Microsoft Copilot Studio training: build your own AI agent for your organisation
Practical in-company training in Microsoft Copilot Studio: the low-code platform that lets you build AI agents that answer questions, trigger processes and support employees and customers around the clock. We teach the new Copilot Studio experience, in which you describe your agent in plain language instead of drawing conversation flows. Tailored to your organisation.
AI agents that relieve your organisation immediately
Organisations using Copilot Studio agents see an immediate drop in routine queries at IT, HR and customer service. Without extra staff, without programming. And since the June 2026 rebuild, a first working agent takes minutes rather than days.
Choose the training that fits your team
Copilot Studio offers two variants: the Introduction (half day) for those who want to discover what Copilot Studio is and what it can mean for their organisation, and the Hands-on day for those who want to start building themselves. In a preliminary conversation we determine together which variant fits your team and objectives best.
Copilot Studio Introduction · ½ day
What is Microsoft Copilot Studio?
A clear introduction to the platform and what sets it apart from regular chatbots and other Microsoft tools.
- From Power Virtual Agents to Copilot Studio: what changed?
- The June 2026 rebuild: nine configuration tabs became four
- What AI agents are and how they differ from regular chatbots
- Position in the Microsoft ecosystem: Power Platform and Microsoft 365
- Conversational agents and autonomous agents: two worlds
- Demo: an existing agent in action
What can you build? Use cases and architecture
Discover the most impactful applications for your organisation and understand how the platform is technically structured.
- The five use cases that pay off fastest: IT helpdesk, HR questions, customer service, knowledge management and starting processes
- When to choose Copilot Studio and when Power Automate or Azure AI?
- How an agent is put together: engine, instructions, knowledge sources, tools, skills and channels
- Demo: an agent answering from SharePoint documents without you writing a single topic
- Real-world examples: what are organisations doing with it?
How does it work? Core concepts
The building blocks you need to understand before getting started with Copilot Studio.
- Three engines under one platform: standard, GitHub Copilot and Copilot chat
- The new experience: describing your agent in plain language, with instructions, knowledge and tools on one screen
- Generative orchestration: the agent picks its own topic, tool and knowledge source
- Classic orchestration: fixed conversation routes on trigger phrases, and when you want them
- Skills and memory: reusing ways of working and remembering preferences
- Channels, licences and Copilot Credits: what does your organisation need?
Your first agent: hands-on exercise
Getting started: you build a simple agent yourself and publish it to Teams.
- Create an agent by describing it in plain language
- Set the name, tone and boundaries in the instructions
- Connect one knowledge source (SharePoint or website)
- Test the agent on the Preview tab
- Publish to Teams: demo
- Personal action plan: three concrete steps for your organisation
Introduction or Hands-on day?
The Introduction is ideal for decision-makers, IT managers and business analysts who want to understand what Copilot Studio is and whether it is relevant for their organisation. The Hands-on day is for those who want to start building themselves and leave with a working agent at the end of the day.
Want agents built instead of learning to build yourself? See our Copilot Studio agent building and coaching.
Related trainings
Copilot Studio works closely with other Microsoft tools. Combine this training with:
- Power Automate, for powerful agent flows and integrations
- SharePoint, as a knowledge base for your agent
- Microsoft 365 Copilot, for AI in your daily apps
Morning: building in the new Copilot Studio
The new Copilot Studio
We start with the new experience, because building in it is noticeably faster than in the old one.
- The New experience toggle on the home page: what changes and what stays
- From nine configuration tabs to four: Build, Preview, Evaluate and Monitor
- Creating an agent by describing it in plain language, instead of drawing conversation flows
- Writing instructions that set identity, tone and scope – including what the agent explicitly does not do
- Watching along on the Preview tab while you build
Which engine runs underneath?
The engine determines what your agent can do, where you publish, and when consumption starts counting.
