Document Control

Version	Date	Author	Description
0.1	January 2026	Security Team	Initial draft
0.9	February 2026	Security Team	Internal review and legal approval
1.0	February 2026	Security Team	Public release

1 Executive Summary

1.1 Purpose

This Security Whitepaper provides a comprehensive overview of the security practices, technical safeguards, and compliance measures implemented by Agentys Inc. ("Agentys") to protect customer data processed through the Agentys AI email management platform. It is intended for current and prospective customers, partners, auditors, and other stakeholders who require assurance regarding the security posture of our platform.

1.2 Scope

This document covers all aspects of the Agentys platform security, including but not limited to: infrastructure, application, data protection, AI model governance, identity and access management, monitoring, incident response, business continuity, and regulatory compliance. The controls described herein are aligned with industry-recognized frameworks including NIST Cybersecurity Framework (CSF) 2.0, ISO/IEC 27001:2022, ISO/IEC 42001:2023, and SOC 2 Trust Services Criteria.

1.3 Our Commitment

Agentys processes sensitive email data—including correspondence, attachments, and metadata—on behalf of professionals across legal, financial, healthcare, and corporate sectors. This responsibility demands the highest security standards. We are committed to maintaining an information security management system (ISMS) that meets or exceeds the requirements of:

• SOC 2 Type 2 — Security, Availability, and Confidentiality Trust Services Criteria
• ISO/IEC 27001:2022 — Information Security Management Systems
• ISO/IEC 42001:2023 — Artificial Intelligence Management Systems
• GDPR — General Data Protection Regulation (EU) 2016/679
• Loi 25 — Quebec's Act to modernize legislative provisions as regards the protection of personal information (Law 25)

2 Company Overview

2.1 About Agentys

Agentys is an artificial intelligence platform designed to transform email management for professionals and enterprises. By leveraging advanced natural language processing and machine learning, Agentys automates the drafting, prioritization, scheduling, and categorization of email communications—enabling knowledge workers to reclaim significant productive time while maintaining the quality and nuance of their professional correspondence.

2.2 Founding Principles

Agentys was founded with a clear mandate: build an AI platform that professionals can trust with their most sensitive communications. From day one, our engineering and product decisions have been guided by three core principles:

1. Security by Design: Security requirements are defined at the architecture phase, not retrofitted after deployment. Every component undergoes threat modeling before implementation.
2. Privacy by Default: Data minimization, purpose limitation, and user consent are embedded into every feature. We collect only the data required to deliver the service and retain it only as long as necessary.
3. Transparency: We believe customers have the right to understand how their data is processed, stored, and protected. This document is one expression of that commitment.

2.3 Corporate Information

Legal Name	Agentys Inc.
Headquarters	Montreal, Quebec, Canada
Industry	Enterprise Software / Artificial Intelligence
Product	AI Email Management Platform
Security Contact	security@agentys.io
Privacy Officer	privacy@agentys.io

3 Security Architecture

3.1 Architectural Overview

The Agentys platform is built on a modern, cloud-native architecture designed to deliver security, scalability, and reliability. Our architecture follows a defense-in-depth model, implementing overlapping security controls at every layer—from network perimeter to application logic to data storage. NIST CSF: PR.AC ISO 27001: A.8

3.2 Multi-Tenant Architecture with Tenant Isolation

Agentys employs a multi-tenant architecture in which each customer's data is logically isolated from all other tenants. Isolation is enforced at multiple levels:

• Database-Level Isolation: Each tenant's data is stored in logically separated database schemas with row-level security (RLS) policies enforced at the database engine level. Cross-tenant queries are architecturally impossible.
• Application-Level Isolation: Every API request is authenticated and scoped to the requesting tenant. Middleware validation ensures that no operation can access data outside its tenant boundary.
• Compute-Level Isolation: AI inference workloads are executed in isolated containers with dedicated resource allocations, preventing any data leakage between tenant processing jobs.

