Skip to main content

Enterprise Deployment

Enterprise Deployment Options in Arato

Updated over a month ago

Arato.ai offers two enterprise self-hosting deployment models:

  1. Hybrid Deployment - You retain full control of your data by connecting Arato to your own Elasticsearch deployment, while using Arato Cloud for application services and metadata.

  2. Fully Managed Deployment on Your Cloud (AWS) - Arato is deployed entirely within your AWS environment. All resources live in your infrastructure and never leave your cloud account.

Both options are designed to provide flexibility, security, and transparency based on your organizational and compliance needs.

1. Hybrid Deployment: Use Your Own Elasticsearch

This option allows you to self-host the data layer using your own Elastic Cloud Serverless deployment, while the Arato application layer runs in the cloud. You retain ownership of all sensitive data (e.g., datasets, logs, prompt versions), and only metadata is stored in Arato Cloud.

This setup is ideal for organizations that want full control over their data without needing to manage the Arato application infrastructure.

Arato uses Elasticsearch (or OpenSearch) as the central data store for:

  • Prompts and prompt versions

  • Datasets and experiments

  • Log events and analytics

Arato Cloud accesses your Elasticsearch cluster to read/write data, while keeping user identities, configurations, and metadata in its managed layer.

Setup Instructions:

1. Create an Elastic Serverless Deployment

  • Visit Elastic Cloud Serverless.

  • Create a new deployment under the Observability template.

  • Once the deployment completes, proceed to the next steps.

2. Retrieve Connection Credentials

  • Go to Project Settings → Integrations.

  • Click Connection Details (top-right corner) and copy your Elasticsearch endpoint.

3. Create an API Key

  • Navigate to Project Settings → Management → API Keys.

  • Click + Create New API Key:

    • Name: arato-api-key

    • Permissions: Full Access

    • Copy the encoded API key after creation (you will need this only once).

4. Provide Configuration Values

  • Please send us the following information:

    • Elasticsearch Endpoint

    • Encoded API Key with full permissions

  • We will add these to your Arato account configuration. Once complete, your Elastic deployment will be used as the primary data lake for all Arato functionality.

💡 A screen-sharing session is strongly recommended to verify connectivity and correct configuration. You can book one here.

2. Fully Managed Deployment on Your AWS Infrastructure

In this model, Arato is deployed into your own AWS environment via AWS Organizations. All components are sandboxed in a dedicated AWS sub-account, isolated from your core infrastructure. Arato’s enterprise team handles the full deployment and ongoing management.

To run Arato entirely within your cloud, we use AWS Organizations, which allows the creation of isolated sub-accounts with separate policies, billing, and resource boundaries.

We offer a Managed On-Premise Deployment model:

  • Our team fully manages the deployment and maintenance of Arato.

  • Resources are isolated in a new AWS account under your AWS Organization.

  • No access to your primary AWS infrastructure is required or possible.

Security Model

  • All Arato resources are sandboxed in a dedicated AWS sub-account.

  • The deployment follows the principle of least privilege: Arato has access only to the resources it installs.

  • No visibility into or access to other VPCs, databases, or services in your primary AWS account.

Setup Instructions:

Create a New AWS Account in Your Organization

Did this answer your question?