Read capability and evidence status

Versioned governance intent, assigned independent review and explicit activation, with execution limits kept separate.

Supported environments: Authorized tenant console

Evidence: Proposal versions; Reviewer decisions; Audit events

Inputs: Seven-domain governance proposal; Current reviewer assignments; Attributed evidence references and attestations

Outputs: Versioned proposal and digest; Review decisions; Activation and change history

Dependencies: Current tenant membership; Supported lifecycle and validation rules

Controls: Server-derived authority; Exact-content approval binding; Author separation; Explicit activation

Security considerations: Tenant-scoped access; Current membership and assignment checks

Verified: 2026-11-06

Qualification: Implemented authoring and lifecycle controls; runtime applicability and deployment qualification remain separate.

• Active descriptive Alignment holds ordinary text and payload execution until its semantics have a supported executable adapter.
• A reference or attestation is not independent verification of source truth.

A tenant project registry with current-member accountability, declared responsibilities, workload references and exception history.

Supported environments: Authorized tenant console outside managed evaluation configuration

Evidence: Immutable change events; Linked original assessment evidence; Event-time project evidence where recorded

Inputs: Project name and purpose; Current accountable member; Optional responsibility roles; Application/environment reference

Outputs: Stable draft project record; Revision history; Accountable exception responses

Dependencies: Current tenant membership; Configuration-manager authority for writes

Controls: Optimistic revision checks; Tenant-scoped workload uniqueness; Transactional audit and outbox

Security considerations: Managers manage; responsible members have scoped read access; Responsibility metadata does not grant approval permissions

Verified: 2026-11-06

Qualification: Registry and access boundaries qualified with synthetic PostgreSQL and console workflows; no enterprise inventory completeness claim.

• Registration remains draft and does not admit runtime traffic.
• Exception response does not alter the original policy result or automatically resume a held request.

A separately enabled, narrowly qualified authorization lifecycle for the deterministic payload boundary.

Supported environments: Explicitly enabled controlled environments

Evidence: Authorization and activation history; Canonical project context and release audit

Inputs: Exact project/workload/policy revisions; Owner acknowledgement; Independent officer decision; Manager activation; Bounded validity window

Outputs: Immutable authorization subject; Sticky application enrollment; Event-time execution context

Dependencies: Current owner and approver membership epochs; Supported payload-sanitize-v1 boundary; Server feature enablement

Controls: Independent owner/reviewer/activation responsibilities; Exact scope binding; Final authority refresh; Suspension and revocation hold

Security considerations: Default-off enablement; Enrolled applications cannot silently fall back to legacy authority

Verified: 2026-11-06

Qualification: Controlled local lifecycle, real disposable PostgreSQL concurrency and synthetic HTTP qualification; not general enterprise availability.

• No unrestricted agent, provider or financial authority.
• Disabling the feature holds enrolled workloads.
• Active unsupported Alignment still holds execution.

Explicit time-bounded administration and immutable, independently reviewed four-role policies.

Supported environments: Explicitly enabled authorized console

Evidence: Immutable policy decisions; Assignment and delegation audit

Inputs: Platform-issued delegation; Current reviewer assignments; Versioned review-policy proposal

Outputs: Delegation history; Independent decisions; Effective review-policy revision

Dependencies: Server enablement; Current membership epochs; Four distinct reviewers

Controls: No self/onward delegation; No author approval; Expiry and revocation; Policy-change invalidation

Security considerations: Tenant scope; Role removal/rejoin does not revive old authority

Verified: 2026-11-06

Qualification: Implemented behind a default-off server flag; controlled qualification does not imply enabled customer access.

• Reduced-review policy activation is not qualified.
• Administrative delegation is not permission to approve another person's decision or change financial limits.

Adoption contexts for explaining proportionate accountability; expanded profile configuration and contextual reporting are being developed.

Supported environments: Planning and scoped evaluation

Evidence: Scoped evaluation evidence, where available

Inputs: People, shared work and managed workload context; Intended control boundary

Outputs: Agreed governance scope; Proposed ownership and evidence requirements

Dependencies: Qualified integration; Organizational authority

Controls: Existing supported controls apply only where integrated

Security considerations: Context labels confer no permissions

Verified: 2026-11-06

Qualification: Explanatory architecture; expanded configuration and reporting remain In Development.