3.3 Data Encryption

All customer data is protected by encryption both at rest and in transit, using industry-standard cryptographic algorithms:

Layer	Standard	Implementation
Data at Rest	AES-256-GCM	All databases, object stores, and backups are encrypted using AES-256 with keys managed through a dedicated KMS. Encryption keys are rotated annually.
Data in Transit	TLS 1.3	All external and internal communications use TLS 1.3. Legacy protocols (TLS 1.0, 1.1, and SSL) are explicitly disabled. HSTS headers are enforced with a minimum max-age of one year.
Key Management	Envelope Encryption	Data encryption keys (DEKs) are wrapped with key encryption keys (KEKs) stored in a hardware security module (HSM)-backed key management service with FIPS 140-2 Level 3 validation.

NIST CSF: PR.DS-1, PR.DS-2 ISO 27001: A.8.24

3.4 Zero-Trust Network Architecture

Agentys implements a zero-trust network model based on the principle of "never trust, always verify." Key elements include:

• Identity-Based Access: All service-to-service communication requires mutual TLS (mTLS) authentication. No implicit trust is granted based on network location.
• Micro-Segmentation: Network policies enforce least-privilege communication paths between services. Each microservice can only communicate with explicitly authorized peers.
• Continuous Verification: Authentication tokens are short-lived and continuously validated. Session anomalies trigger re-authentication.

NIST CSF: PR.AC-5 SOC 2: CC6.1

3.5 Secure API Gateway

All external API traffic is routed through a secure API gateway that provides:

• Authentication & Authorization: OAuth 2.0 / JWT-based authentication with scope-based authorization for all endpoints.
• Rate Limiting: Configurable rate limits per tenant, per endpoint, and per IP address to prevent abuse and ensure fair resource allocation.
• Input Validation: Strict schema validation on all inbound requests. Malformed or oversized payloads are rejected at the gateway before reaching application logic.
• Logging & Telemetry: All API requests are logged with correlation IDs for end-to-end traceability. Sensitive data fields are masked in logs.

4 Data Protection

4.1 Email Data Handling

The Agentys platform processes email data to provide its core AI-powered management capabilities. We recognize the sensitivity of email communications and have implemented stringent protections around the handling of this data. ISO 27001: A.8.10

• Email Content Encryption: All email content (bodies, subjects, headers, and attachments) is encrypted at rest using AES-256-GCM. Decryption occurs only at the point of processing within isolated, ephemeral compute environments.
• Metadata Protection: Email metadata (timestamps, sender/recipient addresses, thread identifiers) is encrypted and stored separately from content, enabling limited-scope operations without exposing full message content.
• Attachment Handling: Attachments are stored in encrypted object storage, separate from email content databases, with independent access controls and audit logging.

4.2 Authentication with Email Providers

Agentys integrates with email providers (Google Workspace, Microsoft 365, and others) exclusively through OAuth 2.0 authorization flows:

• No Password Storage: Agentys never collects, transmits, or stores user email passwords. Authentication is delegated entirely to the email provider through OAuth 2.0.
• Minimal Scope: We request only the minimum necessary OAuth scopes required to deliver the service. Users can review and revoke access at any time through their email provider's security settings.
• Token Security: OAuth refresh tokens are encrypted at rest and stored in a dedicated, hardened credential vault with strict access controls and audit logging.

4.3 Data Residency

Agentys offers customers a choice of data residency regions to comply with local data sovereignty and regulatory requirements:

Region	Data Center Location	Applicable Regulations
Canada	Montreal, QC / Toronto, ON	PIPEDA, Loi 25
United States	US-East / US-West	SOC 2, CCPA
European Union	Frankfurt, DE / Paris, FR	GDPR, EU AI Act

Data residency selection is configured at the organization level during onboarding and is enforced at the infrastructure layer. Cross-region data transfers, if necessary for operational purposes, comply with applicable transfer mechanisms (e.g., Standard Contractual Clauses for EU data). GDPR: Art. 44-49

4.4 Data Lifecycle Management

Agentys implements automated data lifecycle management to ensure data is retained only for as long as necessary:

• Retention Policies: Configurable retention periods by data type, with defaults aligned to the minimum required for service delivery.
• Automated Purging: Expired data is automatically and irreversibly purged through cryptographic erasure (destruction of encryption keys) and physical overwrite.
• Account Termination: Upon account closure, all customer data is purged within 30 days, with a certificate of destruction available upon request.

