How does Callmine protect your data?
Callmine encrypts data at rest and in transit, isolates workspaces, retains per plan, logs admin actions, and discloses every sub-processor.
TL;DR
Data is encrypted in transit (TLS) and at rest. Workspace credentials get
a second layer of application-level encryption keyed to our secret
manager. Workspaces are isolated at every database query. Admin actions
write to a 14-event audit log at /audit.
Encryption
In transit, every API call between your browser, Callmine, Gong, HubSpot, Slack, and the LLM providers runs over HTTPS/TLS. There is no plaintext-on-the-wire path.
At rest, all data in Google Cloud SQL (Postgres) and Google Cloud Storage is encrypted by Google's infrastructure-level keys. On top of that, workspace integration credentials — your Gong access key, OAuth refresh tokens, HubSpot tokens, Slack tokens — are encrypted application-side with a symmetric key stored in Google Secret Manager. A read of the database alone does not expose those credentials.
Workspace isolation
Every Callmine workspace is a Clerk Organization. Every database read
and write is scoped to a workspace_id, enforced at the query layer.
Cross-workspace data leakage is a class of bug we treat as a critical
incident — we have automated tests that fail the build if a new query
omits the scope.
Retention
Reports are retained per plan. Trial and Free retain for 30 days.
Starter retains for 90 days. Growth retains for 1 year. Scale retains
indefinitely. Older data is hidden from the UI past the retention
window; it is not deleted from storage until you request deletion via
hello@callmine.ai. See Privacy and DPA for the
full data-handling agreement.
Audit log
Workspace admins can review a per-workspace audit log at /audit.
Fourteen event types are recorded — including integration changes,
billing actions, member changes, schedule edits, and report exports —
with actor identity, timestamp, and a structured diff blob. Logs are
filterable and paginated.
Sub-processors
Callmine relies on the following sub-processors:
- ·Google Cloud Platform — compute (Cloud Run), database (Cloud SQL), storage (Cloud Storage), secret management (Secret Manager).
- ·Clerk — authentication, organization membership, session management.
- ·Stripe — billing, subscription management, payment processing.
- ·OpenAI — LLM analysis (selected models per workspace plan).
- ·Anthropic — LLM analysis (selected models per workspace plan).
The current list lives at Sub-processors. We give 30 days' notice via that page before adding a new sub-processor.
What we don't do
We do not train any model on your call data. We do not share your data with third parties beyond the sub-processors above. We do not retain analysis prompts, transcripts, or report outputs on the LLM providers' side — every analysis runs through their API tiers, which do not retain inputs for training by default.