Aegis InsightOps Secure / Trust & Security

Your data never leaves your tenancy. Our architecture is built around that.

Q: What happens to our data if we terminate the service?

Your data stays in your AWS tenancy. Terminating Secure removes IVI's control-plane access within 30 days and deletes IVI-side metadata. Your telemetry in your tenancy is yours to retain or delete per your own policies.

The Secure platform is designed from the data plane outward. Your telemetry stays in your AWS environment, private AI inference runs against your data only, and every action is logged, scoped, and auditable.

Data plane

Your AWS tenancy

AI inference

Private, no training

Identity

Your IdP, your RBAC

Attestation

SOC 2 in progress

Where your data lives

The data plane is in your tenancy. Not ours.

Aegis InsightOps Secure uses a hybrid architecture. The intelligence layer runs in IVI's AWS control plane. Your telemetry, prompts, and query results live in your AWS tenancy and never leave it. Cross-account communication uses AWS PrivateLink; there is no public internet exposure for your data.

Your AWS tenancy (data plane)

Your security telemetry, at rest in your accounts
Cribl normalization workers you control
OpenSearch indices encrypted with your KMS keys
Audit logs written to your write-only S3 bucket
All prompts, queries, and model responses

AWS PrivateLink

No public internet

IVI AWS control plane

Authentication and RBAC orchestration
Platform configuration and tenant metadata
Integration templates and connector library
Aegis engineer access via break-glass workflow
No customer telemetry, ever

How AI is used, and how it isn't

Private inference on Amazon Bedrock. No training on your data.

The language model behind Secure runs on Amazon Bedrock within the AWS service boundary. Every contractual and technical guarantee Bedrock provides about data handling is inherited by Secure.

No customer data trains any model.

Bedrock's data handling contract prohibits training on customer inputs or outputs. Your prompts, queries, and response content are not used to improve the underlying foundation models. This is contractual with AWS, not just a Secure setting.

Inference runs in your region.

Bedrock requests are routed to the AWS region you specify during onboarding. Your queries and your data never traverse regions you haven't approved. Cross-region inference is off by default.

Guardrails filter PII and sensitive content.

Bedrock Guardrails run between your application and the model, redacting or blocking PII, financial identifiers, and other sensitive patterns before they reach the model. Guardrails are configurable per customer during deployment.

Prompts and responses stay yours.

Every prompt submitted and every response generated is logged to audit storage in your tenancy. The logs are queryable by your team and never exposed to other tenants. IVI engineers cannot access them without a break-glass workflow you approve.

Identity and access

Your IdP, your RBAC, your audit trail.

Secure does not issue user accounts. Every person who touches the platform does so through your identity provider. Every query is attributable to a named user. Every access is logged.

Federation with your IdP

SAML 2.0 and OIDC federation with Okta, Microsoft Entra ID, Ping, and any standards-compliant IdP. We do not issue local accounts. If an employee leaves your organization, their access to Secure terminates the moment you disable them in your IdP.
RBAC mirrors your organization

Secure inherits your group membership, role assignments, and access policies. Your security team can see security data. Your IT team can see IT data. Your SOC analysts see what their role permits. Nothing bleeds across.
Named-user attribution

Every query, every investigation, every action is attributed to a specific identity. Audit logs capture who asked what, when, and what they received in response. No service accounts run queries on behalf of users.
Break-glass access for Aegis engineers

IVI engineers do not have standing access to your environment. When operational work requires access, a time-bound role is assumed through a break-glass workflow with your approval. Every session is recorded end to end, and the recording is yours.
Audit log immutability

Audit logs are written to an S3 bucket in your tenancy configured as write-only and object-lock enabled. Logs cannot be modified or deleted by either IVI or your own operators for the retention period you define.

Compliance and attestations

Honest about where we are, and what we inherit.

Secure is a newer service. We are deliberate about what we can and cannot claim, and we document the infrastructure-layer attestations we inherit from AWS separately from the service-layer attestations we are pursuing ourselves.

Inherited from AWS

Infrastructure-layer attestations

Because Secure runs entirely on AWS (Bedrock, OpenSearch, Aurora, S3, EKS, and related services), it inherits AWS's infrastructure-layer attestations. These cover physical security, datacenter operations, and AWS service-level controls.

