You run Microsoft 365Reaches your documents, email and meetings
Strongest at
Work built from documents and email you already hold
The catch
Inherits the state of your content and permissions. Rewards a tidy tenant, punishes a messy one.
Which AI platform should your business use?
Almost certainly the one you already pay for. Here is the honest read on where these platforms genuinely differ, and where the difference is marketing.
The short answer
Microsoft Copilot, ChatGPT, Claude and Google Gemini do most of the same things. For the ordinary work of a professional services business, drafting, summarising, research and first-pass documents, the differences between them are smaller than the marketing suggests and smaller than the cost of switching. What separates businesses getting a return from those that are not is almost never which platform they chose. It is whether anyone decided what the tool is for, who owns the output, and what gets checked before it reaches a client.
So the useful question is not which platform is best. It is what your business already pays for, and what it would take to get value from that.
Where the platforms genuinely differ
Four differences are real and worth knowing. Almost everything else on a vendor comparison page is not.
What it can already see
The largest real difference, and the one nobody markets well. Copilot can reach the documents, email and meetings inside your Microsoft 365 tenant. Gemini can reach your Workspace. An outside tool can be connected to the same content, but someone has to configure and maintain that. If the work depends on your own content, the difference in setup outweighs almost everything else.
What your plan actually includes
Increasingly the assistant arrives inside a subscription you already hold, but not uniformly. Microsoft sells Copilot both bundled into some business plans and as a paid add-on to others, so two firms on Microsoft 365 can be in quite different positions. Read the licence before you compare it against anything.
How your content is handled
This differs by tier more than by vendor. The business and enterprise plans of the major platforms treat what you type differently from the free consumer app carrying the same name. Most businesses have never checked which one their staff are signed into.
Where it looks things up
Only matters if you care about being found. These systems search different indexes, so being visible to one is not being visible to all.
Where the difference is mostly marketing
Everyday writing quality. Summarising a meeting, drafting a proposal, tidying an email: all four are good enough that the choice will not decide whether the work improves. Benchmark differences are real in a laboratory and rarely survive contact with an ordinary Tuesday.
Feature lists, too. A capability one platform announces tends to appear on the others within a quarter or two. Choosing on a feature comparison usually means choosing on a gap that closes before the rollout finishes.
The four, at a glance
What each one suits, and what it costs you, in the terms that decide a real choice rather than a feature count.
You run neither, or want a second opinionWorks only from what you give it
Strongest at
Reasoning and drafting from nothing
The catch
Knows nothing about your business unless someone tells it, every time.
You run neither, or write long documentsWorks only from what you give it
Strongest at
Long documents and holding a consistent voice
The catch
Fewer of your staff will have used it, so a rollout needs more deliberate onboarding.
You run Google WorkspaceReaches your Docs, Gmail and Drive
Strongest at
Work built from Docs, Gmail and Drive
The catch
Least written about, so you will find less help when you get stuck.
Read the first line of each before the rest. What you already run decides more of this than everything else put together.
Which one are you on?
Find your situation below. Each one leads to what to do about it.
Already paying for Microsoft 365
The most common position, and the one where value most often goes missing. See getting value from Microsoft Copilot.
Weighing Copilot against ChatGPT
A fair question with an answer that depends on your work, not on which is better. The short version: if the work runs on documents you already hold, the tool that can reach them starts ahead. If it does not, you are choosing on preference.
Running on Google Workspace
The same logic as Microsoft: the assistant that can see your own content starts ahead. See Gemini for Workspace businesses.
Using ChatGPT or Claude directly
Common where one person got started on their own, usually on a personal login. The first job is moving that onto something the business controls, before anyone worries about which tool is better. See ChatGPT for business.
How we help
We work in the stack you already run
What you already pay for
Most businesses hold licences they are not using. The first job is finding out what is already included before anything new gets bought.
One job, made repeatable
A named workflow that runs the same way twice, in the tool your team already opens, rather than a pilot beside the work.
Rules people can follow
What client information is allowed near which tool, who owns the output, and what gets checked. A page, not a framework.
The team, not just the tool
Training on your own work rather than generic examples, so the people who have to use it can see whether it is genuinely faster.
Questions we get asked
Which AI platform is best for a small or mid-sized business?
Usually the one already included in something you pay for, though what that means varies. Microsoft sells Copilot bundled into some business plans and as a paid add-on to others. Google includes Gemini across the paid Workspace tiers, but Business Starter is basic access and does not reach your Docs and Drive. For ordinary drafting and summarising the gap between the major platforms is smaller than the marketing suggests, and switching costs real time.
Is Copilot better than ChatGPT?
For different things. Copilot can see the documents, email and meetings already in your Microsoft 365 tenant, natively and with no connector to maintain. ChatGPT starts from what you put in front of it and can be connected to the same sources, which is a setup job somebody owns. If your work depends on your own content, that access matters more than anything else.
Do we need more than one AI platform?
Most small and mid-sized firms do not, and paying for several is a common and avoidable cost. The exception is where one tool is genuinely better at a job you do constantly, and someone owns the decision about which tool does what.
We already pay for Microsoft 365. Do we still need a separate AI subscription?
Check what your plan already includes before buying anything. Microsoft sells Copilot both bundled into some business plans and as a paid add-on to others, so two firms on Microsoft 365 can be in quite different positions. Agents published to Microsoft 365 Copilot are included for anyone licensed for it; agents that reach SharePoint or Graph connector content are billed on metered consumption, and Microsoft ships those switched off.
Is it safe to put client information into these tools?
It depends which plan your staff are signed into, and most businesses have never checked. The business and enterprise tiers of the major platforms handle your content differently from the free consumer apps of the same name. We have written about how to check your own setup.
Should we switch platforms to get better results?
Rarely. In the businesses we work with, the difference between getting value and not getting value is almost never the platform. It is whether anyone decided what the tool is for, who owns the output, and what gets checked before it leaves the business.
What about the Chinese models like DeepSeek or Qwen?
Capable, and markedly cheaper. For an Australian professional services firm the question is not capability, it is where your data goes and who can reach it. If you hold client information subject to Australian privacy obligations, that is a question you would be answering early and largely alone. Cheaper is not the constraint you are up against.
Should we run an open-source model ourselves?
Running open weights on your own infrastructure removes the data question almost entirely, which is a real advantage. It replaces it with servers to maintain, models to update and someone whose job that is. For a firm without an IT function that trade rarely comes out ahead, and the businesses where it does tend to know already.
Microsoft, Copilot, ChatGPT, Claude and Gemini are trademarks of their respective owners. Korbai is independent and not affiliated with any of them.
Get started
Start with what you already have.
Book a discovery call and tell us what your business runs. We will tell you straight whether you need anything new.