:root { --fin-bg:#f7f3ea; --fin-ink:#162013; --fin-card:#fffdf7; --fin-line:#d8cdb7; --fin-green:#1f6b45; --fin-gold:#a16409; --fin-red:#9f1239; } body { background:var(--fin-bg); color:var(--fin-ink); } .fin-shell { padding:1.2rem; max-width:1480px; margin:0 auto; } .fin-hero { border:1px solid var(--fin-line); border-radius:24px; padding:1.35rem; background:linear-gradient(135deg,#fff9ec,#edf6e6); box-shadow:0 18px 42px rgba(35,31,22,.08); } .fin-kpis, .fin-grid { display:grid; grid-template-columns:repeat(4,minmax(0,1fr)); gap:.85rem; margin-top:1rem; } .fin-grid.two { grid-template-columns:repeat(2,minmax(0,1fr)); } .fin-panel, .fin-card { border:1px solid var(--fin-line); border-radius:20px; background:var(--fin-card); padding:1rem; box-shadow:0 12px 30px rgba(35,31,22,.06); } .fin-panel { margin-top:1rem; } .fin-label { text-transform:uppercase; letter-spacing:.08em; font-size:.72rem; font-weight:900; color:#6b5d45; } .fin-value { font-size:2rem; font-weight:950; line-height:1.05; margin:.25rem 0; } .muted { color:#667056; font-size:.86rem; } .mono { font-family:ui-monospace,SFMono-Regular,Menlo,Monaco,Consolas,monospace; } .fin-warning { border:1px solid #d8a235; background:#fff6dd; color:#5c3b00; border-radius:16px; padding:.8rem; margin-top:.8rem; font-weight:750; } table { width:100%; border-collapse:collapse; font-size:.9rem; } th, td { text-align:left; border-bottom:1px solid #eadfca; padding:.68rem .55rem; vertical-align:top; } th { color:#594b35; font-size:.72rem; text-transform:uppercase; letter-spacing:.06em; } .proof { display:inline-flex; margin:.18rem .28rem .18rem 0; border:1px solid var(--fin-line); border-radius:999px; padding:.36rem .7rem; text-decoration:none; color:var(--fin-green); font-weight:850; background:#fff; } .fin-button { cursor:pointer; display:inline-flex; margin:.18rem .28rem .18rem 0; border:1px solid var(--fin-green); border-radius:999px; padding:.56rem .9rem; color:#fff; font-weight:900; background:var(--fin-green); } .note { color:#667056; font-size:.9rem; line-height:1.45; } @media (max-width: 980px) { .fin-kpis, .fin-grid, .fin-grid.two { grid-template-columns:1fr; } .fin-shell { padding:.8rem; } table { display:block; overflow-x:auto; } }
Sphere capability

FinOps Command Center

Tracks incremental compute cost, token consumption, model routing, SAP/S4 call instrumentation, cache reuse, budget governance, and ROI against business value. Spend comes from LLMAuditRecord; value comes from ValueEventRecord.

AI spend
$0.0008
Computed from token ledger
Business value
$1,722,248,180.00
Reported value events
Net benefit
$1,722,248,180.00
Value minus compute cost
ROI multiple
2055188758949.9x
Only shown when cost and value exist
Creates fresh FinanceEvent, AgentExecution, LLMAuditRecord, ValueEventRecord, evidence and replay rows.
Universal Decision Cost Runtime

Every agent execution receives a cost envelope

Envelope coverage
100.0%
5,000 / 5,000 executions
Cost / agent decision
$0.000000
Includes deterministic zero-LLM envelopes
Explicit cost counters
10.0%
500 instrumented executions
SOR counter coverage
0.0%
0 executions with SAP/SOR counters

Every AgentExecution receives a FinOps envelope; explicit cost/SOR counters remain the hardening target for production-grade cost claims.

Agent Cost Runtime

Cost per agent, mission, model and decision

AgentCallsTokensCostCost / decisionValueROILatencyModels
enterprise-decision-runtime
Unmapped
1958$0.0008$0.000838$0.00n/a184.0 mscompact-local-governed-model
Mission Cost Runtime

Rollup by finance mission

MissionCallsTokensCostCost / callAgents
Unmapped1958$0.0008$0.000838enterprise-decision-runtime
Token Intelligence Runtime

Top token consumers

Unmapped
958
enterprise-decision-runtime · 100.0% of tokens · avg 958.0

Waste alerts: 0. Audit question: Can BP identify top token consumers and unnecessary context loading?

