The product

Financial control for every stage of AI investment.

Understand spend, establish ownership, anticipate where spend is heading, govern material decisions and report the AI position from one financial model.

What Finance gets out of it

Four outcomes, one financial model.

See

One financial position across every AI provider, cloud platform and coding assistant — with a named owner behind each euro.

Predict

A forecast per initiative with visible assumptions and a confidence interval, set against the budget that was actually approved.

Control

Guardrails, thresholds and approvals bring Finance into the few decisions that matter — and stay quiet on the rest.

Prove

A documented value case and value KPI next to every euro — so the next increment of AI investment is a decision Finance can back.

How it is used

From portfolio, to initiative, to the one decision that needs you.

The same model, at three levels of zoom. Most weeks you stop at the first.

Portfolio 01
forecast FY26 €2.14M
approved guardrail €2.35M

Four initiatives, one owner each. Headroom is intact.

Portfolio within guardrails.
Initiative 02
Developer Copilot Review
forecast €315k · +18% vs plan
owner CTO

Seat rollout from 180 to 265 accounts for €31k of the move. Assumptions visible, open to challenge.

traced to source billing
Decision 03

Raise the guardrail to €370k for the remainder of FY26, on the back of a measured value case.

value evidence Outcome measured
approver CFO

One decision, taken deliberately, six weeks before it would have surfaced as a variance.

Approved and audit-logged.

Illustrative product example — not customer results

Understand AI spend Anticipate spend Ownership & guardrails Control decisions Prove value Communicate position
Understand AI spend · Establish ownership

Every euro of AI spend, with a name against it.

Slice AI and cloud spend by initiative, business unit, model or team, and map it onto a real org hierarchy so every cost has an accountable owner.

  • Interactive analysis — no SQL; drill from an initiative to the exact resource or model
  • Multi-level org hierarchy, 5+ levels deep, with a named owner at each node
  • Unowned spend always visible — financial ownership tracked as a number
AI spend by owner Illustrative scenario — example data
Acme Corp · AI portfolio€2.14M
Customer Support AI · VP Support€420k
Developer Copilot · CTO€315k
Sales Assistant · CRO€182k
Document Intelligence · COO€96k
AI platform & infra · VP Platform€1.13M
Not yet owned€86k · 4%
96% of AI spend has a named owner · unowned spend stays visible until it is resolved · ✓ audit-logged
Forecast vs approved budget Illustrative scenario — example data
Customer Support AI €420kWithin guardrail
Developer Copilot €315kTrending over
Sales Assistant €182kWithin guardrail
AI platform & infra €1.13MWithin guardrail
Forecast outside approved plan +€190k

Driven almost entirely by one initiative — flagged 6 weeks before it would have shown up as a variance.

every forecast carries visible assumptions · confidence interval shown per initiative
Anticipate where spend is heading

See where spend is heading while you can still shape it.

Forecast each AI initiative against the approved plan, with the assumptions Finance needs to challenge the number.

  • Forecast vs. actual vs. approved budget, side by side
  • Confidence intervals, not just point estimates
  • Every assumption behind a forecast is visible and challengeable
See how this works +

Established forecasting methods run in parallel — classical time-series, seasonality-aware, exponential smoothing and a deep-learning multi-horizon model — and the best fit per initiative is scored against your own history. No proprietary black box. You can pin a model per business unit if your team has a preference, and method, historical error and assumptions stay visible on every number.

no data-science team required · method visible on every number · ✓ audit-logged

Set financial guardrails

Set the financial guardrails that let teams keep moving.

A guardrail is the room an initiative has to operate in. Set the amount, scope it to an initiative or a business unit, name a finance owner and a business owner, and teams can spend inside it without asking. Finance only hears about it when the position moves toward the edge.

  • Headroom visible at all times, per initiative and per business unit
  • A finance owner and a business owner on every one
  • Hard or soft, in any currency, with a full draft → active → archived lifecycle
See how this works +

Guardrails can be hard or soft, in any currency, with the total distributed across periods however your planning cycle works. Every guardrail runs a real lifecycle — draft → active → archived — and every transition is audited, so the boundary a team was operating under on any given date can still be reconstructed later.

Guardrail · Customer Support AI FY26
On plan
€440,000 / annual
€282k committed · 64%€158k headroom
Period type
Quarterly
Forecast risk
Low
Finance owner
finance.lead@
Business owner
VP Support
€158k of room left this period · ✓ no action required
Decisions · Finance
1 waiting
Needs your approval

Scale Support AI to three more regions

Raises the annual guardrail from €440k to €610k. Value case validated last quarter.

Approve Ask a question

Everything else is inside its guardrail.

4 initiatives · 1 decision this month · ✓ audit-logged
Control material decisions

Bring Finance into the few decisions that matter.

Route material AI spend decisions through explicit policies and delegated approvers, and write every state transition to an append-only trail. Scaling an initiative, raising a guardrail or onboarding a new model becomes a recorded decision with a name, a date and a rationale attached — and everything below the threshold stays quiet.

