AI SPEND

Every AI dollar, accounted for.

EngLedger meters every AI request across providers, ties the spend to the work it produced and puts it next to your engineering cost: attributed, trended, defensible.

This period · $8,900 in AI, on what exactly
$7,650
86% of AI spend · attributed
attributed
740 sessions tickets 8 initiatives: spend tied to the work it produced.
$1,250
unattributed · shown, not hidden
remainder
Not yet tied to delivery. The honest remainder, on its own line, never folded into the total.

“On what, exactly?” is the whole product. Spend reads as attribution, and the part that cannot be attributed is flagged in amber and shown on its own line, never hidden.

Attribution is where it starts, not where it ends. The same readout points at the next move:

61%of spend
GPT-4o · 34% of requests
next
One model is carrying 61% of spend on a third of the requests: review those sessions and estimate what a lighter default would save.
METERS SPEND ACROSS
AnthropicOpenAIGoogle GeminiLlamaMistralQwenDeepSeekLocal LLMs
Attribution, not adoption

Who used AI is the easy question.
What the spend bought is the real one.

🔌
Every request, every provider
Every request records cost, tokens, model, latency and the identity of the person behind it, whether it went to Anthropic, OpenAI or a model you host yourself. One spend view a single-vendor tool cannot show.
🔗
Spend tied to delivery
AI sessions attach to the tickets and initiatives they touched, so the headline reads as attribution: what share of spend produced shipped work. The unattributed remainder is shown, not folded into a total.
🛡️
Budgets and limits, enforced
Set spend budgets by org, team or person, with model and rate limits, and choose what happens at the cap: block, restrict to approved models, rate-limit or alert only. Break-glass overrides restore access the moment someone needs it.
🧭
Session intent
Sessions are classified by intent (feature work, bugfix, spike), giving you the productive mix of AI spend rather than only its size.
📒
One ledger with labour cost
AI tooling only means something next to the engineering cost it sits inside. EngLedger is the one place both live, under the same evidence rules and the same period controls.
How the metering works

One settings change, not an implementation project.

01
Connect your AI tools
One endpoint change in your AI clients and every request is metered on its way to the provider. No agent installs, no plugin per tool.
02
Every request is recorded
Cost, tokens, model, latency and the person behind the request, captured from the first token onward.
03
Sessions attach to work
The same identity graph that resolves Jira, GitHub, Bitbucket and Asana accounts resolves AI usage to one person, and sessions to the tickets they touched.
04
Spend lands in the ledger
AI cost arrives next to labour cost, classified and period-controlled like every other number finance relies on.

Metering works alongside your existing work-tool connections, and the attribution feeds the same ledger described on the Engineering Ledger page. The step-by-step flow is documented in the metering setup guide.

AI meteringMetering
AI
All providers
● 49,832 requests this period
Captured per request
Cost and tokens→ Ledger
Model and latency→ Reports
Person identity→ Attribution
The AI Tax

One ratio the board already
knows how to read.

The AI Tax is AI tooling spend as a percentage of total engineering cost, trended month over month and forecast forward. A provider invoice tells you what AI cost; the AI Tax tells you what it cost relative to the engineering it supports. That is the form the board question actually takes.

It is not a standalone vanity number. The same figure is embedded in the cost report, where labour and AI tooling appear as one cost picture per person and per project, built on the same ledger that carries your capitalisation numbers.

Cost pictureThis period
Labour
$110,600
AI tooling
$8,900
AI Tax
7.4%trended
// Readout
AI tooling is 7.4% of engineering
cost this period, trended monthly.
The spend reports

Where the money is going,
and what it bought.

📈
Spend & Forecast
Spend trend and forecast, the AI Tax, and spend against throughput. Where the money is going and where it will be by quarter end.
🎯
Work & Value
Ticket-level AI spend lookup, session intent, and spend by tag. What any given piece of work cost in AI, and what kind of work the spend favours.
Get started today

Know what your AI spend delivered. By this afternoon.

Connect your tools and read your first AI spend readout today.

14-day trial. No credit card. Setup in minutes, not weeks.