- Three engines: standard, GitHub Copilot and Copilot chat, and what each is meant for
- The GitHub Copilot engine: reasons its way through multi-step tasks and finds another route when a step fails
- Word, Excel, PowerPoint and PDF files the agent creates and edits itself
- The standard engine: fixed rules, predictable answers, and the meter only starts after publishing
- The Copilot chat engine: adding your own knowledge to Microsoft 365 Copilot Chat
- You choose when creating the agent: switching engines afterwards is not possible
Everything together on the Build tab
What used to be spread across nine tabs now sits side by side on one screen.
- Instructions, knowledge, tools, skills and model in a single side panel
- Choosing the AI model that does your agent’s reasoning
- Adding tools with which the agent retrieves data or sets something in motion
- Setting boundaries: what the agent does and does not cover
- Steering by rewriting the instructions rather than dragging nodes around
Skills, memory and connected agents
The three additions that set the new experience apart from the old one most clearly.
- Skills: a way of working you capture once and reuse across several agents
- Uploading a skill, writing one yourself, or having one drafted from a description
- Turning on memory: the agent remembers preferences per user between conversations
- Connected agents: one agent at the front door that hands off to specialists
- Adding MCP servers as tools, to open up external systems
Knowledge sources
Let your agent answer from the documents, sites and systems you already have.
- Answers from your own knowledge, without you writing topics by hand
- Connect a SharePoint site as knowledge base
- Drag files in, add public websites and FAQ pages
- Dataverse, Azure AI Search, Dynamics 365, Salesforce, ServiceNow and Azure SQL as sources
- Test and fine-tune answer quality
Afternoon: taking action, publishing and managing
Tools, actions and workflows
Letting your agent act: retrieve data, create records and start processes.
- What a tool is: prompts, agent flows, connectors, MCP servers and HTTP calls
- The redesigned workflows designer: actions, decisions and AI steps on a single canvas
- Creating an agent flow from within Copilot Studio
- Retrieving data from, for example, a SharePoint list or Dataverse and returning it
- Prebuilt connectors: which are available and how you use them
- Using a flow’s output as a variable in the agent
Classic agents: authoring the conversation
The standard engine with conversation routes you write yourself, for everything that must not be left to chance.
- Turning generative orchestration off: fixed sequence, control and auditability
- Writing trigger phrases so the agent recognises the user’s question
- Conversation nodes: Message, Question, Condition and Redirect
- System topics: Conversation Start, Fallback, Escalate and End of Conversation
- Entities, variables and slot filling: recognising, remembering and intelligently asking for data
- Reaching the classic experience through the home page toggle or Other ways to build
Channels and publishing
Making your agent live where your people already work.
- Available channels: Teams, SharePoint, your own website, WhatsApp and telephony
- From development environment to production: the publication route
- Configure authentication: anonymous or signed in with Entra ID
- Every agent gets its own Microsoft Entra Agent ID, and what that means for permissions
- Handing off to a colleague when the agent cannot resolve the question
- Publishing the agent to Teams and testing it straight away
Evaluating, monitoring and managing
The two tabs that sit next to Build in the new experience, plus the management side around them.
- Evaluate: assembling test sets and measuring answer quality instead of estimating it
- Monitor: reviewing completed tasks, files the agent accessed, and activity
- Reading analytics: conversation outcomes, topic success rates and escalations
- Governance: who can build agents? DLP policies and environment management
- Licences and Copilot Credits: where consumption starts counting per engine
- Management through solutions, environments and import/export
Wrap-up and own projects
Closing the day with a working agent and a plan for tomorrow.
- Each participant briefly shows what they built
- Joint Q&A and working through sticking points
- Next steps: what does your organisation need to go live?
- Personal action plan: get started tomorrow
- Further learning: Microsoft Learn, Power CAT and adoption materials
Preparation for the Hands-on day
Participants receive a short setup guide in advance for setting up a Copilot Studio trial account. An active Microsoft 365 account is required. Basic Power Automate knowledge is an advantage but not mandatory, we cover everything needed in the training itself.