4.5 Right to Deletion

In compliance with GDPR Article 17 (Right to Erasure), Loi 25, and other applicable privacy regulations, Agentys supports data subject deletion requests. Upon receiving a verified deletion request, Agentys will:

1. Acknowledge the request within 48 hours.
2. Complete deletion of all associated personal data within 30 calendar days.
3. Confirm deletion in writing to the requesting party.
4. Ensure deletion propagates to all backups within the subsequent backup rotation cycle (maximum 90 days).

GDPR: Art. 17 Loi 25: s. 28 SOC 2: P6.5

5 AI Security

5.1 AI Governance Framework

Agentys recognizes that the deployment of AI in the context of sensitive email communications introduces unique security and ethical considerations. Our AI governance framework is aligned with ISO/IEC 42001:2023 (AI Management Systems) and addresses risks specific to AI-powered email management. ISO 42001: 6.1

5.2 Data Isolation in AI Processing

AI models deployed by Agentys are designed with strict data isolation principles:

• No Data Retention Between Sessions: AI models do not retain user data between processing sessions. Each inference request operates on a fresh context, and all intermediate data is purged upon completion.
• No Customer Data Used for Training: Agentys does not use customer email content, metadata, or any other customer-provided data for training, fine-tuning, or improving AI models. Our models are trained exclusively on synthetic and licensed datasets.
• Isolated Inference Environments: Each tenant's AI inference workloads execute in isolated containerized environments with dedicated memory and compute resources. No state is shared between tenants.

5.3 Model Output Safety

Given that Agentys assists with composing professional email communications, the safety and accuracy of AI model outputs is of paramount importance:

• Output Review Pipeline: All AI-generated email drafts pass through a multi-stage review pipeline that includes content safety filters, tone analysis, and factual consistency checks before being presented to the user.
• Hallucination Mitigation: We employ retrieval-augmented generation (RAG) techniques and grounding mechanisms to minimize the risk of factually incorrect content in generated emails.
• Guardrails: AI outputs are constrained by configurable guardrails that prevent the generation of harmful, inappropriate, or non-compliant content.

5.4 Human-in-the-Loop Controls

Agentys enforces a mandatory human-in-the-loop paradigm for all email operations:

• User Approval Required: Every AI-generated email requires explicit user approval before it is sent. No email is transmitted without the user reviewing and confirming the content.
• Edit Capability: Users can edit any AI-generated content before approval, ensuring the final communication reflects their intent and judgment.
• Audit Trail: All user approval and editing actions are logged to provide a complete audit trail of email operations.

ISO 42001: A.7.3 NIST AI RMF: MG-3.2

5.5 Adversarial AI Protections

Agentys implements protections against adversarial attacks targeting AI systems:

• Prompt Injection Defense: Input sanitization and structured prompting techniques mitigate prompt injection and jailbreaking attacks.
• Data Poisoning Prevention: Strict separation between customer data and training pipelines eliminates the risk of data poisoning through customer inputs.
• Model Integrity: AI model artifacts are cryptographically signed and verified before deployment. Any tampering with model files is detected and blocked.

6 Access Control

6.1 Identity and Access Management

Agentys implements a comprehensive identity and access management (IAM) program to ensure that access to systems, data, and resources is granted on a least-privilege basis and continuously monitored. NIST CSF: PR.AC ISO 27001: A.5.15-A.5.18

6.2 Role-Based Access Control (RBAC)

All access within the Agentys platform and corporate systems is governed by role-based access control:

• Predefined Roles: Access roles are predefined with specific permission sets aligned to job functions. Custom roles can be configured for enterprise customers.
• Least Privilege: Users are assigned the minimum permissions necessary to perform their functions. Elevated access requires explicit approval.
• Separation of Duties: Critical operations (e.g., production deployments, key management, access provisioning) require involvement of multiple authorized individuals.