• These are contexts, not pricing tiers or automatically discovered estates.
• No broad profile reporting or universal end-user coverage is claimed.

Configured policy checks and evidence within supported integration boundaries, with legal interpretation kept separate.

Supported environments: Supported routed requests; Bounded assessments and evaluations

Evidence: Decision and final-boundary records

Inputs: Approved technical policy; Supported content; Reviewed scope

Outputs: Findings; Policy decisions; Enforcement evidence

Dependencies: Enabled validators; Exact policy assignment; Qualified integration

Controls: Input evaluation; Supported transformation and revalidation; Withholding; Traceable policy version

Security considerations: Identity and policy resolved server-side

Verified: 2026-11-06

Qualification: Available for supported evaluation and qualified integrations; no regulatory certification.

• Technical controls do not establish legal compliance or cover every obligation.
• Regulatory research coverage is not executable jurisdiction coverage.

An expanding operating model for AI observability, consumption and control across qualified organizational integrations.

Supported environments: Architecture and controlled development

Evidence: Current decision evidence only within supported boundaries

Inputs: Future qualified observation and usage sources; Ownership and integration inventory

Outputs: Planned operational visibility and reporting

Dependencies: Network and provider integrations; Attribution and completeness definitions

Controls: Existing Gateway controls remain bounded to routed requests

Security considerations: Unknown visibility must remain unknown

Verified: 2026-11-06

Qualification: Network-oriented harness, Shadow AI Intelligence and Tokenomics remain under development.

• No organization-wide observation or autonomous remediation is claimed.
• Dashboard illustration is not measured enterprise capacity.

Planned enterprise egress integrations beyond the existing routed Gateway boundary.

Supported environments: Proposed customer network integrations

Evidence: Actual Gateway receipts, not inferred organization-wide coverage

Inputs: Qualified traffic sources; Destination inventory; Approved data boundary

Outputs: Planned observation and enforcement coverage

Dependencies: Actual traffic routing; TLS and identity design; Failure-mode qualification

Controls: Existing inline evaluation for supported routed traffic

Security considerations: Destination controls at the configured adapter boundary

Verified: 2026-11-06

Qualification: Existing Gateway is implemented; enterprise network harness integrations are not released.

• Not a universal network interception product.
• Traffic that bypasses an integration is outside its known enforcement coverage.

Planned source-based discovery and attribution of AI use outside the approved operating model.

Supported environments: Future qualified integrations

Evidence: Source, observation time and confidence will be required

Inputs: Approved source records; Identity and application inventory

Outputs: Planned attributed observations and review queues

Dependencies: Source access; Attribution quality; Deduplication and coverage definitions

Controls: Human review of inferred observations

Security considerations: Least-privilege collection and data minimization

Verified: 2026-11-06

Qualification: Design direction; do not interpret existing Discovery Scanner findings as Shadow AI inventory.

• No current all-tool discovery capability.
• Observed traffic does not by itself prove unsanctioned use or identify the accountable person.

Planned usage, allocation and financial visibility with explicit measurement provenance.

Supported environments: Future qualified provider and billing integrations

Evidence: Usage/rate/source provenance required

Inputs: Observed usage; Versioned rates; Attribution and invoice context

Outputs: Planned consumption and allocation reporting

Dependencies: Provider-specific usage semantics; Currency and rate provenance; Financial ownership design

Controls: Separate actual, estimated and unavailable values

Security considerations: Financial responsibility metadata is not spending authority

Verified: 2026-11-06

Qualification: Existing usage metadata fields do not establish complete FinOps or realized savings.

• No released enterprise budget admission or invoice reconciliation.
• Missing usage is not zero; execution budgets are not monetary limits.

Deterministic evaluation of supported ingress and egress content under exact assigned policies.