Want agents built or guidance while building rather than learning to build yourself? See our Copilot Studio agent building and coaching.
Why choose this Copilot Studio training?
No abstract theory, but hands-on building with use cases from your own organisation.
Immediately applicable
You practise with use cases from your own organisation. By the end of the day you have a working agent that you can improve or roll out tomorrow.
100% tailored
In a preliminary conversation we map your industry, work processes and knowledge level. The training and use cases are adapted accordingly, so every exercise feels familiar.
Experienced Microsoft trainer
Over 15 years of experience with the Microsoft ecosystem, and up to date with the new Copilot Studio experience and rapid Power Platform developments.
No programming knowledge needed
Copilot Studio is designed for people without a development background. In the new experience you describe your agent in plain language, without writing a single line of code.
Small group, personal attention
With a maximum of 12 participants there is room for questions, discussion and personal guidance. You work on your own use case, not a generic example.
Two formats, one choice
Choose the half-day if you want to explore what Copilot Studio can mean for your organisation, or the full day if you want to start building straight away. We help you choose during the intake call.
Copilot Studio for every department
AI agents add value in almost every business process. Select your department below and discover the concrete scenarios we cover in the training.
IT helpdesk deflection
Handle routine queries automatically so your service desk focuses on complex issues.
- Password reset agent: step-by-step guidance for end users
- Handle software requests via an agent flow
- Answer frequent IT questions from your knowledge base
- 30–50% ticket reduction with well-implemented agents
Knowledge management and documentation
Making internal IT knowledge accessible to employees via an agent.
- Making handbooks and procedures searchable via an agent
- Connecting a SharePoint knowledge base as an intelligent FAQ
- Summarising incident logs and signalling trends
- Supporting onboarding of new employees
Incident management
Faster response to reports through smart routing and escalation.
- Structure and categorise incident reports via an agent
- Automatically create a ticket in your ticketing system (e.g. ServiceNow or Dataverse)
- Keep the user informed of status
- Escalate to a live agent for complex cases
HR FAQ and self-service
Answer frequent HR questions without employees having to call an HR colleague every time.
- Explain leave policy and employment conditions via an agent
- Make policy documents searchable for employees
- Guide employees through forms and procedures step by step
- 4–6 hours saved per new employee during onboarding
Onboarding assistant
Helping new employees find their way with a personal onboarding agent.
- Work through onboarding checklist via an agent
- Answer practical questions in the first weeks
- Refer to the right person or department
- Guide introduction to tools and systems
Learning and development
Making learning offerings accessible and helping employees find the right training.
- Search training catalogue via an agent
- Learning recommendation based on role and objective
- Integrate knowledge quizzes and exercises in a conversation flow
- Track progress and certifications
FAQ and routine queries
The majority of your customer queries consist of repetitive questions that an agent can answer flawlessly.
- Automatically answer opening hours, delivery times and return policies
- Look up product information and prices via the agent
- Direct the customer to the right department or employee
- 60–70% of routine queries handled without human intervention
Order and status tracking
Keep customers informed of their order or request without manual lookup work.
- Retrieve order status via a connection to your system
- Proactively inform customers of delays or changes
- Structure returns and complaints via a conversation flow
- Measure customer satisfaction at the end of the conversation
Escalation and handoff
Smartly switching to a human agent when the AI cannot answer the question.
- Configure the Escalate topic for smooth handover
- Pass conversation summary to the agent
- Integration with Omnichannel for Customer Service
- Reduce wait times and service desk load
Knowledge management for employees
Procedures, manuals and work instructions always available quickly for the shop floor.
- Make technical documentation searchable via an agent
- Request step-by-step work instructions while performing a task
- Guide inspection checklists and safety procedures
- Support onboarding of operational staff
Process automation
Automatically process recurring requests and notifications without manual intervention.
