Defense in depth, top to bottom
Controls span infrastructure, application, and the AI pipeline.
Tenant isolation
Every row of every table is scoped by tenant, with row-level security enforced in the database and privileged access gated against the tenant boundary.
Encryption in transit and at rest
TLS 1.2+ on every client and server connection. AES-256 encryption at rest for all tenant data via managed Postgres. Integration credentials stored with separately scoped encryption keys.
Verified integration delivery
Inbound events from connected integrations and payment providers are cryptographically verified before we act on them. Unsigned, tampered, or replayed events are rejected.
Hardened web surface
Strict browser security headers — content security policy, transport security, and anti-framing protections — enforced platform-wide. Bot protection on every authentication page.
Prompt-injection defense
User input is sanitized before reaching any AI model. No transcript content is interpolated into a prompt in raw form. We regularly test for the latest prompt-injection patterns.
Continuous monitoring
Real-time error and performance monitoring across the application and the AI pipeline. Anomalies trigger immediate on-call alerts.
Your conversations stay yours
Sensitive data is removed before it is stored, and your content is never used to train an AI model.
Sensitive-data removal before storage
Our redaction layer automatically detects and redacts common payment, banking, and personal contact identifiers from call transcripts before they are stored or analyzed. It is on by default — no opt-in required.
AI providers do not train on your data
The AI providers that process your call content — Anthropic, OpenAI, and AssemblyAI — contractually exclude API data from model training under their commercial terms. Your call content is processed, not learned from.
Data rights, on demand
Right to access, rectify, delete, and export is supported through the admin portal. Customer-initiated deletion is processed within 30 days. GDPR, CCPA, and US state privacy laws are honored regardless of where you are based.
No advertising or tracking cookies
We use essential and preference cookies only today — authentication, session, and CSRF. No third-party advertising trackers or behavioral ad networks; any future analytics would be consent-gated.
Built to stay up
Managed infrastructure, automated backups, and a resilient AI pipeline.
Managed infrastructure
The platform runs on Vercel’s global edge network and a managed Postgres cluster with automated daily backups and point-in-time recovery. We benefit from the operational maturity of these providers rather than reinventing it.
Resilient AI pipeline
AI calls retry automatically on provider errors. Long-running analyses run on a durable queue, designed so retries do not duplicate work and a failed worker does not lose jobs.
Backups and continuity
Automated daily backups with point-in-time recovery; full retention and recovery detail is in the security packet. The application is stateless and re-deployable in minutes, and webhook events are persisted before processing so a downstream outage does not silently drop data.
Reviewing us formally?
Request the security packetWhat we attest to today
SOC 2 Type II readiness in progress. GDPR and CCPA-ready. PCI handled by Stripe. We list frameworks honestly — what is in place, what is in progress, and what is handled by our processor.
SOC 2 Type II
In progressReadiness program in progress, targeting Q4 2026. Control documentation and policies are available under NDA today.
GDPR
ReadyData processing agreement available on request. Data subject rights honored. Cross-border transfers covered by Standard Contractual Clauses.
CCPA & US state privacy
ReadyCalifornia, Colorado, Connecticut, Virginia, and Utah privacy rights supported. No sale or share of personal information.
PCI DSS
Handled by StripeWe never store your credit card information. All payment processing is handled by Stripe, a PCI DSS Level 1 service provider — the highest level of certification in the payments industry. Every billing event is cryptographically verified before we act on it.
Controls you can configure
The same controls a security team expects from a modern SaaS — and a few designed for the realities of sales conversations.
SAML 2.0 single sign-on
Per-tenant SAML SSO configurable from the admin portal. Works with Okta, Microsoft Entra, Google Workspace, and any standards-compliant IdP. Available on the Enterprise plan.
Multi-factor authentication
Multi-factor authentication with trusted-device support. A verified second factor is required for privileged actions, with tenant-level enforcement available.
Granular roles and permissions
Admin, manager, and rep roles with a fine-grained permission system. Managers are always scoped to their team. Permission grants are logged.
Append-only audit log
Every privileged action — admin changes, impersonation sessions, billing events, integration credential changes — is logged with full context and available for compliance review on request.
Impersonation controls
Support access requires explicit impersonation sessions, time-bounded and read-only. Destructive operations are blocked during impersonation, and every session is recorded in the append-only audit log.
Compliance rules engine
Define disclosure, scripting, and regulatory rules per tenant. Every call analysis is checked against your rule set in real time. By-rep and by-rule dashboards for legal and risk teams.