Model Routing Intelligence

Deterministic-first economics

Recorded calls
1
LLM share
100.0%

Deterministic-first routing is required for policy thresholds, duplicate lookup, SLA checks, and SAP field validation.

Tier distribution: {'balanced': 1}

Context Cost Intelligence
82.0%
capture_enabled

LLMAuditRecord input/output token totals until agent evidence emits context_used counters.

SAP Call Intelligence
0
capture_enabled

No SOR gateway counters found in LLMAuditRecord.evidence_json for the selected window.

Cache Intelligence
0.0%
opportunity_only

Prompt/input hash repeat analysis; actual cache_hit counters not yet emitted.

Learning Optimization
0.0%

Positive trend means average tokens per decision decreased over the selected window.

Budget Governance

Alert, throttle, switch model, or require approval

MissionSpendBudgetUsedValueAction
Controls$0.0000$1,800.000.0%$0.00monitor
FP&A$0.0000$1,800.000.0%$0.00monitor
O2C$0.0000$3,000.000.0%$875,000.00monitor
P2P$0.0000$2,500.000.0%$856,000.00monitor
R2R$0.0000$2,800.000.0%$1,625,000.00monitor
Tax$0.0000$1,200.000.0%$0.00monitor
Treasury$0.0000$1,500.000.0%$1,875,000.00monitor
Unmapped$0.0008$750.000.0%$0.00monitor

Budgets are configured governance controls for the demo environment, not tenant-approved production limits.

FinOps Replay

Execution economics trace

ExecutionAgentModelTokens in / outCostValueSAP countersReplay
0f4f1d7b-960enterprise-decision-runtimecompact-local-governed-model812 / 146$0.000838$0.00capture_enabledOpen replay
SOR Writeback Proof

Confirmed document versus governed intent

Ledger attempts
8
Rows in governed writeback ledger
Dry-run / intent
1
Policy-gated, not production posting
Confirmed 2xx writes
7
Non-dry-run successful adapter calls
Confirmed docs
0
Successful writes with document reference

Confirmed SOR posting is claimed only for non-dry-run 2xx writebacks that return a document reference. Dry-run rows prove governed intent and policy-gated execution, not production posting.

Customer production SOR credentials and approved non-dry-run adapters are required before a production document reference can be represented as confirmed.

WritebackSOREntityActionModeHTTPProofCreated
5a5ac854-34bsap-s4hanainvoice
BP-SVC-884220
block_paymentlive attempt2002xx without document ref2026-07-13T19:39:06.118915
d4055aea-023sap-aribasupplier_task
VEND-8842
request_supplier_duplicate_confirmationlive attempt2022xx without document ref2026-06-29T08:18:13.810135
cf628e91-3aesap-ecc-apinvoice
INV-LIVE-DUP-1778594510
apply_payment_blocklive attempt2022xx without document ref2026-06-29T08:10:04.056990
f85e7ed6-7a2sap-s4-financejournal
R2R-WORKSHOP-JE-001
stage_correction_journallive attempt2022xx without document ref2026-06-29T08:10:04.056990
0e8f56e4-230servicenowfixed_asset_request
FA-WORKSHOP-CAP-001
create_asset_controller_review_tasklive attempt2022xx without document ref2026-06-29T08:10:04.056990
a3f18773-525microsoft-graphapproval_message
P2P-WORKSHOP-DUP-001
publish_supervisor_approval_cardlive attempt2022xx without document ref2026-06-29T08:10:04.056990
3d3d6ba0-9aftreasury-tmscash_position
TREASURY-WORKSHOP-001
stage_cash_position_commentarydry-run202dry-run intent2026-06-29T08:10:04.056990
385efcf3-d48sap-aribasupplier_task
BP-VEND-8842
request_supplier_duplicate_confirmationlive attempt2022xx without document ref2026-06-29T08:10:04.056990
CFO AI Economics Dashboard

AI spend: $0.0008

AI savings/value: $1,722,248,180.00

Cost per decision: $0.000838

Cost per collection: $0.000000

Cost per close: $0.000000

Data Basis

Cost source: LLMAuditRecord

Value source: ValueEventRecord rows linked to agent_id, assistant_id, or execution_id

SAP source: No SOR gateway counters found in LLMAuditRecord.evidence_json for the selected window.

Cache source: Prompt/input hash repeat analysis; actual cache_hit counters not yet emitted.

Budget source: BP workshop governance baseline

Live FinOps API SOR Writeback Proof API AI Cost & Usage Legacy Runtime FinOps