6.3 Multi-Factor Authentication (MFA)

Multi-factor authentication is mandatory for all access:

• Employee Access: All Agentys employees are required to use MFA (TOTP or hardware security keys) for access to all corporate and production systems. SMS-based MFA is not permitted due to known vulnerabilities.
• Customer Access: Agentys supports and encourages MFA for all customer accounts. Enterprise plans include enforced MFA policies configurable by the customer's administrator.

6.4 Privileged Access Management (PAM)

Access to production environments and sensitive systems is governed by a strict privileged access management program:

• Just-in-Time (JIT) Access: Privileged access is not persistent. Engineers request temporary, time-limited access to production systems through an automated approval workflow. Access is automatically revoked upon expiration.
• Session Recording: All privileged access sessions are recorded and available for audit review.
• Break-Glass Procedures: Emergency access procedures are documented and tested, with all break-glass events triggering immediate security team notification and post-event review.

SOC 2: CC6.2, CC6.3

6.5 Access Reviews

Access permissions are subject to regular review to ensure ongoing appropriateness:

Review Type	Frequency	Scope
User Access Review	Quarterly	All employee access across corporate and production systems
Privileged Access Review	Monthly	All accounts with administrative or elevated privileges
Service Account Review	Quarterly	All automated service accounts and API keys
Offboarding Verification	Within 24 hours	Terminated employee access revocation confirmation

6.6 Single Sign-On (SSO) Integration

Agentys supports enterprise SSO integration through industry-standard protocols:

• SAML 2.0: Full support for SAML 2.0-based identity providers (e.g., Okta, Azure AD, OneLogin).
• OpenID Connect (OIDC): OIDC support for modern identity providers and custom integrations.
• SCIM Provisioning: Automated user provisioning and deprovisioning through SCIM 2.0, ensuring user lifecycle events in the identity provider are reflected in Agentys in real time.

7 Infrastructure Security

7.1 Cloud Infrastructure

The Agentys platform is hosted on SOC 2 Type 2 and ISO 27001-certified cloud infrastructure providers. Our infrastructure-as-code approach ensures consistent, auditable, and repeatable deployments across all environments. NIST CSF: PR.IP

• Infrastructure as Code (IaC): All infrastructure is defined, versioned, and deployed through code. Manual configuration of production infrastructure is prohibited.
• Immutable Infrastructure: Production servers are treated as immutable. Updates are deployed as new instances; in-place modifications are not permitted.
• Environment Separation: Development, staging, and production environments are fully isolated with separate accounts, networks, and credentials.

7.2 Network Segmentation

Agentys implements multi-layered network segmentation to contain potential security incidents and limit lateral movement:

• Virtual Network Isolation: Production, staging, and corporate networks operate in separate virtual networks with no direct routing between them.
• Micro-Segmentation: Within the production environment, network policies enforce service-level segmentation. Each microservice is confined to its authorized communication paths.
• Private Endpoints: Databases, key management services, and internal APIs are accessible only through private network endpoints. No direct internet exposure.

7.3 Web Application Firewall (WAF)

All public-facing endpoints are protected by a Web Application Firewall (WAF) that provides:

• Protection against OWASP Top 10 vulnerabilities (SQL injection, XSS, CSRF, etc.)
• Custom rule sets tailored to the Agentys application layer
• Geolocation-based filtering capabilities
• Bot detection and automated threat mitigation
• Real-time alerting on blocked and flagged requests

7.4 DDoS Protection

Agentys employs multi-layer DDoS protection to ensure service availability:

• Network Layer (L3/L4): Always-on volumetric DDoS mitigation with the capacity to absorb multi-terabit attacks.
• Application Layer (L7): Behavioral analysis and rate limiting to detect and mitigate application-layer DDoS attacks.
• Auto-Scaling: Infrastructure auto-scales in response to traffic spikes, providing additional resilience against resource exhaustion attacks.

7.5 Container Security

All Agentys services run in containerized environments with the following security controls:

• Image Scanning: All container images are scanned for vulnerabilities in the CI/CD pipeline. Images with critical or high-severity CVEs are blocked from deployment.
• Minimal Base Images: Containers are built on minimal, hardened base images with unnecessary packages and tools removed.
• Runtime Security: Container runtime behavior is monitored for anomalies (e.g., unexpected process execution, file system modifications, network connections). Deviations from the expected profile trigger alerts.
• Non-Root Execution: All containers run as non-root users with read-only file systems where feasible.