  • Rule-based decision policies + time-bounded delegations
  • One decisions inbox per approver — Finance sees only the exceptions that need a decision
  • Append-only audit trail across guardrails, approvals and ownership
See how this works +
Decision policies
Rule-based — "requires approval above €50k", "requires review if forecast exceeds guardrail". Health monitored per policy.
Delegations
Time-bounded delegation of approval authority to colleagues. Auto-expires.
Decisions inbox
One inbox per approver, holding only what needs a call. Pending → approved / rejected / cancelled.
Append-only audit trail
Every material state transition is recorded — guardrails, approvals, users, ownership.
Prove value

Put the value case next to the investment.

Each initiative carries an expected outcome, the KPI that would show it, an owner, and an honest evidence status — so an expectation is never read as a result.

  • An expected outcome and a value KPI on every initiative
  • Evidence status shown, so an expectation is never read as a result
  • Value case and spend reported side by side, from the same model
Value cases
Illustrative scenario — example data
InitiativeValue caseValue KPIEvidence
Support AIReduce cost per contact 18%Cost per resolved ticketMeasuring
Developer CopilotReduce cycle time 12%Median PR lead timeInstrumented
Sales Assistant+4% conversionQualified-to-closeAwaiting baseline
value case ≠ realised return · evidence status is stated, never inferred
How evidence is graded

Evidence gets stronger as an initiative earns it.

Every initiative shows where it stands on the ladder, right next to the spend. Reaching the top rung is what turns scaling into a decision Finance can back.

  1. Step 1
    Case documented

    The expected outcome is written down and owned.

  2. Step 2
    KPI instrumented

    A measurable value KPI is defined and being collected.

  3. Step 3
    Outcome measured

    Enough history exists to compare against the baseline.

  4. Step 4
    Value validated

    The business owner and Finance agree the outcome holds. This is where scaling gets funded.

Initiatives that have not reached step 1 are shown as Awaiting baseline — visible, not hidden.

Communicate financial position

Turn the live AI position into board-ready reporting.

The Report Builder is a drag-and-drop editor with 15+ building blocks — position cards, forecast vs guardrail, initiative tables, ownership coverage and insight panels. Export to PDF, set a schedule, and the AI financial position lands in the board's inbox automatically.

  • 15+ building blocks, configurable per report
  • Pixel-perfect PDF export, stored and shareable
  • From board number to source billing in one trace
AI position
Forecast vs guardrail
Ownership
Spend trajectory
What changed
Initiative table
Value case vs spend
+ Add block
Explain what changed

Answer "why did this move?" in plain English.

The built-in assistant has live access to your AI and cloud inventory. Ask why an initiative's forecast moved and get a real answer tied back to the usage that caused it — the explanation you need before you take it into a budget conversation.

You: Why is the Developer Copilot forecast 18% above plan?
Assistant: The forecast rose by €48,400 against plan, driven by three changes:
  1. Seat count grew from 180 to 265 after the platform team rollout on 4 May — €31k.
  2. Average tokens per session up 24% since the agent mode default changed — €12k.
  3. Evaluation runs against a larger model added €5.4k in the last two weeks.
Sources: 3 providers · cost model v17 · change log · owner: CTO
Connect the AI portfolio

The AI bill isn't one invoice anymore.

Model providers, coding assistants, the infrastructure underneath, and the finance systems it all has to land in. Read-only ingestion, tenant-isolated data and traceability back to source billing.

AI spend — the fastest-moving part of the bill
OpenAI
Model & token spend unified into €
Anthropic
Model & token spend unified into €
Google Gemini
Model & token spend unified into €
Google Vertex AI
Model & inference cost ingestion
Amazon Bedrock
Model & inference cost ingestion
Azure OpenAI
Model & inference cost ingestion
Cursor
AI coding assistant spend
Claude Code
AI coding agent spend
GitHub Copilot
Seat & usage spend
ChatGPT
AI assistant seat spend
Microsoft Copilot
AI assistant seat spend
Infrastructure — including what AI runs on
Google Cloud
Cost, inventory & utilization via BigQuery
AWS
Multi-cloud cost ingestion
Microsoft Azure
Multi-cloud cost ingestion
Finance — where the numbers have to land
NetSuite
Ledger & cost-centre alignment
Exact Online
Ledger & cost-centre alignment
DATEV
Ledger & cost-centre alignment
Read-only and multi-tenant by design. We never need write access to your cloud or AI accounts. Each customer is an isolated tenant, and every figure traces back to source billing. Coverage varies by plan and by provider API availability. The full architecture — data stores, encryption and tenant identity — is documented on Security & Trust.

See where your AI financial control actually stands.

Nine questions on spend visibility, ownership, forecasting, guardrails and value. No cloud access, no sales call. Or book a 25-minute walkthrough on your own billing exports.