Finance overview
Finance (/finance, under Spend) is the overview for the person who answers for the AI bill. Every figure on it is the figure of a page beneath it, and every tile opens that page. The total is the same AI spend that Home, Costs and Executive show for the same period.
What AI spend counts
Section titled “What AI spend counts”AI spend is money billed or owed for AI in the period:
| Part | Basis | Where it comes from |
|---|---|---|
| Seats | Contract price | Each subscription’s seats (the larger of seats bought and seats held) × the monthly price, prorated to the period. |
| Extra usage and charges | Receipt | Charges recorded under Subscriptions, counted on the day they were charged. |
| API billed by vendors | Vendor invoice | What each connected billing connector (OpenAI, Anthropic, Azure OpenAI, Bedrock) charged for usage. |
| Cloud AI services | Vendor invoice | AI services on the AWS, Azure or Google Cloud bill, less any service an AI vendor connector already bills. |
| Usage priced for unbilled vendors | Estimate | Usage on a vendor with no connected bill: tokens the collector saw and the cost instrumented applications recorded, at Shield360’s prices. |
A vendor whose bill is connected is counted from that bill, and its usage is not counted again. Connecting OpenAI’s bill does not hide Anthropic usage: Anthropic stays an estimate until its bill is connected too.
What plan usage would cost at API rates is shown as Plan value. It is context for the seats, never spend.
The tiles
Section titled “The tiles”| Tile | Shows | Opens |
|---|---|---|
| AI spend | The total and its change on the previous period of the same length | Costs |
| Backed by bills | The share of the total from invoices, receipts and contract prices; the rest is estimated | Actual vs estimate |
| Budgets | How many enabled budgets are over or forecast to go over | Budgets |
| Paid for, unused | Seats nobody holds in the period, and seats with no use in 30 days | Subscriptions |
| Savings found | What fixing Cost of code’s findings could save a month (an estimate) | Cost of code |
| Plan value | Plan usage at API rates against the seats’ cost | Costs |
Below them: what the spend is made of, spend by vendor, spend by department (with what no department carries as its own row, so the rows add up to the total), and what was left out and why.
How is this calculated?
Section titled “How is this calculated?”How is this calculated? on a tile opens its working:
- the sum in words, for example Seats $600.00 + API billed by vendors $100.00 = $700.00;
- each part’s line items, with how each amount was arrived at and its source;
- what was left out, with the amount and the reason;
- the period and time zone, the latest bill line, models with no price, and subscriptions in other currencies;
- whether the parts, vendors and departments add up to the total. If one does not, the page says so: that is a bug to report.
The AI spend tiles on Home and Costs link here for the same working.
CFO pack
Section titled “CFO pack”CFO pack in the header downloads the period as a PDF, or as a CSV with every line item. Both carry the same working as the page, so the figures can be rebuilt from the pack alone. Each download is recorded in the audit log.
POST /api/finance/ledger {"timeLimit": {"start": "…", "end": "…"}, "timezone": "Europe/London"}GET /api/finance/export?format=pdf|csv|html&start=…&end=…&tz=…Both need cost read access and read only the caller’s own environment.
Known limits
Section titled “Known limits”- Vendor and cloud bills arrive by the day. A period that starts or ends mid-day counts those days’ bills in proportion to the hours covered.
- Amounts in currencies other than USD are added as entered, not converted. The working says when this happens.
- Spend alerts look back hours, so they use seats and priced usage rather than this total.