Platform Data Security Overview
Last updated on: 2026-07-08
The general philosophy is that all customer data is treated with the utmost care and attention, and this document aims to capture how the security and privacy around user data is treated.
1. Opper Platform
The Opper platform is deployed in Stockholm, Sweden region within Amazon Web Services (AWS). In order to access the Opper platform, users need to be authenticated.
The platform consists of a web application and an API service. Access to the platform requires being authenticated with a valid user, or having a valid API key.
Opper is using auth0 as an external authentication provider, which handles user flows such as sign up, login and forgotten password.
2. Data Privacy
Opper does not use customer data to train models.
The data sent to the model provider consists of several parts. When making a request through the gateway, the prompt together with the input of the request is sent to the selected model. If the request uses one or more indexes, matching snippets from the index are additionally passed to the model. The snippets passed are based on what is found in the index based on the input.
What Opper stores depends on the plan:
- Gateway (default): request and response content is not stored. Opper only stores metadata about each call, such as model, token counts, latency, status and cost, which powers usage analytics and billing.
- Control Plane: tracing is enabled, and request inputs and outputs are stored in an encrypted Amazon RDS database in order to provide traces, sessions and evaluations.
- Enterprise: data retention is configurable, including zero retention.
Model providers may retain request and response content for abuse monitoring depending on the route. See the sub-processor list for per-provider retention detail.
For a list of model providers in use, please see the model directory: /models.
3. Indexes
The Opper platform provides functionality to manage uploaded documents in the form of indexes. After creating a new index, the user can upload one or more documents to the platform. These indexes can then be associated with one or more functions, and used as context when providing an answer to the question asked, together with the prompt in the function. Data is indexed and vectorized. This data is stored in AWS RDS.
4. Data storage
Application data and uploaded files data is stored in AWS RDS. Encryption at rest is enabled with an AWS KMS managed key.
Uploaded data is stored in a private S3 bucket, and the objects are segregated per organization. The S3 bucket contents is encrypted using Server-side encryption with Amazon S3 managed keys (SSE-S3).
5. Data transmission
Service to service communication is limited using a VPC and specific public and private subnets exist to segregate public and private communication.
Communication between the API service and S3 is using the AWS SDK, and is subject to the TLS encryption set by AWS. All communication between the browser and the Opper platform is encrypted using TLS.
6. Lifecycle, backups and deletion
When an index is deleted, all associated files and indexes are deleted.
When a function is deleted, all associated events are deleted.
Backups are stored in the AWS Backup service and are encrypted using an AWS managed key. Only Opper engineers have access to restore backups if needed, no other personnel at Opper have access.
Realtime backups are enabled for Aurora and S3 for point-in-time-recovery. Daily snapshots are retained for 5 weeks. Weekly snapshots are retained for 14 months.