissions on the principle of least privilege using a role matrix.

DLP

Roll out Copilot Studio DLP policies step by step and interpret the violations.

Compliance

Prepare a checklist that maps KVKK and internal policies to technical settings.

Lifecycle and publishing

Manage releases to production, sharing and orphaned agents through an approval process.

Incident investigation

Find in the audit logs who made a change to an agent, and when.

Cost

Keep consumption under control with credit allocation and a per-agent limit.

Management model

Prepare a Governance Playbook with roles, a request process and an audit routine.

Frequently Asked Questions

There are three separate programs: Agent in a Day runs 1 day, Microsoft Copilot 2 days and Agent Development 2 days. Training hours are 09:30 to 16:30.

It can be delivered online, on site or hybrid. The format is set according to your preference.

Agent in a Day and Microsoft Copilot need no prior experience. Agent Development assumes basic agent-building experience in Copilot Studio; Agent in a Day is the natural entry point for it.

If the Copilot toolset is new to you, start with the Microsoft Copilot program. If you want to build your own agent straight away, take Agent in a Day. To move existing agents to enterprise level, continue with Agent Development.

No licence purchase is required for the training; the free Copilot Studio trial environment is used. If the agent you build goes into production, Copilot Studio licensing is assessed separately.

Each program takes up to 10 participants. Participants receive an IBA Consultancy Certificate of Attendance at the end of the training.