# How to run AI-built tools inside your own cloud

> Substrait runs the whole tool, front end, back end and data, inside the company's own cloud account or data centre. Only policy and deployment instructions cross the boundary; data never leaves.

URL: https://substrait.build/problems/keep-ai-built-tools-in-our-own-cloud/  
Last updated: 2026-11-06

**The question:** Our data cannot leave our own cloud tenant. Can AI-built tools run inside it?

Yes. Substrait runs the whole tool, front end, back end and data, inside your own cloud account or data centre. Substrait sends policy and deployment instructions only, and your data never leaves your perimeter.

## Why this matters

Many AI builders run the tool on the vendor's own servers.
Your data then leaves your perimeter, and the tool runs where you cannot see it.
For a company with customer data, regulators or an infosec team, that ends the conversation.

## What Substrait does

**The whole tool is moved in, not proxied.**
Substrait runs the front end, the back end and the data inside your cloud account or data centre.

**Only instructions cross the boundary.**
Substrait's control plane sends policy, connector grants and deployment instructions.
Your environment sends back status and decisions only, never data.
The link is outbound from your side, so Substrait never opens a connection into your network.

**The unmodified tool.**
Substrait runs the tool your builder produced, as it is.

**Your keys.**
Substrait supports encryption keys that you manage.
Each organisation also uses its own AI model key, so your spend and your data terms stay yours.

## Four ways to run Substrait

| Setup | Where it runs |
|---|---|
| Substrait's cloud | Both the control plane and your tools on Substrait's cloud. The standard setup. |
| Hybrid | The control plane on Substrait's cloud; your tools and data in your own cloud account |
| Self-hosted | The control plane and your tools both in your environment |
| Air-gapped | Everything in your data centre, with no outbound connection |

## Questions and answers

### Does 'your own environment' mean on-premise only?

No. Your own cloud account on AWS, Google Cloud or Azure is your own environment. It runs under your identity and access rules, your encryption keys and your bill. A data centre you run is also your own environment.

### Does Substrait keep a copy of our data?

No. When Substrait runs in your environment, the tools and their data stay in your account and region, and Substrait holds no copy.

### Can we start on Substrait's cloud and move later?

Yes. Many companies start on a standard setup on Substrait's cloud, and move into their own cloud when they want. Some stay on Substrait's cloud for good.

## Related

- [On Substrait's cloud, or inside yours](https://substrait.build/capabilities/your-own-cloud/): Start on a standard setup on Substrait's cloud with nothing to install, or run Substrait in your own cloud account or data centre. In your environment, the whole tool and its data stay inside your perimeter.
- [Security, configured exactly for your organisation](https://substrait.build/security/): Security, Substrait's third pillar. Every release is checked before production, and each organisation sets how strict that check is: its own risk tiers, blocking rules, severity names, approvers and deadlines, with a full audit trail nobody can alter.
- [Company data, granted by its owner](https://substrait.build/capabilities/governed-data-access/): A tool on Substrait reads company data only after the data owner approves it, for a set time, through one gateway that holds the credential, checks every call and records every read.
- [Solutions for every starting point](https://substrait.build/solutions/): Substrait serves three kinds of company: teams already building with AI, companies ready to start whose work still runs on spreadsheets, chat and paper, and companies stuck on a closed legacy vendor system. Each has its own way in.