Sub-processors
Ingest0r's infrastructure runs on Google Cloud. The services below process or store data on our behalf to run the product. This list is derived from our infrastructure configuration and is updated when the infrastructure changes.
Google Cloud
| Service | What it is used for |
|---|---|
| Cloud Run | Runs the services you use: the console, the reporting service behind Performance, the builder (for workspace admins who have it), the builder's isolated test runs of a connector, the report checks, and our public marketing site. Also runs our jobs: the sync jobs, the weekly backup and the report builds |
| BigQuery | Stores synced rows, the standard tables built from them, your reports' tables and weekly backups (Managed tier). Report queries run as BigQuery jobs in a separate Google Cloud project of ours, reading the same data |
| Firestore | Workspace metadata: connected sources, account IDs, sync history, invites, and logins (email address and a bcrypt hash of the password). Also what your team builds: connector definitions saved in the builder (never credentials) and each report's title and SQL. Daily backups of this metadata |
| Secret Manager | Access tokens for sources that need one, the API credentials of connections to connectors your team built, and our own service secrets |
| Cloud Storage | The builder's daily usage counters per workspace (workspace identifiers and counts only) and, while a builder session is open, its working state (the conversation, and the draft connector or report with its test results) so the session survives a builder update. Keys typed in the builder's key dialog, and your data's rows, are never saved; anything typed into the chat is. Deleted when the session ends; otherwise Cloud Storage deletes it automatically, normally within 48 hours of its last save |
| Vertex AI | Generates the builder's answers. While a workspace admin uses the builder to build a connector or a report, the model receives what they type (including anything they paste), the text of the public documentation pages the builder reads, and the results of its checks. For a connector: the table names and row counts of test runs made with the admin's API key (never the rows themselves, never the key), the error messages of those runs with the key removed (at most 700 characters each; an error can quote part of what the API answered), and the status, a few headers and up to 20 KB of the body of unauthenticated requests to the API's own host. For a report: the workspace's table and column names and types, and for each test its row count, each column's null rate and how much it reads (never a value). Not used while your syncs run or while your report tables refresh |
| Cloud Logging | Service and audit logs, which can include IP addresses and workspace identifiers. A copy of the audit logs is kept for 400 days |
| Pub/Sub | Passes console requests to the sync job: sync, connect, pause, resume and remove requests (workspace identifiers, the source details you enter such as an account, property, spreadsheet or site ID, a built connector's settings, and a client name — never a credential's value), team changes (the email addresses of the invited or removed person and of the teammate acting), invite acceptance and password changes (a bcrypt hash of the new password), connector definitions saved in the builder, and reports' titles and SQL. Messages are kept for at most 6 hours. Never synced rows |
| Cloud Scheduler | Starts the hourly sync, the hourly report builds and the weekly backup |
| Cloud Load Balancing and Certificate Manager | Receives your HTTPS connection to ingest0r.ai and terminates TLS |
| Artifact Registry | Stores our application images (no customer data) |
Warehouse tier
If your workspace uses the Warehouse tier, synced rows and reports are written to BigQuery in your own Google Cloud project, in the region you choose. Ingest0r's own project keeps only the workspace metadata listed above.
Questions
To ask about this list, email support@ingest0r.ai.