Private AI that runs on your servers or your cloud
Use AI while keeping your data under your organization's control.
Private AI means running AI where your organization controls the data, on your own servers or in your own cloud account, instead of sending documents to a public AI service. BUG MAI can configure Private AI with access control and logs, and chooses the setup with you after reviewing your data in discovery.
Private AI: AI that runs under the organization's own control, for example on its own servers or in its own cloud account.
It is used when documents are sensitive or internal rules do not allow sending data to an outside service.
§ 01 — Steps
How it works
- 01
Classify the data
You provideWhich documents and systems the AI may touchYou getA written data map, including what must never leave - 02
Choose where it runs
You provideYour servers, cloud account or any constraintsYou getA recommended setup with the trade-offs explained - 03
Pilot inside your boundary
You provideA sample dataset and people to try itYou getResults measured against the target, with data staying where agreed - 04
Operate and decide
You provideYour go or no-goYou getAn operating plan if you go
§ 02 — Connects to
What it can connect to
What we can connect to depends on the exports or interfaces your systems offer. We check this during discovery.
- On-premise servers
- Your cloud account
- Login and identity systems
- Document stores
- Existing apps
§ 03 — Safeguards
Data and control
Each item below is something that can be configured, as agreed in the scope.
- Where it runs
- Can be configured to run on-premise or in your own cloud account.
- Access
- Can be configured with role-based access to who can use the AI and which documents they can see.
- Logs
- Can be configured to record who asked what, and when.
§ 04 — Compare
On-premise, your own cloud and public AI services
| Topic | On-premise | In your own cloud | Public AI service |
|---|---|---|---|
| Where data is processed | On servers you own | In a cloud account you control | In the provider's service, possibly abroad |
| Who runs the hardware | You supply and maintain servers | The cloud provider runs the hardware | The AI provider |
| Best when | Data must stay inside your building or network | You want control of data without running servers | Documents are not sensitive and outside processing is acceptable |
§ 05 — Related
Related services and pages
§ 06 — FAQ
Common questions
Do we need our own servers?
Not necessarily. Private AI can run in your own cloud account. We recommend a setup after looking at your data and constraints in discovery.
Is it as capable as public AI services?
The capability of models that run privately can differ from public services. We test on your own task in the pilot and tell you the limits plainly.
Does any part use an overseas provider?
If a design uses an overseas AI provider, we say so in writing before work starts, and you decide.
§ 07 — Get started