Agent FinOps

Cost attribution across 42 discovered agents, read from platform telemetry. GenTrust reports spend — it never bills, caps or stops it.

Total spend (30d)

US$2,313

Read from platform telemetry

Period budget

US$2,500

Fleet budget, pro-rated to the period

Utilisation

93%

Alerts only — spend is never stopped

Headroom remaining

US$187

Left before the hard threshold

Agents with spend

8 without telemetry

Total tokens

420.6M

Input and output combined

Spend by platform

SAP (LeanIX) is inventory-only and reports no cost.

SAP (LeanIX): telemetry unavailable · 8 agents

Spend trend

Daily attributed cost over the selected period.

Token usage split

Input versus output tokens by platform.

Cost attribution

Showing 42 of 42 agents · highest cost first

Name	Platform	Owner	Model	Calls	Tokens	Cost	Trend
Customer Operations	Copilot Studio	Tom Whitfield	gpt-4o-mini	47,508	23,516,471	US$204.28
IT	Copilot Studio	Yuki Tanaka	gpt-4o-mini	45,193	12,427,985	US$194.33
Patient Services	Copilot Studio	Grace Mbeki	gpt-4o	38,240	8,336,358	US$175.90
Corporate Affairs	Microsoft Fabric	Clara Nowak	gpt-4o	45,660	34,062,448	US$155.24
Quality & Regulatory	Azure AI Foundry	Priya Raman	gpt-4o	33,111	16,323,507	US$135.75
Clinical Affairs	Azure AI Foundry	Helena Voss	gpt-4o	31,197	21,058,195	US$134.15
Corporate Affairs	Microsoft Fabric	Clara Nowak	gpt-4o-mini	30,338	31,157,566	US$106.18
Supply Chain	Microsoft Fabric	Aisha Bakker	gpt-4o-mini	43,336	35,362,176	US$104.01
Marketing	Copilot Studio	Sofia Marchetti	gpt-4o	37,100	10,388,000	US$103.88
Quality & Regulatory	Azure AI Foundry	Helena Voss	gpt-4o-mini	23,067	8,004,249	US$80.73
Commercial	Copilot Studio	No owner	gpt-35-turbo	19,874	14,945,624	US$79.50
People	Copilot Studio	Ben Ashworth	gpt-4o-mini	31,608	15,677,403	US$75.86
Corporate Affairs	Copilot Studio	Yuki Tanaka	gpt-4o-mini	21,782	5,772,222	US$71.88
Manufacturing	Microsoft Fabric	Rajesh Nair	gpt-4o	23,027	13,816,285	US$64.48
Medical Affairs	Azure AI Foundry	Helena Voss	gpt-4o	27,699	29,748,238	US$60.94
Quality & Regulatory	Azure AI Foundry	Priya Raman	gpt-4o	13,627	2,888,924	US$54.51
People	Copilot Studio	Ben Ashworth	gpt-4o-mini	13,947	11,534,169	US$48.81
Commercial	Copilot Studio	Sofia Marchetti	gpt-4o	13,743	4,480,237	US$46.73
Quality & Regulatory	Microsoft Fabric	Priya Raman	gpt-4o-mini	39,519	10,393,593	US$43.47
Quality & Regulatory	Azure AI Foundry	Helena Voss	gpt-4o	11,466	5,148,368	US$42.43
Manufacturing	Microsoft Fabric	Rajesh Nair	gpt-4o	37,794	14,475,067	US$41.57
Finance	Microsoft Fabric	Marcus Lindqvist	gpt-4o-mini	14,099	5,343,521	US$38.07
Commercial	Copilot Studio	Sofia Marchetti	gpt-4o-mini	35,982	33,895,045	US$35.98
Quality & Regulatory	Azure AI Foundry	Rajesh Nair	gpt-4o	16,214	6,388,316	US$35.67
Manufacturing	Azure AI Foundry	Rajesh Nair	gpt-4o-mini	10,262	4,566,576	US$33.86
Finance	Microsoft Fabric	Marcus Lindqvist	gpt-4o-mini	36,948	16,256,900	US$29.56
Supply Chain	Microsoft Fabric	Aisha Bakker	gpt-4o-mini	31,348	7,523,400	US$25.08
Clinical Affairs	Azure AI Foundry	Helena Voss	gpt-4o	8,136	5,858,228	US$22.78
Quality & Regulatory	Azure AI Foundry	Priya Raman	gpt-4o	4,512	4,484,549	US$18.95
Clinical Affairs	Azure AI Foundry	No owner	gpt-4o	5,266	1,279,693	US$16.33
Legal	Azure AI Foundry	Clara Nowak	gpt-4o	2,558	2,138,715	US$11.26
Supply Chain	Microsoft Fabric	Aisha Bakker	gpt-4o-mini	2,831	2,295,941	US$11.04
Quality & Regulatory	Azure AI Foundry	Priya Raman	gpt-4o	2,343	763,838	US$7.97
Commercial	Copilot Studio	Sofia Marchetti	gpt-4o-mini	539	336,840	US$1.78
Finance	SAP (LeanIX)	Marcus Lindqvist	SAP AI Core — joule-orchestration	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
Customer Operations	SAP (LeanIX)	Tom Whitfield	SAP AI Core — foundation-model-hub	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
Supply Chain	SAP (LeanIX)	Aisha Bakker	SAP AI Core — joule-orchestration	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
Supply Chain	SAP (LeanIX)	Rajesh Nair	SAP AI Core — foundation-model-hub	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
People	SAP (LeanIX)	Ben Ashworth	SAP AI Core — joule-orchestration	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
Procurement	SAP (LeanIX)	Marcus Lindqvist	SAP AI Core — foundation-model-hub	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
Procurement	SAP (LeanIX)	Marcus Lindqvist	SAP AI Core — joule-orchestration	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—
Supply Chain	SAP (LeanIX)	No owner	SAP AI Core — foundation-model-hub	Telemetry unavailable	Telemetry unavailable	Telemetry unavailable	—

