How can a business use AI without handing its data to a public chatbot?
Put the AI somewhere the business controls: a private assistant in your own cloud account, or a business plan of a managed assistant under company sign-in and admin settings. Then decide which documents it may search, who can use it and what gets logged.
Three ways to set it up
| Option | Where it runs | Fits |
|---|---|---|
| Managed AI assistant | The vendor's business plan (Claude, Copilot, Gemini or ChatGPT), under your company sign-in and admin settings | A team that wants a familiar assistant with company controls |
| Private assistant | Your own AWS or Google Cloud account, reaching models through zero-retention gateways | Search over your own documents, from $1,150 a month on Managed Cloud & AI for Business |
| Full private AI | Your own cloud, with agents and retrieval workflows built for your processes | Repeat work, such as intake, quoting or reporting, run on your own data |
Decide these before you switch it on
- Which folders, mailboxes and systems the AI may read, and which it may not.
- Who may use it, through company sign-in with multi-factor authentication.
- What is logged, and who reviews it.
- What it may send, change or delete on its own, if anything.
- An acceptable-use policy your team signs.
Who builds it
On Managed Cloud & AI for Business, Mike Golliet leads AI implementation and agent workflows, and Keegan Justis, our Chief Technology Officer, leads cloud security architecture. The environment sits behind the same security operations centre as the rest of your systems.