SOC 1 Type II, SOC 2 Type II, SOC 3
ISO 27001, ISO 27017, ISO 27018, ISO 27701
PCI DSS Level 1
HITRUST CSF
FedRAMP Moderate and High (where AWS region supports)
HIPAA-eligible services under AWS BAA

AWS attestation documentation is available through AWS Artifact under your own AWS account.

IVI service-layer

Our own attestations

IVI is pursuing a SOC 2 Type II attestation that covers the Aegis platform controls specific to IVI: personnel access, change management, customer onboarding and offboarding, incident response, and tenant isolation.

SOC 2 Type II: in progress
HITRUST CSF: architecture-aligned (not certified)
NIST CSF: aligned control framework
Evidence package available under NDA during procurement review

We will not claim attestations we don't hold. If a control is specifically required for your buying decision, ask during the assessment and we will tell you straight where we sit.

Healthcare customers

HITRUST-aligned architecture for healthcare environments

Secure's architecture is designed to be deployable within a HITRUST CSF-aligned environment using HIPAA-eligible AWS services under your existing AWS BAA. IVI does not currently execute its own BAA as a service provider; customers who require a direct BAA with IVI should raise this during the assessment so we can scope the engagement accordingly.

Data handling

What we store, where we store it, and how long.

Our goal is that nothing about data handling should be a surprise. If the answer to any of the questions below is not what you need, tell us during the assessment and we will scope accordingly.

Security telemetry

Stored in your AWS tenancy. Encrypted at rest using KMS keys you own. Never leaves your accounts.

Prompts and model responses

Stored in your tenancy audit log. Not used for training. Accessible to you, not to IVI without a break-glass request.

Platform metadata

Tenant configuration, integration settings, and RBAC mappings are stored in the IVI control plane. No customer telemetry. Replicated across availability zones.

Encryption in transit

TLS 1.3 for all external connections. AWS PrivateLink for cross-account traffic. Zero public internet exposure for customer telemetry.

Encryption at rest

AWS KMS customer-managed keys per tenant. You hold the key material, you control rotation, you can revoke access.

Data retention

Configurable per customer. Hot, warm, and cold tier retention are set during the assessment based on your compliance requirements. No published defaults; this is a conversation.

Deletion on offboarding

Terminating the service removes IVI control-plane metadata within 30 days. Your data in your tenancy is yours to retain or delete per your own policy; the service has no ability to reach in after offboarding.

Backups

Operational backups of platform metadata are retained for disaster recovery per AWS best practice. No customer telemetry in IVI backups.

Operational security

How Aegis engineers operate. How IVI operates.

A managed service is only as secure as the team operating it. We treat our own operational posture with the same rigor we expect of our customers.

Personnel vetting

Background checks for every engineer with potential access to customer environments. Annual security training. Non-disclosure obligations in employment contracts.
Change management

Infrastructure changes flow through IaC pull requests with peer review. Production deployments require approval from a second engineer. Emergency change procedures are documented and audited.
Vulnerability management

Continuous scanning of IVI platform images and dependencies. Critical vulnerabilities patched within contractual SLAs. Penetration testing by an independent third party on a regular cadence.
Incident response

24/7 monitoring of the Aegis control plane. Customer notification for incidents affecting service availability or data handling within 72 hours, consistent with common breach-notification norms. Post-incident reports delivered to affected customers.
Access to customer environments

No standing access. Every engineer session into a customer environment is initiated through a break-glass workflow, authenticated through customer SSO, time-bound, and session-recorded. Recordings are retained per customer retention policy.

Subprocessors

Two named subprocessors. Full list in the DPA.

Secure uses a small number of named subprocessors to deliver the service. The two foundational ones are named publicly here. The full subprocessor inventory, including any that support operational tooling, is maintained in our Data Processing Addendum available during procurement review.

Amazon Web Services (AWS)

Infrastructure, AI inference, data storage

AWS is the cloud platform under Secure. Bedrock provides AI inference, OpenSearch and S3 provide data storage, and AWS Organizations provides tenant isolation. Data stays in the AWS region you select.