NIST CSF: PR.IP-1 SOC 2: CC7.1

8 Monitoring & Incident Response

8.1 Security Monitoring

Agentys operates a 24/7 security monitoring program that provides continuous visibility into the security posture of the platform: NIST CSF: DE.CM ISO 27001: A.8.15, A.8.16

• Security Information and Event Management (SIEM): Centralized log aggregation and correlation across all infrastructure, application, and security layers. Security events are analyzed in real time for indicators of compromise (IOCs) and anomalous behavior.
• Automated Threat Detection: Machine learning-based anomaly detection identifies unusual patterns in authentication, API usage, data access, and network traffic.
• Alerting: Security alerts are triaged and routed through an automated escalation framework with defined response times based on severity.

8.2 Log Management

Comprehensive logging supports both security monitoring and compliance requirements:

Log Type	Retention	Encryption
Security events (authentication, authorization)	1 year	AES-256 at rest
Application logs	90 days	AES-256 at rest
Infrastructure / system logs	90 days	AES-256 at rest
Audit trail (admin actions)	2 years	AES-256, immutable storage
API access logs	1 year	AES-256 at rest

All logs are stored in tamper-evident, append-only storage. Sensitive fields (e.g., email content, PII) are redacted or masked prior to logging.

8.3 Incident Response Plan

Agentys maintains a formal Incident Response Plan (IRP) that defines procedures for detecting, containing, eradicating, and recovering from security incidents. The IRP is aligned with the NIST Computer Security Incident Handling Guide (SP 800-61r2). NIST CSF: RS.RP

8.3.1 Incident Severity Classification

Severity	Definition	Response Time	Escalation
Critical (P1)	Active data breach, service compromise, or exploitation of a critical vulnerability	15 minutes	CISO, CEO, Legal
High (P2)	Confirmed security event with potential for data exposure or service disruption	1 hour	CISO, Engineering Lead
Medium (P3)	Suspicious activity under investigation; no confirmed impact	4 hours	Security Team Lead
Low (P4)	Minor security events, policy violations, or informational alerts	24 hours	Security Analyst

8.3.2 Incident Response Phases

1. Detection & Analysis: Initial triage, scope assessment, and evidence preservation.
2. Containment: Immediate actions to limit incident impact (e.g., network isolation, credential rotation, service suspension).
3. Eradication: Removal of the root cause, including malware, compromised accounts, or vulnerable code.
4. Recovery: Restoration of affected systems and services from verified clean backups, with monitoring for recurrence.
5. Post-Incident Review: Comprehensive analysis of incident timeline, root cause, response effectiveness, and lessons learned. Findings are documented and drive improvements to security controls.

8.4 Breach Notification

In the event of a confirmed data breach involving personal data, Agentys will:

• Notify the relevant supervisory authority within 72 hours of becoming aware of the breach, as required by GDPR Article 33.
• Notify affected data subjects without undue delay if the breach is likely to result in a high risk to their rights and freedoms (GDPR Article 34).
• Notify the Commission d'acces a l'information du Quebec (CAI) for incidents involving Quebec residents, as required by Loi 25.
• Provide affected customers with a detailed incident report including nature of the breach, data affected, remediation steps, and preventive measures implemented.

GDPR: Art. 33, 34 Loi 25: s. 3.5

9 Business Continuity & Disaster Recovery

9.1 Business Continuity Program

Agentys maintains a formal Business Continuity Plan (BCP) and Disaster Recovery Plan (DRP) to ensure resilience against disruptive events, including natural disasters, infrastructure failures, cyberattacks, and pandemic scenarios. Our BCP/DRP is reviewed and updated annually. NIST CSF: PR.IP-9 ISO 27001: A.5.29, A.5.30

9.2 Recovery Objectives

Metric	Target	Description
Recovery Point Objective (RPO)	1 hour	Maximum acceptable data loss. Continuous replication ensures data loss does not exceed 1 hour in the event of a disaster.
Recovery Time Objective (RTO)	4 hours	Maximum acceptable downtime. Full service restoration to a secondary region within 4 hours of a declared disaster.
Uptime SLA	99.9%	Monthly uptime commitment, equivalent to a maximum of approximately 43 minutes of unplanned downtime per month.

