Use cases
Four teams, four AI questions, four reconciliations today. Engineering, finance, security review, and AI program leadership instead inspect the same evidence trail.
The same no-tagging attribution record supports engineering, finance, security review, and AI program decisions.
Engineering lens
Engineering sees service, repository owner, deploy SHA, model or provider change, expected cost delta, and confidence state in one record. Triggers include a model changed in a PR, a fallback route changed, a reasoning model introduced, or a batch job whose volume spiked.
Decision Should this model choice ship, be reviewed, or be shaped toward a lower-cost option?
Illustrative example
Pre-deploy decision record
Finance lens
Finance reads spend split into owned, contested, and unresolved. Each is tied to a budget path with its confidence basis, so a chargeback line is defensible before close.
Decision Which AI spend can be attributed, forecasted, and defended?
Illustrative example
Budget attribution ledger
Security review lens
Security sees shared keys, workload owner, resolver basis, Gate posture, access review state, and unresolved ownership. Everything stays advisory by default; Gate enforcement applies only on explicit opt-in.
Decision Which usage is safe to leave advisory, and which usage needs review before any Gate opt-in?
Illustrative example
Access review record
AI program leadership lens
AI program leadership sees workflow-level adoption and team-level consumption patterns as an aggregated projection of the attribution graph.
Decision Which teams have embedded AI into real workflows, which teams only have licensed access, and where would enablement change adoption?
Illustrative example
Adoption projection
Design partner review
No production data is required for the first conversation. Start with one workload, one unclear owner or budget path, and the decision your team cannot make confidently today.
Useful starting points