Copilot Training or Claude Training: Which Should Your Team Do First?
Part of AI Workshops
Copilot or Claude training first? It depends on the licences you already pay for and where the work lives. How to decide, and what good training covers.
In short
Train first on the tool your team already has and will use every day. If the working day is Outlook, Teams, Word and Excel and you hold Microsoft 365 licences, Copilot training is the sensible start, because Copilot works inside those apps under the permissions your tenant already sets. If the work lives in systems outside Microsoft, such as accounting, ecommerce, a CRM or industry software, or people need to build repeatable workflows of their own, Claude training usually pays off sooner. Either way, the training that changes behaviour is the same: each person picks one recurring job, times it, builds it with the tool on their real work, and checks the time again afterwards.
Teams in New Zealand and Australia ask this one a lot, usually because they already pay for Microsoft 365 and someone has also been using Claude and getting good results. Now the owner wants everyone trained and is not sure which tool to train on first.
The honest answer is that it depends less on which tool is better and more on two things you already know: what you pay for, and where the work lives.
Start with what you already pay for
If your team is on Microsoft 365, Copilot is probably already there. Microsoft includes Copilot Chat with eligible Microsoft 365 licences, and it shows up in Outlook, Teams and the Copilot app. There may well be people in your business using it right now.
That changes the question. Before buying anything or booking any training, find out what people are already using AI for. Those uses are your training plan: they tell you which jobs matter and who is already ahead.
The licence level matters too, and it is easy to miss. Microsoft's own documentation draws a clear line. Copilot Chat without the Microsoft 365 Copilot add-on answers from web data, and if you want it to work with your documents you have to upload them or have them open in Outlook or Teams. With the add-on licence, Copilot reaches your organisation's emails, files, meetings and calendars through Microsoft Graph, scoped to what each person already has permission to see, and works inside Word, Excel, PowerPoint and OneNote.
So "Copilot training" for a team on the included Chat is a different course from Copilot training for a team with the add-on. Know which one you have before you book.
Where Copilot is strongest
Copilot's advantage is that it is already where a Microsoft-based team works. If most of the day is Outlook, Teams, SharePoint, Word and Excel, Copilot can summarise a long thread, draft a reply, turn a meeting into actions, draft a document from existing files or analyse a spreadsheet without anyone switching tools.
It also works inside the tenant your IT team already manages: the same sign-in, the same permissions, the same admin centre and the same data protection terms as your email and files. For a business with a Microsoft-shaped IT setup, that is a real reason to start there. No new vendor, no new identity system, and no argument with whoever looks after security.
For a team like that, Copilot training first makes sense. The quickest gains are usually in the inbox, in meetings and in the documents people write every week.
Where Claude fits
Claude tends to pay off sooner in three situations.
The work lives outside Microsoft. If the systems that matter are Xero, Shopify, a CRM, a booking system or your own industry software, Claude's connectors reach into them. Copilot is strongest inside Microsoft 365. A finance team that spends its day in an accounting system, or an ecommerce team that lives in its store, gets more from a tool that connects to where the work actually is.
People need to build repeatable workflows. In Claude, a job someone does every week can be written once as a skill: the steps, the checks, the format, the tone. Then it runs the same way every time, for anyone who uses it. That is the difference between a tool people ask questions of and a tool that does a defined part of the job.
You want the business's own context in every answer. Claude projects and context files tell it who you sell to, how you write and what your numbers mean. Done properly, this is what turns general answers into answers about your business.
None of that makes Copilot the wrong choice. It means the right first tool depends on the work.
A simple way to decide
Ask three questions.
- Which licences do we already hold? If you have Microsoft 365 Copilot add-on licences sitting partly unused, start there. Unused seats are the most expensive kind.
- Where does each team's work actually happen? Map it roughly: inbox and documents, or another system. Teams whose work is mostly in Microsoft apps start with Copilot. Teams whose work is mostly elsewhere start with Claude.
- Who will build, not just ask? The people who will turn their own jobs into repeatable workflows are usually the ones who get the most from Claude.
Most businesses end up with an answer like "everyone gets the tool that is already included, and the heavy users and the teams working outside Microsoft get Claude". Usage is never even. A few people will use AI for hours a day, most for minutes. Train the heavy users on the tool they will get the most from.
Our Copilot vs Claude comparison goes into the detail: what each includes, where each processes data, and which kinds of team each suits.
What good training looks like, either way
The tool matters less than the way it is taught. Training that changes how a team works looks much the same for Copilot or Claude.
It is on real work. Not example prompts about planning a holiday. Each person brings one job they repeat every week and builds it with the tool, in the room.
It starts with the rules. What may go into the tool and what may not, which plan you are on and what it does with your data, and who to ask. One page, written before the first session.
It is measured. Each workflow is timed before anything is built and again once it runs. Without a baseline, nobody can say whether the training worked, only whether people enjoyed it.
It is followed up. A single day starts the habit. The habit sticks when someone checks a week or two later that the workflow is still being used and the time has come down.
We run Claude training days for teams on exactly that pattern, and our workshops cover the same ground for teams still choosing. If your team is on Microsoft 365 and Copilot is the right first step, we will say so.
Questions people ask
Can we use Claude inside Microsoft Copilot?
In some Microsoft 365 licensed experiences, yes. Microsoft lists Anthropic as a subprocessor and offers Claude models in Copilot, though not to every user by default. Copilot vs Claude covers where each runs and under whose terms.
Is free Copilot Chat enough to train a team on?
It is a reasonable place to start learning to prompt well. Microsoft's own documentation says Copilot Chat without the add-on licence answers from web data, and you have to give it your work content yourself. Training on it is useful, but it will not show people what Copilot does with their email, files and meetings.
Should we train on both Copilot and Claude?
Only if people will use both. Most teams do better learning one tool properly on their real work, then adding the second for the people and jobs that need it. A day on two tools tends to leave people competent in neither.
Read next on this site
Written by

Founder, Sentry AI
James Oldham founded Sentry AI in Auckland and runs it: an AI engineer who moves teams across New Zealand and Australia onto Claude and ChatGPT, teaches it in the room, and does the engineering on the work described in these posts.
- Founder and sole director of Sentry AI Solutions Limited (NZ company 9399152, incorporated 20 January 2026)
- Built custom AI systems for enterprise teams, including the NBA, before founding Sentry AI
- Delivered the voice, knowledge-graph and agent systems in the named case studies on this site, across recruitment, healthcare, retail and financial services
- Taking Sentry AI through its own ISO 27001, SOC 2 and ISO/IEC 42001 programme
- Auckland University of Technology
Ask an AI about this page
Opens the assistant with this page loaded: read it, summarise it, cite it.