9.3 Multi-Region Redundancy

The Agentys platform is architected for multi-region availability:

• Active-Passive Configuration: Primary workloads run in the customer's selected region with a warm standby in a geographically separated secondary region.
• Cross-Region Replication: Databases and object stores are continuously replicated to the secondary region, ensuring data availability in the event of a regional outage.
• DNS Failover: Automated DNS failover directs traffic to the secondary region within minutes of a primary region failure detection.

9.4 Backup Strategy

• Automated Backups: Full database backups are performed daily, with incremental backups every hour. All backups are encrypted using AES-256.
• Backup Verification: Backup integrity is verified through automated restoration testing on a weekly basis.
• Offsite Storage: Backups are stored in a geographically separated region from the primary infrastructure, with independent access controls.
• Retention: Daily backups are retained for 30 days, weekly backups for 90 days, and monthly backups for 1 year.

9.5 Disaster Recovery Testing

Agentys conducts formal disaster recovery testing to validate the effectiveness of recovery procedures:

• Annual Full DR Test: A complete failover to the secondary region is performed annually, simulating a total loss of the primary region.
• Quarterly Tabletop Exercises: Scenario-based exercises involving engineering, security, and leadership teams to rehearse response procedures.
• Backup Restoration Tests: Monthly tests to verify that backups can be successfully restored and that data integrity is maintained.

SOC 2: A1.2, A1.3

10 Compliance & Certifications

10.1 Compliance Program Overview

Agentys maintains a comprehensive compliance program managed by a dedicated Governance, Risk, and Compliance (GRC) team. Our compliance posture is validated through independent third-party audits and continuous monitoring against applicable regulatory and industry frameworks.

10.2 Certifications & Attestations

Framework	Scope	Status
SOC 2 Type 2	Security, Availability, Confidentiality Trust Services Criteria	Certified (annual audit)
ISO/IEC 27001:2022	Information Security Management System (ISMS)	Certified (3-year cycle with annual surveillance)
ISO/IEC 42001:2023	AI Management System (AIMS)	Certified (3-year cycle with annual surveillance)
GDPR	Processing of personal data of EU data subjects	Compliant (DPA available upon request)
Loi 25 (Quebec Law 25)	Protection of personal information of Quebec residents	Compliant (Privacy Impact Assessments conducted)

10.3 Regulatory Compliance Details

10.3.1 GDPR Compliance

As a data processor handling personal data of EU data subjects, Agentys complies with all applicable requirements of the General Data Protection Regulation (EU) 2016/679:

• Data Processing Agreements (DPAs) executed with all customers and sub-processors.
• Data Protection Impact Assessments (DPIAs) conducted for high-risk processing activities.
• Records of processing activities maintained as required by Article 30.
• Data subject rights (access, rectification, erasure, portability, restriction, objection) supported through documented procedures.
• Standard Contractual Clauses (SCCs) implemented for cross-border data transfers where applicable.

10.3.2 Loi 25 Compliance (Quebec Privacy Law)

As a company headquartered in Quebec, Agentys fully complies with Loi 25 (Act to modernize legislative provisions as regards the protection of personal information):

• Designated Privacy Officer with reporting to executive leadership.
• Privacy Impact Assessments (PIAs) conducted for information systems involving personal information.
• Transparency requirements met through published privacy policy and incident notification procedures.
• Consent management aligned with Loi 25 requirements for express consent.

10.4 Third-Party Security Assessments

In addition to certification audits, Agentys engages independent third-party security firms to conduct:

• Annual Penetration Testing: Comprehensive penetration testing of the platform, including network, application, and API layers, conducted by CREST-accredited firms.
• Continuous Vulnerability Assessment: Ongoing automated vulnerability scanning of all internet-facing assets.
• Red Team Exercises: Periodic adversary simulation exercises to test detection and response capabilities.