- Structure and forward purchase requests via an agent flow
- Initiate approval workflows from a conversation
- Query inventory information from your system
- Compile reporting data on request
Autonomous agents
Agents that respond to events and act without human initiation.
- Agent that responds to a new row in a database or spreadsheet
- Automatic notifications at thresholds or deviations
- Schedule and execute periodic tasks via a time trigger
- Multi-agent orchestration for complex processes
Finance FAQ and self-service
Quickly answer frequent financial questions from employees and managers.
- Explain expense policy and budget guidelines via an agent
- Clarify procurement processes and approval thresholds
- Look up VAT rules and cost categories
- Guide employees to the right form or system
Process support
Execute recurring finance processes faster and more accurately.
- Create a purchase order via a structured conversation flow
- Retrieve invoice status from your ERP or Dataverse
- Initiate approval workflow from the agent
- Work through compliance checklist for expenses
Reporting and insights
Making financial data available faster to the right person.
- Query budget status per department
- Compile variance reporting on request
- Support periodic close with a checklist agent
- Combine and present data from multiple sources
Make-or-buy decision
Understanding when building yourself with Copilot Studio is the right choice and when it is not.
- ROI analysis: when does an agent pay for itself?
- Copilot Studio vs. custom Azure AI solutions: pros and cons
- Typical lead time and investment for a simple agent
- Organisational prerequisites for successful implementation
Governance and control
Deploying agents responsibly with the right rules and management structure.
- Who can build agents? Roles and responsibilities
- Data Loss Prevention (DLP) policies for agents
- Environment strategy: development, test and production
- Licences and Copilot Credits: manage and monitor costs
Adoption and change management
Ensuring the agent is actually used and valued.
- Introducing employees to a new agent
- Using analytics to measure and improve adoption
- Promoting agents via internal communication
- Growth scenarios: from one agent to an agent ecosystem
Everything you need to know
Approach
The training combines short theory moments with extensive hands-on exercises. We use use cases and scenarios from your own organisation or industry, so you can immediately translate what you have learned to your own situation.
Prior knowledge
Basic knowledge of Microsoft 365 is desirable. Programming experience is not required: Copilot Studio is designed for business professionals. For the Hands-on day, some knowledge of Power Automate is an advantage, but we cover the essentials in the training itself. In the preliminary conversation we align the level with the participants.
Licences
Hands-on exercises require access to Copilot Studio. A free trial account is sufficient for the training. Participants receive a setup guide in advance. We are happy to advise on licence options for production use after the training.
Materials & follow-up
After the training each participant receives a comprehensive summary, the practice agents built and a reference card with the key concepts. Feel free to get in touch with questions after the training.
Customisation & combinations
Want to combine this training with Power Automate, SharePoint or a deep-dive on a specific use case (e.g. IT helpdesk automation)? We are happy to put together a custom programme.
From request to result
Every Copilot Studio training is carefully prepared and tailored to your organisation.
Getting acquainted
A short conversation to understand your wishes, target group, industry and existing Microsoft environment. No obligation.
Tailoring the programme
We design a programme with use cases and exercises from your own organisation and industry. Including a licence setup guide sent in advance.
Delivering the training
Interactive, practical and hands-on. Online or on-site. Participants leave with a working agent.
Follow-up
Summary, practice agents and the option of follow-up support. So what was learned sticks and your agent goes live.
What participants say
"A very educational training delivered with great enthusiasm. Interactive, room for our own cases and lots of practical tips and tricks I can use straight away."
"Learned a lot about smarter ways to achieve the same results. The skill level in our team varied a lot, but the trainer was able to handle this flexibly."
"I finally understand how to deploy AI agents for our processes. The practical exercises were an eye-opener. By the end of the day I had a working helpdesk agent ready."
