Skip to content

Synthetic demo · No customer connectors attached · Fixture snapshot 24 Sep 2026, 06:15 UTC

Lifecycle

AI use-case inventory and stage matrix

Each row is a registered AI use case; each column is a lifecycle stage from the coverage contract. Unknown is drawn with a dashed outline and not applicable is drawn flat, so neither can be mistaken for a pass.

Observed at 24 Sep 2026, 06:15 UTC5 of 5 fixture use cases shown · stage badges are not release-gate verification

AI lifecycle × DSPM cycle

The same eight-step review repeats within each of the 14 lifecycle stages; this is an illustrative planning view, not an executed control.

Sample action · A01 / C01

Inventory the purpose register and owner.

Enumerate assets and movement; record inventory denominator and inaccessible surfaces.

Evidence expectation

A dated approved purpose and named owner, with scope and reviewer attached.

Illustrative frontend only · no collection, enforcement or verification is performed here.

Open related sample screen →

Filters

Narrow the matrix by stage, model, dataset or environment.

PassPartialFailStaleUnknownNot applicable

Stage and control matrix

Select any cell to read the recorded note and evidence age for that stage.

Lifecycle stage status per AI use case
Use caseA01A02A03A04A05A06A07A08A09A10A11A12A13A14Owner / evidence
Customer Support CopilothighproductionPriya Ramanlast evidence 5 min ago · review due 30 Nov 2026
Contract Clause AnalystcriticalproductionDaniel Okoyelast evidence 9 h ago · review due 15 Oct 2026
Internal Code AgenthighstagingMei Tanakalast evidence 1 h ago · review due 2 Oct 2026
HR Insights AssistantcriticaldevelopmentNo accountable ownerlast evidence 16 d ago · review due 30 Sep 2026
Air-Gapped Field Advisormediumair-gappedSofia Lindqvistlast evidence 3 d ago · review due 1 Dec 2026

Stage detail

Nothing selected yet.

Select a cell

Choose any status cell in the matrix to read its recorded note, evidence age and linked records.

Stage reference

A01–A14 of the lifecycle coverage contract.

Scroll table sideways for more columns →

StageRequired security work
A01 — Purpose and ownershipUse case, owner, permitted data and actions, retirement responsibility.
A02 — Acquire data and providersOrigin, permissions, processing conditions, regions and unknowns.
A03 — Discover and collectInventory and classify raw, shadow and transient data with coverage receipts.
A04 — Prepare and labelCleaning, joins, chunking, labelling, synthetic data and inherited constraints.
A05 — Select model and architectureBase weights, licensing, retention options, isolation and supply chain.
A06 — Train and adaptRun and input manifests, training identities, secured checkpoints.
A07 — Build RAG, agents and toolsSource versions and ACLs bound to indexes; memory, tools and identities.
A08 — Evaluate and approveQuality, privacy leakage, adversarial paths, access control and isolation tests.
A09 — Package, release, deployRelease manifest, approval, observed deployment and rollback candidate.
A10 — Infer and actPrompt and output inspection, retrieval authorization, tenant and egress control.
A11 — Monitor and respondDrift detection, blind intervals, investigation and verified closure.
A12 — Change and improveFeedback, reindexing, retraining, provider swaps and rollback governance.
A13 — Retire the AI systemStop activity, drain work, disable identities, inventory retained derivatives.
A14 — Retain, dispose and verifyHolds, expiry, item-level deletion ledger, backups and model residuals.