Run AI on regulated data inside your own cloud

Watch

Why it matters

What this protects

With personal information, account numbers or health data, the question is where the AI runs. Inside your own cloud account the model and the data stay in an environment you control, and you still get the leading models from Anthropic and OpenAI.

Each cloud has a character. Amazon Bedrock is close to turnkey but still needs an expert team. Microsoft Azure AI Foundry has strict controls and needs a team from day one. Google Cloud is fast and AI-friendly, which also makes human error easier. Hive runs both: AI work in Google Cloud, and personal and customer data in Azure.

Lindon Hayes and Igor Tomic's team runs on AWS with the Kortix agent orchestrator, which puts every connection through a sandbox or virtual machine and sets which agents, skills and connectors each person may use.

Game plan

Do it in 6 steps

1. Sort your data into classes: public, internal, confidential and restricted.
2. Decide what may never leave your own cloud account; usually that is everything restricted.
3. Choose the platform your team can actually run: Amazon Bedrock, Azure AI Foundry or Google Vertex AI.
4. Put agents behind sandboxes or virtual machines, with access set per person: which agents, skills and connectors each may use.
5. Read the vendor terms before any data goes in: training, retention, data processing agreement, a BAA for health data, and subprocessors.
6. Bring in an expert team or consultant to run it. This is not a weekend project.

Prompts

Copy and use

Replace the [bracketed] parts. Each prompt keeps you in control: nothing is sent, deleted or shared without your approval.

We hold [types of data, e.g. customer PII, account numbers, health information]. Compare running AI models inside AWS Bedrock, Azure AI Foundry and Google Vertex AI for us:
- what stays inside our own cloud account
- what our team has to manage ourselves
- which agreements we need (DPA, BAA)

Flag anything you are unsure of instead of guessing.

Check your work

You're done when

• You know which data classes you hold and where each may go.
• You've chosen a platform and who will run it.
• You've read the vendor's terms on training, retention and agreements.

Taught live in the AI Practicum on Thursday, October 15, 2026. The videos above are clips from that class.

Keep going

Related modules