# Data Handling

Learn everything data related, including how data is organized and separated, privacy, residency, etc. on Confident AI.

## Data Organization and Separation

When you initially create an account on Confident AI, an **organization** and a **project** within that organization are automatically created for you. Each organization can have multiple **projects** and **users**.

> You should create a separate project for each distinct LLM use
> case, even when multiple use cases share the same
> codebase or business logic. This is because the datasets and metrics will be
> different for each use case.

![Organization, Inc. box holding Projects 1, 2, 3 to n separated by dividers, with arrows from below pointing into multiple projects](https://www.confident-ai.com/media/diagrams/in-house/confident-ai/organization-projects.svg)

*Data Organization and Separation in Confident AI*

As your organization scales, you can:

- Invite more **users** to your **project** (they have access to project data)
- Invite more **users** to your **organization** (they don't have access to project data yet)
- Create more **projects** within your **organization**

All data (test case data, metrics data, dataset data, tracing data, etc.) **is separated at the project level**. Users from one project cannot access data in projects they don't belong to, even if they are within the same organization.

### Organization

An organization represents the top-level container for all your Confident AI resources. Subscription plans and billing cycles are managed at the **organization** level. For example, if you subscribe to the [Team plan](https://confident-ai.com/pricing), all **projects** within your organization will automatically receive Team plan features.

### Project

Projects function as separate workspaces within your organization where data and access permissions are isolated. They allow different teams or use cases within your organization to maintain distinct LLM evaluation workflows.

> After completing the initial onboarding process once you've [created an
> account](/docs/setup-and-installation), you will be the sole
> **user** belonging to your **project** within your **organization**.

```mermaid
graph TD
    ORG[Your Organization] --> PROJ[Your Project]
    ORG --> USER[You]
    USER -. Access .-> PROJ
    PROJ --> DATA[Project Data]
```

## Data Retention

Data retention periods vary based on your [pricing plan](https://confident-ai.com/pricing). **Only the following data types are subject to automatic deletion**:

- Test run data (including metrics data)
- Tracing data (including metrics data)

Your datasets and prompts will be preserved as long as you maintain an active account with us. Here are the data retention periods for test run and tracing data by membership tier:

- **Free**: 1 week
- **Starter**: Unlimited
- **Team**: Unlimited
- **Enterprise**: Unlimited

For extended retention periods, please contact <support@confident-ai.com>.

## Data Privacy

All data processed and stored by Confident AI is encrypted at rest and protected by TLS in transit. We maintain SOC II and HIPAA compliance to meet the most stringent data security requirements for our enterprise customers.

SOC II certifications and HIPAA Business Associate Agreements (BAAs) are available for customers on the **Team plan** and above.

## Data Exportation

We protect against vendor lock-in by ensuring you can export all of your data should you decide to leave Confident AI. This service is available to customers on the **Team plan** or higher, with a 30-day window to submit your request after unsubscribing.

For ongoing use of Confident AI, we provide comprehensive API endpoints that allow you to query and retrieve your data for any downstream processing needs.