11 Vendor Management

11.1 Third-Party Risk Management Program

Agentys recognizes that the security of our platform extends to the security of our vendors and sub-processors. We maintain a formal Third-Party Risk Management (TPRM) program to assess, monitor, and manage risks associated with external service providers. NIST CSF: ID.SC ISO 27001: A.5.19-A.5.23

11.2 Vendor Assessment Process

All vendors that process, store, or have access to customer data undergo a structured security assessment before onboarding:

1. Initial Questionnaire: Vendors complete a comprehensive security questionnaire based on the Shared Assessments SIG (Standardized Information Gathering) framework.
2. Risk Scoring: Vendors are assigned a risk tier (Critical, High, Medium, Low) based on the nature and volume of data they access and the criticality of the service they provide.
3. Evidence Review: For Critical and High-tier vendors, Agentys reviews SOC 2 reports, ISO 27001 certificates, penetration test summaries, and other relevant audit evidence.
4. Contractual Safeguards: All vendor agreements include data protection clauses, security requirements, breach notification obligations, and right-to-audit provisions.

11.3 Ongoing Vendor Monitoring

Vendor risk is not a point-in-time assessment. Agentys conducts ongoing monitoring of critical vendors:

• Annual Reassessment: All Critical and High-tier vendors undergo annual security reassessment.
• Continuous Monitoring: Automated monitoring of vendor security posture through external risk rating services.
• Incident Notification: Vendors are contractually required to notify Agentys within 24 hours of any security incident that may affect Agentys data or services.

11.4 Sub-Processor Management

In accordance with GDPR Article 28 requirements:

• A current list of sub-processors is maintained and published at agentys.io/legal/sub-processors.
• Customers are notified of any changes to sub-processors with a minimum of 30 days advance notice.
• Customers have the right to object to new sub-processors, with a resolution process defined in the Data Processing Agreement.

12 Employee Security

12.1 Personnel Security Overview

Agentys recognizes that people are both the greatest asset and a critical vector in information security. Our personnel security program ensures that all employees, contractors, and temporary workers understand their security responsibilities and are equipped to uphold them. ISO 27001: A.6.1-A.6.8

12.2 Background Checks

All prospective Agentys employees and contractors undergo pre-employment background checks, the scope of which is commensurate with the sensitivity of the role:

• Criminal record verification
• Employment history verification
• Education and credential verification
• Reference checks
• For roles with access to financial data or customer PII: credit history check (where permitted by law)

12.3 Security Awareness Training

All Agentys personnel participate in a comprehensive security awareness program:

Training Type	Frequency	Audience
Security Onboarding	Upon hire	All employees and contractors
General Security Awareness	Quarterly	All employees and contractors
Phishing Simulations	Monthly	All employees
Secure Development Training	Quarterly	Engineering team
Incident Response Drills	Quarterly	Security and engineering teams
Privacy & Data Protection	Annually	All employees and contractors

12.4 Security Policies

Agentys maintains a comprehensive suite of information security policies, including but not limited to:

• Information Security Policy: Overarching policy defining the organization's commitment to information security and the ISMS framework.
• Acceptable Use Policy: Defines acceptable and prohibited use of company systems, devices, and data.
• Clean Desk Policy: Requires that sensitive information is not left unattended on desks or screens. Workstations must be locked when unattended.
• Data Classification Policy: Defines data classification levels (Public, Internal, Confidential, Restricted) and handling requirements for each.
• Remote Work Security Policy: Security requirements for employees working remotely, including VPN usage, device security, and physical workspace requirements.

12.5 Secure Development Lifecycle (SDLC)

Agentys follows a Secure Software Development Lifecycle that integrates security at every phase of the development process:

1. Design: Threat modeling and security architecture review for all new features and significant changes.
2. Development: Secure coding standards, mandatory code reviews, and pre-commit security linting.
3. Testing: Automated security testing in CI/CD (SAST, DAST, SCA). Manual security review for high-risk changes.
4. Deployment: Automated deployment pipelines with security gate checks. No direct production access for developers.
5. Operations: Runtime application security monitoring, vulnerability management, and continuous feedback into the design phase.