Supported environments: Qualified Gateway integrations; Controlled evaluation

Evidence: Policy/version and validator provenance; Final outcome and artifact digest

Inputs: Authenticated request; Supported text or JSON; Exact policy version

Outputs: Rule findings; Engine decision; Transformed or withheld artifact

Dependencies: Application identity; Policy resolution; Supported validators

Controls: Pre-effect evaluation; Server-side transformation; Revalidation; Final release checks

Security considerations: Unsupported input and authority fail closed

Verified: 2026-11-06

Qualification: Bounded route and adapter behavior tested; customer deployment remains separately qualified.

• Detection is scoped and heuristic.
• No complete de-identification, prompt-injection immunity or clinical validity claim.

Bundled validator selection, rule weights and decision thresholds resolved by exact ID and version.

Supported environments: Configured tenant/application/environment

Evidence: Effective pack ID/version; Rule and validator identifiers

Inputs: Approved pack assignment; Validator findings

Outputs: Weighted policy decision; Version provenance

Dependencies: Installed pack and validator versions

Controls: Role-protected assignment; Ambiguous/missing authority rejection

Security considerations: Client headers cannot select arbitrary production authority

Verified: 2026-11-06

Qualification: Bundled technical catalog is implemented; new controls and jurisdiction packs need explicit qualification.

• A pack name is not a certification.
• A Pulse dossier or proposed jurisdiction mapping is not automatically an installed runtime pack.

Registered skills, constrained transport and exact final-publication checks for supported operations.

Supported environments: Supported Gateway and connector boundaries

Evidence: Intent and terminal audit records; Final artifact fingerprint where applicable

Inputs: Server-resolved authority; Registered skill/action; Permitted destination and schema

Outputs: Admitted or held execution; Canonical terminal evidence

Dependencies: Known adapter; Current credential and authority state

Controls: Deadline/cancellation checks; Scope and destination binding; Atomic terminal publication

Security considerations: Proposals cannot expand grants; Unknown effects require reconciliation

Verified: 2026-11-06

Qualification: Qualified mechanisms apply to enumerated boundaries only, with separate simulated paths.

• Not arbitrary agent orchestration or universal tool execution.
• Final commit authorizes preparation/release; it is not proof of external delivery.

Opt-in, time-bounded review of a retained exact output for supported nonclinical applications.

Supported environments: Explicitly enrolled nonclinical applications

Evidence: Review view and decision records; Release evidence

Inputs: Held artifact; Current reviewer authority; Original application audience

Outputs: Approve/reject/revoke history; Eligible release to the original audience

Dependencies: Supported review profile; Unexpired original grant; Encrypted artifact storage

Controls: Exact artifact and context binding; Expiry and revocation; Original audience restriction

Security considerations: Held content excluded from ordinary telemetry

Verified: 2026-11-06

Qualification: Implemented for the named nonclinical profile; integration and data handling require qualification.

• Not clinical approval or general task resume.
• A HELD result does not prove a reviewer saw or approved anything.

Tenant-scoped canonical outcomes, policy provenance and correlated exports with explicit missing/conflicting states.

Supported environments: Authorized console and evaluation workspace

Evidence: Request ID; Policy/version; Outcome/stage; Artifact digest

Inputs: Canonical decision; Terminal audit/outbox; Scoped filter

Outputs: Decision detail; JSON/HTML/PDF evidence where supported; Conflicts and unknowns

Dependencies: Retained evidence; Current read authorization

Controls: Shared projection for totals and excerpts; Event-time context validation; No fabricated missing receipts

Security considerations: Tenant and audience scoping; Derived evidence rather than unrestricted raw payloads

Verified: 2026-11-06

Qualification: Local reconciliation and negative tests support the stated projection; external custody is not certified.

• Hashes are not anonymization or independent attestation.
• Some diagnostic and exact-output review storage has a separate raw-content boundary.

Scoped final outcomes and evidence views with defined windows, excerpts and provenance.

Supported environments: Authorized tenant console; Publication-filtered demonstration telemetry

Evidence: Source and time window; Final outcome versus earlier decision

Inputs: Retained canonical and qualified legacy records

Outputs: Outcome totals; Bounded detail excerpts; Evidence exports

Dependencies: Shared report projection; Publication scope for public data

Controls: Conflict exclusion; Deduplication; Explicit scope and freshness

Security considerations: Public telemetry excludes private tenant data

Verified: 2026-11-06

Qualification: Supported reporting projections tested; customer metrics require agreed denominators and retention scope.

