Control stays with you
AmrudAi’s documented security architecture protects identities, tenant boundaries, credentials and diagnostic evidence across customer-controlled enterprise and MSP deployments. Scoped access, authenticated collection, privacy policies, maker-checker approvals and deterministic AI tool controls define the operating boundaries. These design controls require release-specific validation and do not constitute a compliance certification.
The application experience
See the complete workspace. Enlarge any image to inspect its panels, navigation and evidence at actual size.
Complete the operational workflow
Manage users, roles, resource scope, sessions, API clients and credential profiles in one enterprise operating model. SSO and directory lifecycle, including SCIM where the integration profile supports it, preserve tenant and authorization boundaries.
Monitor Sensor and ingestion quality, storage pressure, retention, housekeeping and restore validation. Distinguish telemetry degradation from target failure, and prepare masked support evidence with the applicable data controls.
Review capacity and license consumption before scope changes. Migration Studio supports mapping, reconciliation and governed adoption; readiness checks, backup evidence and recovery runbooks keep platform operations accountable.
The capability in detail
Bind Sensor and Agent enrollment to deployment, tenant, site and role. Use short-lived, single-use enrollment and unique service identities.
Require encrypted, authenticated Platform-owned telemetry and control paths, with certificate trust checks, rotation and revocation.
Enforce roles and resource scope on the server for inventory, diagnostics, evidence, reports, credentials and actions.
Separate permission to use a credential from permission to reveal it. Keep reusable secrets out of routine telemetry, logs, audit output and AI context, and validate rotated credentials before activation.
Apply command policy, step-up or maker-checker where required, immutable action definitions, impact controls and post-action evidence.
Retain source, timestamps, quality and versioned provenance; preserve RCA revisions and detect tampering with consequential audit records.
Collect focused operational metadata under policy. Exclude unnecessary content, environments, unrestricted command lines and payloads from routine diagnostics. Review explicit Drop, Mask, Hash and Store policies for sensitive fields.
Validate connector destinations, requests, signatures, replay, schemas and rate limits. External tickets and comments cannot become privileged commands.
Use authorized, minimized context, data-egress controls and a deterministic tool firewall. Treat retrieved logs, notes and configurations as untrusted inputs.
Track export, credential revocation, active-data deletion, backup/archive expiry and legal holds through explicit lifecycle states.
Preserve deterministic monitoring without an external AI service. Approved local runtimes and model artifacts follow signed, versioned offline packaging.
Inspect signed packages, software-component records and vulnerability information. Controlled upgrades, rollback, backup and restore workflows preserve a traceable operating history.
From information to action
Agree the customer, sites, network paths, access model and required availability profile.
Validate identities, enrollment, credentials, permissions and connector scope.
Assess supported versions, security tests, package provenance, recovery and measured resource profiles.
Monitor system health, rotate access, review audit records and control changes throughout the lifecycle.
Deployment and operational control
Deploy a dedicated customer environment or a shared MSP service with scoped identity, data isolation and operational governance. Choose standalone Core or two-Core Active/Standby continuity.
Review identity and access controls, authenticated transport, secret handling, privileged-action governance, tenant isolation, retention and audit evidence. Map these controls to your organization’s requirements and review any independent assurance evidence relevant to the deployment.
They must not. Legacy monitored-device compatibility is a separately controlled mode and cannot downgrade the authenticated Platform-owned transport boundary.
No. HA addresses continuity through defined failures; DR requires its own backup, restore, site-loss and recovery evidence. The deployment profile and contractual objectives must state both clearly.
The offboarding model distinguishes active deletion, retained backups/archives, scheduled expiry and legal hold. Final evidence must identify what was deleted, what remains and its governing retention.
No. External input and generated output are proposals or evidence. Server-side identity, resource policy, approvals and audit remain authoritative.
Review controls and their evidence
AmrudAi’s security model describes twelve governance areas spanning identity, transport, permissions, secrets, privileged actions, evidence, privacy, integrations, AI, retention, offline operation and software integrity. These are documented requirements and review boundaries. Their presence in the design does not establish implementation, independent assurance or certification for a particular release or deployment.
| Governance area | Documented control | Boundary to verify |
|---|---|---|
| Identity at every boundary | Scoped enrollment, short-lived single-use authority and unique Sensor or Agent identity. | Enrollment establishes trust; monitored activation still needs the applicable approval. |
| Authenticated internal transport | Encrypted authenticated telemetry and control, certificate trust and lifecycle controls. | Legacy device compatibility cannot downgrade Platform-owned transport. |
| Least-privilege resource access | Server-side roles, tenant/resource scope, action conditions and restricted API access. | Verify permission enforcement and cross-scope negative cases in the selected release. |
| Credentials without unnecessary exposure | Encrypted credential profiles with separate USE and REVEAL permissions. | Reusable secrets stay out of routine logs, audit output, telemetry and AI context. |
| Governed privileged actions | Command policy, step-up controls, approval, bounded targets and post-action evidence. | Model output and external approval cannot grant local execution authority. |
| Evidence integrity | Source, timestamps, quality, versioned provenance and verifiable audit integrity. | Tamper evidence requires a validated mechanism; the design alone is not assurance. |
| Privacy-conscious diagnostics | Bounded operational metadata and field-level Drop, Mask, Hash or Store policies. | Collection scope, retention and field handling must match the approved profile. |
| Bounded access to external systems | Connector destination, authentication, replay, schema and rate controls. | Confirm each connector and permission scope; external text is untrusted input. |
| AI data and tool boundaries | Minimized authorized context, provider/region policy and a deterministic tool firewall. | Untrusted logs, tickets and configurations cannot become privileged instructions. |
| Retention and governed offboarding | Controlled export, credential revocation, active deletion, archive expiry and legal holds. | Active deletion does not imply immediate erasure of retained backup copies. |
| Offline and optional AI operation | Deterministic monitoring without an external model and governed local-provider packaging. | Confirm the approved runtime, model artifact, deployment and resource profile. |
| Software integrity and recovery | Signed packages, component records, vulnerability review, upgrades, backup and restore. | Review release evidence; backup success, recovery and HA are different outcomes. |
Review actual release tests, operating procedures and independent assurance evidence against your organization’s requirements.
AmrudAi documents short-lived, single-use enrollment authority scoped to the tenant and deployment, with optional site, region, pool or role limits. Endpoints generate their private keys locally and establish unique authenticated identities. Enrollment does not automatically authorize normal monitoring: review, permissions and the applicable activation policy remain separate controls.
AmrudAi documents four field-handling choices: Drop omits the field, Mask redacts selected content, Hash replaces it with a derived value, and Store retains permitted content. The configured policy determines the applicable treatment. Retention, access and AI data boundaries still apply; hashing by itself is not a guarantee of anonymity.
AmrudAi treats logs, tickets, configuration text and retrieved content as untrusted evidence in its documented AI design. A deterministic allowlist checks structured action proposals against identity, tenant and resource scope, parameters, targets, approval and resource limits. Model output cannot create privileges, bypass local policy or directly execute a device change.
Publisher: AmrudAi . This content is checked against documented product architecture and cited technical sources. Content updated: . Descriptions of planned or adapter-dependent capabilities are not a release certification. Confirm the supported version and acceptance evidence during evaluation.
Explore the technology, service dependencies and operating requirements that matter to your team.
Search the complete website · Esc to close · Tab to navigate results