Intelligence

AI Cost & Usage

Governed LLM spend for the selected window. This view rolls up the live LLM audit ledger by tenant, agent, model, tokens, latency, and attributed USD cost.
Cost governance proof case
INV-LIVE-DUP-1778594510 · OceanFloor Subsea Services Ltd · £2.8M
Policy FIN-P2P-DUP-04 · Evidence EP-P2P-OCEANFLOOR-LIVE · PO PO-100421 · $2.3M protected.
Latest Runtime Capability Update: governed chat, WIIF assistant portability, Token Economy controls, fail-closed policy, deterministic replay, signed evidence, runtime adoption audit, case-state propagation, and observability readiness.
Open P2P case Evidence pack Replay
Total Cost
$0.00
Calls
1
Input Tokens
812
Output Tokens
146
Avg Cost / Call
$0.00
Token Economy Control Plane

Business outcome per token

Token economy is governed as a platform capability: deterministic-first routing, agent entitlements, prompt contracts, context compression, cache-key governance, quality gates, and value attribution.
Value / 1K Tokens
$13.31
Confidence: estimated
Routing Efficiency
100.0%
Higher means fewer deterministic tasks spent tokens.
Cache Opportunity
0.0%
Repeat prompt+context patterns that can be reused.
Tokens / Dollar Value
75.1373
Lower is better; derived from value ledger or marked estimate.
Mission-level compute cost
Derived from iaf_llm_audit_log by grouping audited LLM calls into finance missions using explicit evidence metadata first and agent naming second. This is computed telemetry, not a financial-impact estimate.
Avoided SAP calls: not instrumented in this ledger yet. Do not claim SAP-read avoidance from token spend or value-protected metrics; use the Enterprise Read Optimization surface until the SOR gateway emits per-decision call counters.
MissionCallsInput TokensOutput TokensCostAvg LatencyAgents / Models
Unmapped1812146$0.00184.0 msenterprise-decision-runtime
compact-local-governed-model
Agent Token Entitlements
Every agent receives an explicit LLM access tier and per-case token budget. Deterministic agents should not consume model tokens.
AgentLLM AccessBudgetCallsTokensOver Budget
enterprise-decision-runtimemoderate4,00019580
Token Waste Detection
The control plane flags repeat prompts, oversized context, wrong-runtime calls, and high-latency model choices.
No token waste alerts detected for the selected window.
By Agent
Cost, volume, and token footprint for governed agent call paths.
AgentCallsInOutCost
enterprise-decision-runtime1812146$0.00
By Model
Attributed spend by approved model with average latency from the live audit stream.
ModelCallsInOutLatencyCost
compact-local-governed-model1812146184.0 ms$0.00