• Displayed totals are not complete enterprise inventory.
• Unavailable data is not a zero result; event timestamps are not end-to-end latency measurements.

Durable assessment of structured cases, with explicit evidence limits and accountable follow-up.

Supported environments: Authorized tenant assessment workspace

Evidence: Assessment inputs and results within authorized storage scope

Inputs: Supported structured case; Declared context and evidence references

Outputs: Assessment findings; Retained assessment record; Linked exception follow-up

Dependencies: Validated input shape; Tenant authorization; Supported deterministic assessment

Controls: Durable scoped processing; Structural validation; History retention

Security considerations: Scoped access and separate uploaded-material retention

Verified: 2026-11-06

Qualification: Implemented deterministic assessment path; do not imply unrestricted model-based reasoning.

• Structural support is not factual grounding or clinical correctness.
• The assessment model broker and authoritative retrieval remain separate unfinished capabilities.

Public no-code exploration of supported policy decisions, governed outputs and inspectable evidence.

Supported environments: Public configured demonstration workspace

Evidence: Request-correlated demonstration records

Inputs: Guided synthetic scenario; Supported non-sensitive custom JSON; Explicit connector mode

Outputs: Observed policy result; Governed output; Demonstration evidence and telemetry

Dependencies: Configured demo policy and workspace

Controls: Server-selected guided fixtures; Enforced custom content; Explicit simulation labels

Security considerations: Demo identity remains separate from signed-in private tenant identity

Verified: 2026-11-06

Qualification: Controlled local behavior and presentation qualified; production customer integrations are not inferred.

• Simulated FHIR/override views are not real provider effects or production approvals.
• Use synthetic or non-sensitive data.

A private, bounded technical evaluation with controlled payload, scenario and supported action exercises.

Supported environments: Enabled hosted or qualified local evaluation workspace

Evidence: Actual local evaluation receipts; Separated fixed-provider scenarios

Inputs: Verified enrollment; Scoped evaluation key; Synthetic or non-sensitive test cases

Outputs: Governed HTTP results; Owner-scoped evidence; Downloaded exercises and reports

Dependencies: Evaluation availability; Activation, quota and expiry rules

Controls: Server-selected policy choices; Shared quota; Ownership and expiry enforcement

Security considerations: Write-only raw keys; Owner-scoped reports; Managed workspace configuration

Verified: 2026-11-06

Qualification: Existing actual HTTP acceptance covers synthetic evaluation boundaries; customer outcomes remain unmeasured.

• No production entitlement, unrestricted provider calls or unlimited traffic.
• Application-kit callback effects are local and bounded; they do not qualify arbitrary tools.

Restricted-distribution local assessment of approved files, with separately qualified source-connection options.

Supported environments: Supported native Windows x64 and macOS Intel packages; Agreed customer assessment scope

Evidence: Assessment source and finding references

Inputs: Approved local files; Optional separately configured SQL Server source

Outputs: Local assessment report; Findings and coverage gaps

Dependencies: Matching release/package qualification; Source permissions and supported formats

Controls: Read-only collection scope; Local evaluation; Customer-controlled report sharing

Security considerations: No automatic result upload in the offline offering; Least-privilege source access

Verified: 2026-11-06

Qualification: Available through a scoped engagement; native artifact, customer driver and source qualification remain specific.

• Findings do not remediate source data or enforce live AI traffic.
• SQL Server and separate PostgreSQL CLI capabilities must not be conflated with every native workspace source type.

U.S.-first, source-referenced state legislative research with published coverage and maturity distinctions.

Supported environments: Public source inventory and reviewed publication paths

Evidence: Source links; Retrieval and publication context

Inputs: Published state source records; Official links and provider metadata; Reviewed obligations where available

Outputs: Source inventory; Legislative status context; Reviewed mappings where published

Dependencies: Source retrieval and publication workflow; Human interpretation and applicability review

Controls: Publication-scoped data; Source/status/date distinctions

Security considerations: Draft/private records excluded from public projection

Verified: 2026-11-06

Qualification: Published state inventory is implemented; source-to-obligation-to-execution completeness varies by reviewed record.

