Trust Center

Security and Trust Center.

An overview of how ProfytAI protects customer compliance artifacts, regulatory obligations, and audit evidence. This page summarizes the security controls, processes, and assurance practices we maintain across our platform. Detailed materials are available to qualified reviewers under NDA.

Last reviewed: May 2026

9 / 9
Common Criteria (CC1-CC9)

Documented controls across CC1-CC9.SOC 2 Type I scope: Security (one of five AICPA categories).

A+
TLS Grade

Qualys SSL Labs (May 2026).No legacy protocols offered.

Verify at SSL Labs
0 / 0
Critical / High Findings

Internal DAST engagement.

Defense-in-Depth Architecture

Defense in Depth, by Design.

ProfytAI's production environment uses cloud-native infrastructure with network segmentation, environment segregation, encrypted storage, managed key lifecycle, and recoverable backup. Each layer limits blast radius under failure or compromise.

Resilient Cloud Architecture

Production workloads run on hardened cloud-native infrastructure with multi-zone deployment, encrypted storage, and tightly scoped administrative access.

Network Segmentation

Public ingress is isolated from protected application and data tiers. Sensitive endpoints sit behind authentication, authorization, and continuous traffic inspection.

Environment Segregation

Production and non-production environments are logically separated to contain cross-environment risk and support controlled evidence handling.

Encrypted Backup and Recovery

Backups are encrypted with managed key controls, retained against documented policy, replicated cross-region, with documented recovery procedures aligned to business continuity requirements.

  • Hosting and Data Residency

    Multi-zone resilient production deployment with cross-region disaster recovery. Specific region and provider details are documented under NDA during vendor due diligence. Customer data does not leave the contracted jurisdiction without explicit authorization.

  • Resilience

    Multi-zone redundant deployment with encrypted backup coverage and documented recovery procedures aligned to business continuity requirements.

  • Environment Segregation

    Logically separated production and non-production environments with controlled administrative access and reviewed change paths.

  • Network Controls

    Restricted public ingress, segmented internal services, network flow logging, and centralized log collection.

  • Backup Protection

    AES-256 encrypted backups with documented retention, point-in-time recovery, and documented restoration procedures.

Encryption and Key Management

Encryption Everywhere: in Transit, at Rest, and in Backup.

ProfytAI applies industry-standard cryptographic primitives across data at rest, data in transit, key lifecycle management, and transport-layer hardening.

Encryption at Rest

AES-256

Customer data, backups, and platform storage are encrypted at rest using AES-256 with managed key lifecycle controls.

Encryption in Transit

TLS 1.2 / 1.3 · HSTS

Client and service communications use TLS 1.2 or higher, with TLS 1.3 supported. HTTP requests are 301-redirected to HTTPS at the edge, and HSTS is preload-eligible and submitted to the Chrome preload list.

No Legacy Protocols

A+ TLS Grade

SSLv2, SSLv3, TLS 1.0, and TLS 1.1 are not offered. The public web tier holds an A+ grade from Qualys SSL Labs across all CloudFront edge nodes (assessed May 2026).

Compliance and Assurance

Internally Assessed. Continuously Evidenced.

ProfytAI's control environment is organized against the AICPA Trust Services Criteria for Security. Twenty-two documented controls are mapped to the Common Criteria CC1 through CC9 and organized for future cross-walk to ISO/IEC 27001:2022 Annex A.

AICPA Common Criteria Coverage

Documented controls are organized against the AICPA Trust Services Criteria for Security (TSP Section 100, Common Criteria CC1 through CC9), covering governance, access, change management, system operations, monitoring, and risk mitigation.

  • Twenty-two documented controls mapped to Common Criteria CC1 through CC9.
  • Evidence collection organized for the Security trust services category.
  • Restricted readiness materials available to qualified reviewers under NDA.

Data Handling Safeguards

Customer data safeguards are documented around ownership, confidentiality, least-privilege use, encrypted handling, and controlled evidence release.

  • Customer data handled under documented confidentiality obligations.
  • Production data access limited to authorized personnel and approved purposes.
  • Restricted data-handling evidence released only under NDA-controlled workflows.

Operational Security

Monitoring, incident handling, continuity planning, and vulnerability workflows support secure operation of the customer-facing platform.

  • Public web surface assessed through internal automated DAST using open-source security testing tools.
  • Security findings handled internally and not published on public trust pages.
  • Operational evidence provided to qualified reviewers under controlled channels.

Identity, Access and Secure SDLC

Least-Privilege Access. Secure-by-Design Delivery.

Workforce access follows the principle of least privilege under enforced multi-factor authentication, group-based authorization, and centralized audit logging. Product delivery runs through peer review, automated pre-merge quality gates, and continuous secrets scanning.