Cribl

Data normalization and routing

Cribl Stream is the data pipeline that normalizes security telemetry before it enters the intelligence layer. Cribl workers run within your AWS tenancy; Cribl the company does not receive your data.

Changes to the subprocessor list trigger advance notice per the Data Processing Addendum. Customers may object to material changes within the notice period.

Questions security teams ask

The things TPRM, GRC, and CISO teams raise during review.

Where does the AI model physically run?

Amazon Bedrock, in the AWS region you specify during onboarding. The model is not deployed on IVI infrastructure; it's an AWS-managed service that runs in your selected region. Cross-region inference is off by default.

Who has access to our prompts and queries?

Only your users, through your IdP. Prompts and responses are logged to audit storage in your AWS tenancy. IVI engineers cannot read them without a break-glass workflow that requires your approval and records the full session.

Do your engineers have standing access to our environment?

No. Every engineer session into a customer environment is initiated through a break-glass workflow with your approval, authenticated through your SSO, time-bound, and session-recorded. No always-on access, ever.

Can we run Secure in GovCloud or another isolated region?

Secure deploys into the AWS region family you specify. GovCloud-specific deployment is possible where AWS Bedrock and the other required services are available, subject to scoping. Raise the requirement during the assessment and we will confirm feasibility for your specific region and compliance scope.

What happens to our data if we terminate the service?

Your data stays where it always was: in your AWS tenancy. Terminating Secure removes IVI's control-plane access within 30 days and deletes IVI-side metadata. The telemetry, logs, and audit records in your tenancy are yours to retain or delete per your own policies; IVI has no ability to touch them after offboarding.

Is customer data used to train AI models?

No. AWS Bedrock's data handling contract prohibits training on customer inputs or outputs. Your prompts, queries, and model responses are not used to train foundation models. This is a contractual guarantee from AWS, not an IVI setting that could change.

Do you have a SOC 2 report?

IVI's SOC 2 Type II attestation for the Aegis platform is in progress. AWS's SOC 2 Type II covers the infrastructure layer and is available through AWS Artifact under your own AWS account. During procurement review, we provide our current evidence package under NDA and discuss attestation timelines if they are material to your decision.

Can we get a DPA, BAA, or other data-specific agreement?

DPA: yes, available during procurement review. BAA: we do not currently execute our own BAA as a service provider. The Secure architecture is designed to be deployable within HIPAA-eligible AWS services under your existing AWS BAA; the implementation pattern is scoped during the assessment. Customers who require a direct BAA with IVI should raise this during scoping so we can address it before SOW.

What is your breach notification SLA?

Customer notification for security incidents affecting data handling or confidentiality within 72 hours of verified detection, consistent with common breach-notification norms. Faster notification for customers subject to regulatory regimes that require shorter windows (e.g., New York DFS, HIPAA in certain scenarios) can be contractually agreed during scoping.

Do you perform penetration testing?

Independent third-party penetration testing of the Aegis platform on a regular cadence. Summary results available under NDA during procurement review. Customer-specific penetration testing of their tenant (e.g., to satisfy internal security requirements) is accommodated under a pre-notified testing window.

Request documentation

Evaluating Secure? Here is what we can share.

This form routes to our security and trust operations. If you are in the middle of procurement review, TPRM assessment, or vendor intake, tell us what you need and we will get back within one business day with what we can share under NDA.

Data Processing Addendum (DPA)
Current security and compliance evidence package under NDA
Architecture diagrams and data flow documentation
Subprocessor inventory and change policy
Answers to your standard TPRM or CAIQ questionnaire

First name *

Last name *

Work email *

Company *

Your role *

What are you looking for? *

Data Processing Addendum (DPA) Security evidence package under NDA Architecture diagrams Subprocessor inventory Answers to our TPRM or CAIQ questionnaire Something else (describe below)

Context (optional)

Resource Directory

78 resources

Featured Resources

other Aegis Total Value Calculator | Intelligent Visibility

Quantify ROI from co-managed network services including monitoring, incident response, and lifecycle management savings.

Managed Services