• State count is not executable pack count or compliance coverage.
• No federal, global or automatically current legal coverage claim.
• Applicability is not inferred from IP address or selected state.

Bounded text-provider adapters and explicitly scoped connector paths.

Supported environments: Qualified tenant provider configuration; Authorized Epic sandbox read/search; Network-free simulated connectors

Evidence: Provider/context metadata and supported terminal receipts

Inputs: Approved endpoint; Tenant-bound secret reference; Supported operation

Outputs: Untrusted upstream result for egress evaluation; Transport and execution evidence

Dependencies: Network and credential qualification; Known adapter contract

Controls: HTTPS destination admission; Redirect rejection; Bounded bytes/retries/deadlines; Tenant-bound credentials

Security considerations: Authenticated encryption for supported stored provider secrets; No provider key in public evidence

Verified: 2026-11-06

Qualification: Adapter contract tests do not prove every upstream model or deployment works.

• No universal provider API compatibility, general MCP/tool gateway or clinical write support.
• Real upstream and customer networks need separate qualification.

A scoped delivery and shared-responsibility engagement for an agreed customer-controlled environment.

Supported environments: Agreed customer VPC or on-premises topology

Evidence: Agreed deployment acceptance records

Inputs: Deployment scope; Infrastructure and identity requirements; Acceptance and support agreement

Outputs: Agreed deployment package and installation plan; Environment-specific qualification evidence

Dependencies: Customer infrastructure; PostgreSQL operations; Network and secret provisioning

Controls: Versioned installation and migration plan; Defined update/backup/incident ownership

Security considerations: Customer-specific network and data-flow review

Verified: 2026-11-06

Qualification: Offered through a coordinated engagement; no customer-specific qualification is asserted by this catalog.

• Not an automatically qualified topology or a promise of air-gapped operation.
• External providers/connectors may cross the deployment boundary.

Workload-specific validation of throughput, concurrency, recovery and operating coverage.

Supported environments: Controlled tests and future customer qualification

Evidence: Per-run receipts and source versions where measured

Inputs: Representative workload; Integration inventory; Latency and recovery acceptance criteria

Outputs: Measured qualification results where actually run

Dependencies: Defined topology; Representative data; Instrumented load and failure tests

Controls: Explicit denominators and test conditions; Separation of illustration and measurement

Security considerations: Capacity must preserve isolation and fail-closed behavior

Verified: 2026-11-06

Qualification: No customer-qualified scale result is claimed; qualification is required for the intended workload.

• The enterprise-scale website example uses illustrative data, not a capacity benchmark.
• No supported agent-count, latency or availability guarantee follows.

Implemented identity, tenant isolation, secret and release controls with documented operational limits.

Supported environments: Supported application and Gateway paths

Evidence: Boundary tests and scoped audit records

Inputs: Authenticated identity; Authorized tenant/app scope; Configured secrets and destinations

Outputs: Scoped decisions and protected evidence

Dependencies: Secure deployment configuration; Session/key lifecycle; Current software releases

Controls: Server-side authorization; Revocation and expiry; Constrained outbound requests; Atomic evidence publication

Security considerations: Defense depends on enabled integrations and deployed operations

Verified: 2026-11-06

Qualification: Implemented controls require deployed configuration, recovery and independent assessment for intended use.

• No assertion of breach prevention, universal security or certification.
• Uploaded material, diagnostic snapshots and held review artifacts have distinct retention and access boundaries.

Early SOC 2 readiness discussions with Vanta; no completed attestation is claimed.

Supported environments: Assurance planning

Evidence: User-confirmed early Vanta discussions; no audit report supplied

Inputs: Proposed control scope; Operating evidence and auditor requirements

Outputs: Readiness planning, when agreed

Dependencies: Defined audit scope; Operating-period evidence; Independent auditor work

Controls: Claims separated from independent attestation

Security considerations: Readiness tooling is not an auditor opinion

Verified: 2026-11-06

Qualification: Planning fact supplied and approved by the owner; no certification evidence inferred.

• LSAS does not claim SOC 2 Type I/II attestation, ISO certification or equivalent external assurance.