Snapshot 298512
Normalized text
Scripts and page chrome removed; this is what change detection compares.
Security at Cradle Last updated: 2026-11-18 Cradle is a communication platform trusted by accounting and bookkeeping firms across New Zealand and Australia. We understand that our customers handle sensitive financial data and client relationships — and that their phone system is critical infrastructure. Security is foundational to how we build and operate Cradle. Infrastructure Cradle runs on Google Cloud Platform (GCP), which maintains SOC 1, SOC 2, SOC 3, and ISO 27001 certifications. Our infrastructure benefits from Google's world-class physical security, network security, and operational practices. Hosting regions: United States for the platform (compute and databases). Call recordings, transcripts, AI notes and message attachments are stored in Google Cloud Sydney for Australian and New Zealand customers and London for UK and European customers, and new call recordings are copied there too. The marketing website and support console run in Google Cloud Sydney. See where your data is stored. Container orchestration: Google Kubernetes Engine (GKE) with hardened node images; the reporting service runs on Cloud Run Network protection: Google Cloud Armor web application firewall with OWASP rulesets for XSS and SQL injection prevention Private networking: Internal services communicate over private networks with restricted access Encryption All data is encrypted in transit and at rest: In transit: TLS 1.3 with SHA-256 enforced across all endpoints. You can verify this at SSL Labs. At rest: AES-256 on all Google Cloud storage and databases. Storage in each Home Region (recordings, transcripts, AI notes and message attachments) is configured to use customer-managed encryption keys held in Cloud KMS. Integration and telephony credentials are encrypted at the application layer with Cloud KMS. Secrets: API keys stored as encrypted secrets in GKE. OAuth tokens encrypted in database. No secrets stored in source code. Authentication Cradle does not implement custom authentication. All users authenticate via single sign-on (SSO) through their existing Google Workspace or Microsoft 365 account. This means: No Cradle-specific passwords to manage or compromise Your organisation's MFA and security policies are enforced automatically Session anomaly detection triggers re-authentication Functional-level access control within the Cradle application Application Security We build against the OWASP Top 10 across all web, mobile, and API surfaces: Broken access control: Functional-level access control within the application, layered on top of infrastructure and SSO-level controls Security misconfiguration: Automated misconfiguration scanning across all services, with security defaults enforced at build time Supply chain security: Dependency integrity verification, removal of unused packages, and secure CI/CD deployment gates Injection & XSS protection: Parameterised queries, allow-list input validation, Content Security Policy, and escaped user output across all endpoints Insecure design: Security threat modelling incorporated into the development lifecycle AI and Data Privacy Cradle uses AssemblyAI for transcription and Google Vertex AI (Gemini) for summaries, notes, report insights and chat. Both are configured so that customer content is not used to train models. We chose Vertex AI specifically for its enterprise data governance: Processing window only: our AI suppliers hold your data only while they process it and for at most 24 hours afterwards, except where a supplier's terms require it to keep a copy for abuse and safety monitoring, for the limited period shown on our sub-processor register, with restricted access and no other use. Google holds Vertex AI working copies for no more than 24 hours and does not use them to train its models. AssemblyAI deletes audio after transcription, and we delete each transcript from AssemblyAI as soon as our own copy is stored; our account is opted out of AssemblyAI's model improvement and benchmarking programmes. No model training on customer data: Google commits to not using Vertex AI customer data for training or fine-tuning models without explicit permission. Cradle has not granted this permission. Responsible AI: Vertex AI is built on Google's AI Principles, with safety filtering, bias monitoring, and content policy enforcement applied to all inputs and outputs. Enterprise certifications: Vertex AI (Gemini) maintains SOC 1, SOC 2, SOC 3, ISO 27001, ISO 27017, ISO 27018, ISO 27701, and ISO 42001 (AI management system) certifications. Processing location: transcription runs in the United States; Gemini requests are served from Google Cloud regions worldwide through the global endpoint. Storage of the resulting transcripts and notes is in your Home Region. We do not use OpenAI or ChatGPT, and no AI service behind Cradle's features trains on customer data or retains it beyond a 24-hour processing window, other than for abuse and safety monitoring as shown on our sub-processor register. Our staff's workplace AI tool. Our staff use Anthropic's Claude (Team plan, commercial terms) with emails, messages and support tickets you send us, our calls and meetings with you, and our customer records (Business Communications, Terms of Use clause 9.10), and to diagnose faults with error reports and logs after an automated process has removed identifiers, typed text and content (Diagnostic Data, clause 9.10), and with our records of your account to reconcile payments and prepare invoices and credit notes (Accounting Records, clause 9.10), with a person approving each change and no payments or refunds made by the tool. It has no access to our error-reporting or logging systems, is never used to access recordings, transcripts, AI notes, attachments or message content held in your organisation's Cradle account, and is not used for training. Anthropic keeps that information until we delete it, then for up to 30 days, and may keep flagged content for up to 2 years. The dates each use starts are on our sub-processor register. Please don't email us recordings or sensitive information. Data Handling Access to your data: customer data is shared only with the sub-processors on our register, for the purposes listed. Cradle staff don't listen to recordings or read transcripts or messages unless you ask us to for support, or we're investigating a security incident. Every recording playback is logged. Least privilege: Role-based access control at both application and infrastructure levels, following the principle of least privilege Call recordings: Governed by our Call Recording Policy and compliant with NZ and Australian telecommunications regulations. Access to call recordings requires explicit permission. Data processing: We maintain a comprehensive data processing register documenting all data collectors, processors, purposes, and retention periods Retention: recordings, transcripts and attachments are deleted automatically under your organisation's retention settings; see the retention schedule. Monitoring and Logging Application monitoring: Sentry for error detection, crash reporting, and performance monitoring (device and diagnostic data only; no call content) Infrastructure monitoring: Google Security Centre for infrastructure and network threat detection Application scanning: Continuous vulnerability scanning of all applications Audit logging: Comprehensive logging including timestamps, user/process IDs, event descriptions, and success/failure status. Audit logs retained for 400 days. Log integrity: Logs stored in Google Cloud Logging with role-based access controls and immutable storage Incident Response Cradle maintains a formal incident response plan with defined severity classifications, escalation paths, and post-incident review processes. Service status is available at our status page. Compliance Standard Status Xero API Security Assessment Certified annually since 2021 OWASP Top 10 Implemented GCP SOC 2 Type II Inherited from infrastructure provider GCP ISO 27001 Inherited from infrastructure provider Vertex AI ISO 42001 (AI Management) Inherited from AI provider NZ Privacy Act 2020 Compliant Australian Privacy Act 1988 Compliant These certifications are held by our infrastructure providers, Google Cloud and Twilio. Cradle does not itself hold SOC 2 or ISO 27001 certification. Our own security programme is documented against those frameworks and our policies are available on request; see Security and data due diligence. Sub-processors We maintain a transparent list of sub-processors that process data on behalf of our customers. Responsible Disclosure If you discover a security vulnerability in Cradle, please report it to security@cradle.io. We appreciate responsible disclosure and will action legitimate reports promptly. Please note that we do not operate a paid bug bounty programme. Questions For security-related questions or to request our detailed security documentation, contact us at security@cradle.io. Last reviewed: 18 November 2026