GetAssurAI
Version 1.0  |  August 2026
getassurai.com
AssurAI Inc. (Delaware C-Corp)
Security Architecture Overview
Enterprise data protection, AI handling, and data residency β€” for regulated industry evaluation
πŸ”’ Confidential β€” For Recipient Only

A β€” Platform Overview

AssurAI is a SaaS GRC platform for internal audit, SOX, and compliance teams. It accelerates evidence testing, workpaper preparation, and audit reporting using AI. The platform is a static HTML front-end backed by serverless Netlify functions, a Supabase PostgreSQL database, and the Anthropic Claude API for AI inference.

Front-end hosting
Netlify CDN β€” SOC 2 Type II
Database
Supabase PostgreSQL (AWS us-east-1)
AI provider
Anthropic Claude API β€” SOC 2 Type II
Architecture
Static HTML + Serverless + PostgreSQL
Data isolation
Row-Level Security (RLS) β€” per org_id
Entity
AssurAI Inc. β€” Delaware C-Corp

B β€” Data Flow Diagram

Client Browser
User's device
β†’
Netlify CDN
Static assets & serverless functions Β· No audit data stored
SOC 2 Type II
β†’
Supabase PostgreSQL
AWS us-east-1 Β· AES-256 Β· RLS isolation Β· PITR backup
SOC 2 Type II Β· ISO 27001
β†’
Anthropic Claude API
AI inference only Β· Zero data retention Β· No training
SOC 2 Type II Β· ZDR
All connections encrypted Β· HTTPS / TLS 1.3 Β· No plaintext at any layer

C β€” Security Controls Summary

Control Implementation Standard / Note
Encryption at rest AES-256 (Supabase / AWS storage layer) Industry standard Β· Always on
Encryption in transit TLS 1.3 enforced on all connections Β· HSTS TLS 1.0/1.1/1.2 rejected
Authentication JWT + Google SSO / Azure Entra ID SSO OAuth 2.0 / OIDC Β· 1-hr token expiry
Authorisation PostgreSQL Row Level Security (RLS) per org_id Database-level Β· Cannot be bypassed
Data retention (AI) Zero data retention β€” Anthropic enterprise API (ZDR) Contractually guaranteed by Anthropic
AI model training Customer data excluded from all model training Anthropic enterprise API policy
Backups Daily encrypted backups Β· 30-day retention Β· PITR AWS standard Β· RTO 4 hours
Evidence integrity SHA-256 hash on every evidence file at ingestion Tamper-evident Β· Re-verified on access
Monitoring Supabase audit logs Β· Sentry real-time error tracking Continuous Β· Anomaly alerting
Network / DDoS Cloudflare WAF + DDoS protection Β· Netlify CDN CDN-level Β· SOC 2 Type II
API key management Encrypted environment variables β€” never in client code Zero secret exposure in front-end
Access control RBAC β€” Admin / Manager / Auditor / Viewer roles Granular per-module permissions
Staff data access By explicit written consent only Β· Logged Β· Time-limited Cannot access client data without consent
Incident notification Within 72 hours of confirmed breach Meets GDPR Article 33

D β€” AI Data Handling Policy

Zero Data Retention: Client data submitted to the Anthropic Claude API is processed in real-time and immediately discarded upon response. Anthropic's Zero Data Retention (ZDR) policy for enterprise API customers means no client data is stored, cached, or used for model training. This is contractually guaranteed by Anthropic's enterprise API terms of service.


No model training: AssurAI audit evidence, workpapers, findings, and financial data are never used to train, fine-tune, or improve any AI model β€” by AssurAI or by Anthropic. AI inference is stateless β€” each request is isolated and independent.


What AssurAI stores: AssurAI stores only the data you explicitly save to the platform β€” workpapers, findings, controls, and engagement files in your dedicated Supabase database. Documents uploaded for AI analysis are processed in memory; only the AI-generated output is stored if you choose to save it.

E β€” AI Provider & Data Residency

AssurAI supports four AI providers. Your data goes to whichever provider you choose β€” under your agreement with them.

🌐
Option 1 β€” AssurAI Default
Provider: Anthropic Claude API
Audit data β†’ Anthropic (US servers) β†’ Zero data retention β†’ Response returned β†’ Data discarded
Suitable for: Most enterprise clients
Key policy: Zero Data Retention (ZDR) contractually guaranteed by Anthropic. Data never stored, never used for training.
SOC 2 Type II ZDR
πŸ”΅
Option 3 β€” Azure OpenAI BYOLLM
Provider: Client's own Azure OpenAI endpoint
Audit data β†’ Client's Azure instance in chosen region β†’ Response returned
Suitable for: Microsoft shops, regulated entities requiring in-region processing (KSA, UAE, EU, UK)
Key benefit: Full data sovereignty. Data never leaves client's Azure environment. Ideal for NCA-regulated Saudi organisations (Azure KSA North).
Regions: Azure KSA North, UAE North, EU West, UK South, US East and others.
βœ… Available KSA compliant
⚫
Option 4 β€” OpenAI GPT-4o BYOLLM
Provider: Client's own OpenAI API key
Audit data β†’ OpenAI API β†’ Response returned
Suitable for: Organisations with existing OpenAI enterprise agreements
Key benefit: Use your existing OpenAI contract and data processing agreement.
βœ… Available
Provider Data Location New Vendor Needed? Status
Anthropic Claude (default) US (Anthropic) No (AssurAI agreement) βœ… Default
Google Gemini (BYOLLM) Your Google Cloud No βœ… Live
Azure OpenAI (BYOLLM) Your Azure region No βœ… Live
OpenAI GPT-4o (BYOLLM) OpenAI servers No βœ… Live

When using BYOLLM, AssurAI acts as the audit workflow layer only. All AI calls route through your provider under your existing data agreement. AssurAI never stores the raw prompt data sent to your provider.

F β€” Infrastructure Certifications

Provider Role Certifications
Supabase (via AWS) Database, authentication, storage SOC 2 Type II Β· ISO 27001 Β· HIPAA eligible Β· GDPR
Netlify CDN, serverless functions (no data stored) SOC 2 Type II Β· GDPR
Anthropic AI / LLM inference β€” zero retention SOC 2 Type II Β· ZDR policy Β· No training on customer data
AssurAI Inc. Application layer SOC 2 Type II certification in progress (2027)

Sub-processor SOC 2 reports available under NDA. Contact security@getassurai.com to request.

G β€” Contact & Enterprise Support

Security queries
security@getassurai.com
Technical review
shakeel@getassurai.com
NDA / Legal
legal@getassurai.com

AssurAI supports: security questionnaires (VSQ / CAIQ / SIG Lite), custom DPAs, mutual NDAs, architecture review calls, penetration test reports (under NDA), and BYOLLM configuration for KSA / Azure deployments.