Getting started
Regions
Recalld regions are independent deployments. Separate accounts, separate keys, separate data, and no synchronisation between them. This is deliberate and permanent.
Available regions
| Region | Host | Key prefix | Status |
|---|---|---|---|
| eu | https://eu.recalld.ai | rcd_eu_ | Live |
| us | https://us.recalld.ai | rcd_us_ | Live |
What independence means
- Accounts do not roam. Signing up in Europe does not create a US account. They are unrelated records.
- Keys do not roam. A key carries its region and is rejected everywhere else. See Authentication.
- Data does not move. Memory written in one region stays there. There is no cross-region replication, failover, or migration path.
- Billing is separate. Plans, credits, and invoices belong to the account, and the account belongs to one region.
Why it works this way
Data residency stops being a claim you have to audit when the data physically cannot leave. Nothing centralises accounts across regions, so there is no path for European memory to be served from anywhere else.
Where the servers physically sit is published on the sub-processor page. Today the EU region runs in France (OVHcloud Gravelines) with backups in Belgium, and the US region in the United States (Virginia).
Choosing a region
Pick the region your data is allowed to live in, then stay there. If you need both, run two accounts and keep the agent identifiers separate in your own application.
Regions in the API
The base URL is the only thing that changes. Paths, request bodies, and responses are identical across regions. The API reference has a region selector on every example.
# Europe
curl https://eu.recalld.ai/v1/models -H "Authorization: Bearer $RECALLD_API_KEY"
# United States
curl https://us.recalld.ai/v1/models -H "Authorization: Bearer $RECALLD_API_KEY"