Footnote: 8 SAP (LeanIX) agents are inventory-only — the platform exposes no usage telemetry, so their cost and tokens are shown as unavailable and excluded from every total above. They are not zero-spend.

Budgets and alerts

Budgets

Attribution thresholds only — GenTrust informs, it never caps or stops spend.

Simulated

Scope

Applies to

Amount (USD)

Cadence

Soft %

Hard %

Soft thresholds alert the owner; hard thresholds alert and escalate to governance. Neither stops spend.

Fleet · Whole fleet

US$2,313 of US$2,500 · monthly · soft 80% / hard 100% · 42 agents · 8 without telemetry (excluded)

Soft breached · 93%

Platform · Azure AI Foundry

US$655 of US$600 · monthly · soft 75% / hard 95% · 13 agents

Hard breached · 109%

Platform · Copilot Studio

US$242 of US$400 · weekly · soft 70% / hard 90% · 11 agents

Within · 61%

Alerts

Threshold crossings and spend anomalies. Every action here is simulated.

Simulated

Soft breached

Fleet · Whole fleet reached 93% of its monthly budget. Owners are informed; spend continues.

Hard breached

Platform · Azure AI Foundry reached 109% of its monthly budget. Owners are informed; spend continues.

Per-agent spend spike

Field Service Knowledge Bot spent 4.1× its 7-day rolling average in the last 24 hours.

Tom Whitfield · detected 17 Sept 2026, 09:15

Hard threshold crossed

IT Access Request Triager passed 100% of its monthly attribution budget. Spend continues — nothing is blocked.

Yuki Tanaka · detected 16 Sept 2026, 08:15

Token burn anomaly

Output tokens for Patient Support Chat Concierge rose 2.6× with flat invocation volume — possible prompt regression.

Grace Mbeki · detected 15 Sept 2026, 07:15