Everything about the Copilot Studio training
The training is suitable for IT professionals, business analysts, process owners, Power Platform makers and anyone who wants to build AI agents for their organisation. No programming experience required. In the preliminary conversation we align the level with the participants.
Basic knowledge of Microsoft 365 is desirable. Experience with Power Automate is an advantage for the Hands-on day, but not mandatory. Programming experience is not required for either variant.
The new one, unless your situation calls for something else. Microsoft rebuilt Copilot Studio in June 2026. You describe your agent in plain language instead of drawing conversation flows, and instructions, knowledge, tools and skills sit together on the Build tab rather than spread across nine configuration tabs. Skills, memory and connected agents are new as well. The classic experience remains fully supported and is reachable through a toggle on the home page; if your organisation already runs classic agents, we cover those just as thoroughly. In the preliminary conversation we agree where the emphasis should lie.
Hands-on exercises require access to Copilot Studio. A free trial account is sufficient for the training. Participants receive a setup guide in advance. We are happy to advise on licence options for production use after the training. The rates per action and two worked examples are in the article Copilot Studio: from idea to working agent.
Microsoft Copilot Studio is the successor to Power Virtual Agents, rebranded in November 2023. The platform has since been fundamentally expanded: from a chatbot builder to a full agent platform that works with Microsoft 365 Copilot and supports autonomous agents. In June 2026 the redesigned building experience was added on top. If your organisation has experience with Power Virtual Agents, the transition is straightforward. We cover the differences explicitly in the training.
Power Automate automates processes and workflows triggered by an event (an email, a new file, a schedule). Copilot Studio builds conversational agents that respond to user questions via chat. The two tools complement each other: from within Copilot Studio you can trigger Power Automate flows as an action, so an agent not only answers but also does something.
Everything you build in Copilot Studio runs on one of three engines, which Microsoft calls the harness. The standard engine is meant for rule-based agents and repeatable conversations, the GitHub Copilot engine for reasoning-heavy, multi-step work, and the Copilot chat engine for extending Microsoft 365 Copilot Chat with your organisation’s knowledge. The GitHub Copilot engine also creates and edits Word, Excel, PowerPoint and PDF files itself, and supports skills and memory. The choice determines what the agent can do and how usage is billed: on the GitHub Copilot engine building and testing already count, while on the standard engine the meter starts only after publishing. You pick the engine when you create the agent; it cannot be switched afterwards. We cover the choice in the training and go through it in the intake conversation. The differences are set out in Copilot Studio: from idea to working agent.
Every new agent on the standard engine starts with generative orchestration. The agent then chooses for itself which topic, which tool and which knowledge source to use, based on the descriptions you give them, asks for any information it still needs, and composes the answer. With classic orchestration the agent selects a topic by matching trigger phrases you write, and you determine every question and every answer using question and message nodes. Generative orchestration works better for open questions; classic orchestration is what you choose when the sequence is fixed and must be auditable, for example for identity verification, approvals or legally prescribed wording. Both are covered in the training, and we help you choose per use case.
Yes, all our trainings are available in-company, at your location or online. We fully tailor the content, use cases and exercises to your organisation, industry and role group.
Yes. In addition to the training we also offer Copilot Studio agent building and coaching. We build agents for you, guide your team while building and publishing, or work in co-creation so your own people gain knowledge transfer at the same time. Get in touch for a no-obligation conversation about the possibilities.
In production Copilot Studio works with Copilot Credits: a consumption model that depends on the number of conversations and the capabilities used. Microsoft offers subscriptions with prepaid Credits and a pay-as-you-go option through Azure. If you hold a Microsoft 365 Copilot licence, certain scenarios (such as extensions for M365 Copilot) are already included. We discuss the licensing strategy for your situation in the intake conversation. The rates per operation and two worked examples are in the article Copilot Studio: from idea to working agent.
Book the Copilot Studio training for your team
Fill in the form on our contact page or call us directly. We will get back to you within two working days to discuss your wishes. No obligation.