NIST CSF: PR.IP-2 SOC 2: CC8.1

13 Vulnerability Management

13.1 Vulnerability Management Program

Agentys operates a proactive vulnerability management program designed to identify, assess, prioritize, and remediate security vulnerabilities across all systems and applications. NIST CSF: DE.CM-8 ISO 27001: A.8.8

13.2 Vulnerability Scanning

Regular automated scanning is conducted across all layers of the technology stack:

• Infrastructure Scanning: Weekly authenticated vulnerability scans of all servers, network devices, and cloud resources.
• Application Scanning: Static Application Security Testing (SAST) on every code commit. Dynamic Application Security Testing (DAST) on every staging deployment.
• Dependency Scanning: Software Composition Analysis (SCA) to identify vulnerabilities in third-party libraries and open-source dependencies. Scans run on every build.
• Container Scanning: Container image vulnerability scanning in the CI/CD pipeline and in the container registry.

13.3 Patch Management

Agentys maintains defined Service Level Agreements for vulnerability remediation based on severity:

Severity (CVSS)	Remediation SLA	Process
Critical (9.0 - 10.0)	24 hours	Emergency patch. Security team lead approval. Immediate deployment to production.
High (7.0 - 8.9)	7 days	Expedited patching through the standard release pipeline with priority scheduling.
Medium (4.0 - 6.9)	30 days	Standard patching cycle. Included in the next scheduled release.
Low (0.1 - 3.9)	90 days	Addressed in routine maintenance. Risk-accepted if mitigating controls exist.

13.4 Penetration Testing

Agentys engages independent, CREST-accredited third-party security firms to conduct annual comprehensive penetration testing:

• Scope: External network, web application, API, mobile application, and cloud infrastructure.
• Methodology: Testing follows OWASP Testing Guide, PTES (Penetration Testing Execution Standard), and NIST SP 800-115 methodologies.
• Reporting: Findings are classified by severity and assigned to engineering teams for remediation within the patch management SLAs defined above.
• Re-Testing: Critical and high-severity findings are re-tested after remediation to verify effective resolution.

13.5 Responsible Disclosure & Bug Bounty

Agentys is committed to working with the security research community to identify and address vulnerabilities:

• Responsible Disclosure Policy: Published at agentys.io/security/responsible-disclosure, our policy provides guidelines for security researchers to report vulnerabilities safely and responsibly.
• Bug Bounty Program: Agentys operates a bug bounty program that provides financial rewards to researchers who discover and report qualifying vulnerabilities. Rewards are commensurate with the severity and impact of the finding.
• Safe Harbor: Security researchers who comply with our responsible disclosure policy are protected under our safe harbor commitment. We will not pursue legal action against researchers acting in good faith.
• Reporting Channel: Vulnerability reports can be submitted to security@agentys.io. We acknowledge receipt within 24 hours and provide an initial assessment within 5 business days.

ISO 27001: A.8.8 SOC 2: CC7.1

14 Contact Information

14.1 Security & Privacy Contacts

For questions, concerns, or requests related to the security and privacy practices described in this document, please contact the appropriate team:

Purpose	Contact	Response Time
General security inquiries	security@agentys.io	2 business days
Privacy inquiries & data subject requests	privacy@agentys.io	2 business days
Vulnerability reporting	security@agentys.io	24 hours (acknowledgment)
SOC 2 report requests	security@agentys.io	5 business days
Data Processing Agreement (DPA) requests	privacy@agentys.io	5 business days
Incident reporting	security@agentys.io	Immediate triage

14.2 Corporate Address

Agentys Inc.
Montreal, Quebec, Canada
Web: agentys.io

14.3 Additional Resources

• Privacy Policy: agentys.io/privacy
• Terms of Service: agentys.io/terms
• Sub-Processor List: agentys.io/legal/sub-processors
• Responsible Disclosure: agentys.io/security/responsible-disclosure
• Status Page: status.agentys.io