Identity and Access Management, Under the Principle of Least Privilege.

  • Multi-factor authentication is enforced for workforce access to production, source-control, and administrative systems.
  • Centralized identity and group-based authorization govern workforce access where supported.
  • Production permissions are granted through groups, not direct user assignments, and production data access is centrally logged.
  • Production access changes require documented approval by accountable leadership.
  • Privileged administrative access is reserved for break-glass scenarios with strong authentication and approval.

Peer-Reviewed Change Management

Every change to the production branch requires peer review and approval before merge, with branch protection enforced at the source-control layer.

Pre-Merge Quality Gates

Build, static type checking, lint, and the regression test suite run automatically before any change can reach production. Failed gates block the merge.

Auditable Release Pipeline

Branch protection on production, release, and hotfix branches blocks force pushes and requires pull-request approval. Conventional-commit format and release notes preserve a structured change history for change traceability and incident reconstruction during audit.

Secrets Hygiene and Controlled Deployment

Secrets are scanned pre-commit and continuously in source. Production deploys are operated by named, authorized engineering personnel only.

AI Governance

How We Govern AI Use.

ProfytAI separates regulatory extraction from interpretation. The extraction stage runs without any LLM involvement so its output is byte-identical on re-execution. The interpretation stage only sees public regulatory text and is always anchored to the source paragraph. Material compliance decisions stay with the customer's compliance team.

Deterministic Regulatory Extraction

Stage 1No LLM Involvement

Regulatory obligations are extracted verbatim from source documents using deterministic code. Exact wording is preserved with machine-readable document-path identifiers, and re-execution produces byte-identical output that regulators can verify independently.

LLM-Assisted Interpretation, Source-Anchored

Stage 2Source-Anchored Output

Interpretation and structuring run against already-extracted public regulatory text only. Every downstream artifact carries a permanent source anchor back to the verbatim source paragraph, and model output is flagged for human review.

Human-in-the-Loop

Material compliance decisions are reviewed and approved by the customer's compliance team. ProfytAI does not auto-publish, auto-attest, or auto-submit to any regulator.

Customer-Data Isolation

Customer restricted data is never sent to the LLM. Only already-extracted public regulatory text and structured prompts reach the model. This is an architectural property, not a policy.

Training Opt-Out and Pinned Models

The LLM provider is contractually prohibited from training on data sent via the API. Model versions are pinned to specific releases and upgrades follow documented regression testing.

Continuous Monitoring and Threat Detection

Detect. Respond. Recover.

ProfytAI operates continuous security monitoring across the platform. Threat telemetry, business continuity awareness, and vulnerability management procedures support secure operation of the customer-facing platform.

Threat Detection

Cloud-native anomaly-detection signals trigger alerts that feed into our incident response process.

Business Continuity and Disaster Recovery

Operational continuity planning, multi-zone resilience, and cross-region backup support service resilience and recovery coordination.

Vulnerability Management

Security findings are triaged, prioritized, and tracked to remediation through a documented risk-based workflow.

Assurance Evidence and Disclosure

Security Assessment Evidence,Released Under NDA.

Security overview, SOC 2 readiness materials, internal assessment summaries, and data-protection documents are released to qualified reviewers under appropriate confidentiality controls. Security concerns are handled through coordinated vulnerability disclosure per RFC 9116.

Assurance Evidence

Released to Qualified Reviewers Under NDA.

Security Overview

Summary of governance, encryption, access management, monitoring, and operational resilience controls.

SOC 2 Readiness Materials

Readiness evidence mapped to the AICPA Trust Services Criteria for Security, including documented control coverage.

Security Assessment Summary

Internal automated DAST findings from the public web surface, provided to qualified reviewers under NDA without exposing non-public test material.

Data Protection Materials

Customer-data handling, encryption, retention, ownership, confidentiality, and processing commitments.

Coordinated Vulnerability Disclosure

Concerns Reach Security Through Approved Channels.

Coordinated through RFC 9116 and the ProfytAI security channel. Out-of-scope activity includes denial-of-service testing, phishing, credential attacks, and any attempt to access customer or third-party data.

Coordinated Through Approved Channels

Customers and partners route security concerns through their established ProfytAI support or security-review channel. Automated scanners follow the published security contact discovery file.

Good-Faith Reporting

Reports include enough detail to reproduce and triage the concern without exposing confidential data in public forums or non-secure channels.

Safe Harbor

ProfytAI will not pursue legal action against security researchers acting in good faith under this policy. Stay within scope, avoid privacy violations and service disruption, and give us a reasonable window to remediate before public disclosure.

Response Commitments

We acknowledge eligible reports within 5 business days and provide status updates at least every 14 days until the issue is resolved.