Somebody in your leadership team has asked which AI tool the business should settle on, and the question usually arrives one step early. The tool matters less than the account it runs on, the data it can reach and the rules your people work to, and once those are settled the choice between Claude, ChatGPT and Copilot gets much smaller.
Cardonet is Certified in the Claude Partner Network, and we are a Microsoft Solutions Partner for Microsoft 365 Copilot. We deploy both, and we put clients on ChatGPT or Gemini where those fit the work better, so what follows is the test we use rather than an argument for any one of them.
The line between them has blurred
A year ago these were three separate products. Today Microsoft offers Anthropic's Claude models inside parts of Microsoft 365 Copilot, starting with its Researcher agent and Copilot Studio, with an administrator deciding whether they are switched on. Claude and ChatGPT both connect to Microsoft 365 and Google Workspace on their business tiers, once an administrator allows it. So "we are a Microsoft business, therefore Copilot" is no longer a complete answer, and neither is "our people like ChatGPT, therefore ChatGPT".
What still separates them is where each one sits in your day, who administers it and what it can open.
Five questions that decide it
1. Which account will your people be on?
Every one of these tools comes in a personal version and a business version, and the terms differ between the two.
| ChatGPT | Claude | Gemini | |
|---|---|---|---|
| Personal plans, where what people type can be used for training unless someone switches it off | Free, Plus and Pro | Free, Pro and Max | Personal Gemini accounts |
| Business plans, not used for training by default | Business and Enterprise | Team and Enterprise | Gemini in Google Workspace |
Microsoft 365 Copilot belongs with the business tiers, which are not used for training by default according to each vendor's published business terms. On the personal side, Claude's position comes from the consumer terms Anthropic introduced in 2025. So compare tiers, because your people's personal ChatGPT account and a company Claude Team seat are further apart than ChatGPT and Claude are.
2. Does the tool need to read your documents?
Here Copilot is a different kind of thing. Microsoft 365 Copilot, the paid version, searches and answers from whatever the signed-in person is allowed to open across email, Teams, SharePoint and OneDrive. Copilot Chat, which comes with most Microsoft 365 plans, works from the web and from whatever you paste or upload, and does not search your work data. Copilot Chat or Microsoft 365 Copilot sets out the difference in full.
Claude and ChatGPT reach your documents through connectors an administrator turns on, and they read what the connected person can read. Whichever route you take, the boundary is the same: a connected tool inherits the person's permissions. If those have never been checked, read what Copilot can already see before you decide, because the answer applies to all three.
3. What work is it for?
For a first draft of an email, a rewrite of your own paragraph or a summary of a public document, the three perform close enough that the difference will not decide anything. The differences show up at the edges, and they show up on your work rather than on published benchmarks.
Copilot's advantage is that it lives inside the applications your people already have open: a summary of the meeting they just left, a draft reply in Outlook, a first pass at a spreadsheet formula. Claude tends to suit long documents, careful reading and writing where tone matters, and work that is becoming a repeatable process.
ChatGPT has the widest familiarity, since most of your people have used it at home, which makes moving them onto a business account the shortest step of the three. If your business runs on Google Workspace, Gemini is already included in the business plans, and that is worth testing before you buy anything else.
4. Who will administer it?
Copilot sits inside the Microsoft 365 admin your IT team already runs, with the same sign-in, retention settings and audit trail. Claude and ChatGPT business tiers each bring an admin console of their own. Although that is no reason to rule them out, it does cost something: another place to add and remove people, another register entry and another step in your leaving process that has to happen every time.
5. What does it cost in total?
The seat price is the smallest part and the easiest to compare, so check it on the vendor's own page when you buy. The larger costs sit around it. Before a tool reads your documents, the permissions need tidying. Every tool needs your people shown what it is for, and something has to tell you when the output starts to drift.
Two tools for two groups can cost less than one tool that half your people never open. What Microsoft 365 Copilot costs works through all of this for Copilot.
Where each one tends to fit
These are starting points, and the test on your own work settles it.
- Microsoft 365 Copilot fits a business that lives in Outlook, Teams, Word and Excel, where the help people want is inside those applications and where you are prepared to check permissions before switching it on.
- Claude fits work built on long documents and careful writing, teams that want one assistant across Microsoft 365 and other systems, and work that is on its way to becoming an automated process. It is also available inside parts of Copilot, which can settle the question for a Microsoft business. For the business plans, Claude Team or Claude Enterprise has the detail.
- ChatGPT fits a business where most people already use it personally, because the move to a company account is quick and the habits come with it.
- Gemini fits a Google Workspace business, since it is already in the plan you pay for.
Two of them side by side is often the right answer, for example Copilot for everyone and Claude for the team that writes your proposals. Each one you add needs its own line in the register and its own step in the leaving process.
How we make the choice with you
In our route the tool is chosen in Adopt, after Govern has settled what data may go in and who may use what. Choosing first and writing rules later is how a business ends up with a tool its policy then has to work around.
In the test itself, your first team picks five to ten pieces of their own work and runs the same tasks through each candidate on a business account. The people who do the work judge the results against how long it takes today. Whatever wins goes into the register with the reason it won, so the next time someone asks why the business uses it, the answer is written down.
If your people are already using one of these on personal accounts, that is where to start, and is it safe to use ChatGPT at work covers it. If you would rather see where you stand first, ten questions take about two minutes and send nothing anywhere.