AI FinOps.

The economics of AI: what it creates and what it truly costs, carried to earnings, to the balance sheet and to the valuation. Not the cloud bill, which answers to the CIO — the questions a CFO, an operating partner and a shareholder actually ask.

PRACTICE 02 · ACCOUNTED FOR · CFOS AND INVESTORS

Download the practice brief
What AI FinOps means

Nobody validates a case written by the party that sold the work.

Finance directors no longer accept business cases written by the firms delivering the projects. Odyssey sells no execution, so our figures carry no interest of ours. Where we framed the strategy, we audit only the objectives agreed upfront with finance, against a frozen baseline. Behind that independence: more than forty transactions, on both sides of the table.

FIGURES THAT SURVIVE DUE DILIGENCE

Nobody validates a case written by the party that sold the work.

Finance directors no longer accept business cases written by the firms delivering the projects. Odyssey sells no execution, so our figures carry no interest of ours. Where we framed the strategy, we audit only the objectives agreed upfront with finance, against a frozen baseline. Behind that independence: more than forty transactions, on both sides of the table.

FIGURES THAT SURVIVE DUE DILIGENCE

The deliverables

What AI FinOps produces, exhibit by exhibit.

AI FinOps · TCO-AI

Infrastructure is only the entry ticket.

Technology is only about 15% of the cost. Change management and shadow costs decide the true cost of AI — and appear on no invoice.

Expected outcomeCloud FinOps steers the ~15% on the invoice; Odyssey AI FinOps steers the full 100%, carried to the P&L and the valuation.
FormatIn AI Business Case and AI ROI Audit
Exhibit · AI FinOpsIllustrative · signed by a partner
TCO-AI, the full costGenerative AI project · orders of magnitude
Technology
15%
Change management
25%
Integration & ops
20%
Data
15%
Governance
10%
Shadow costs
15%
What the invoice shows · tech ≈ 15%
Inference, tool calls and API tokens
Volume × price × context, across agent runs
LLM licences and subscriptions
Paid seats × actual usage rate
Hosting and cloud
Compute, vector database, storage
Hardware
On-premise GPUs; rare in mid-market firms
What the invoice doesn’t show · the other 85%
Data
Preparation, quality, labelling
Integration & operations
Connectors, orchestration, maintenance, human review
Change management
Training, adoption, process rework
Governance
EU AI Act, evaluation, oversight
Shadow costs
Dormant licences, shadow AI, unused seats
Steering AI on its invoice means steering about 15% of its cost — the CFO’s question is the full 100%.

SELECTED CASE · AI PRICING STRATEGY

Weda — AI pricing in healthcare software →

Why AI FinOpsThe gain exists. No one measures it through to EBIT.
80%see productivity improve
37%see it in the income statement
6%turn it into a material earnings lever
Source: McKinsey, The state of AI in 2026: On the road to ROI, August 2026 — survey of 1,719 respondents in 97 countries.Measure what your AI actually returns

IN SCOPE

What the practice covers.

What the practice covers.

Four questions, in a fixed order. What the investment creates. What price captures it. What it costs, fully loaded. And how the resulting case holds when an investor, an acquirer or an audit committee takes it apart.

Business case and value realised

What the case promised, what the investment has actually returned, and the gap between them.

Pricing of AI-enabled products

Capturing the value created, under gross margins the market is resetting.

Due diligence and value creation plans

For funds: the AI thesis of a holding, tested before the investment and steered after it.

IN SCOPE

What the practice covers.

Four questions, in a fixed order. What the investment creates. What price captures it. What it costs, fully loaded. And how the resulting case holds when an investor, an acquirer or an audit committee takes it apart.

Business case and value realised

What the case promised, what the investment has actually returned, and the gap between them.

Pricing of AI-enabled products

Capturing the value created, under gross margins the market is resetting.

Due diligence and value creation plans

For funds: the AI thesis of a holding, tested before the investment and steered after it.

OUT OF SCOPE

What we do not do.

We do not tune infrastructure, we do not audit code, and we do not sell a dashboard. Where a mandate requires technical examination, the code and data audit is carried out by specialist partners; the financial stakes of data stay within our ROI Audit, and the fund keeps a single interlocutor and a single report.

No infrastructure optimisation

Cloud and inference spend answer to the CIO. A different question, a different discipline.

No code or data audit

The technical layer is for specialist partners. The financial stakes of data sit within our ROI Audit.

Portfolio decisions belong to AI StratOps

What to commit to, and in what order, is settled in the other practice.

HOW TO ENTER

Three levels within the practice.

Some Signatures are bought directly online, at a fixed fee; others are scoped with you in a written proposal. Corporates usually enter through AI Business Case or AI ROI Audit; funds through AI Strategy Due Diligence, most often by introduction.

AI Business Case
AI ROI Audit
AI Strategy Due Diligence
All Signatures
How AI ROI is measured
Entry diagnostic — AI Strategy Readiness

€990, within three business days. Bought online.

Signatures

AI Business Case · AI Pricing Strategy · AI ROI Audit · AI Strategy Due Diligence. Fixed scope, published fee.

Bespoke mission

Measurement installed and method transferred to your finance teams — scoped and priced after a call with a partner.

ENGAGEMENT MODEL

How we engage.

Engagements begin with a focused 30-minute conversation to test fit. From there we propose a scoped engagement with a firm timeline, named deliverables and a partner present throughout: the same partner from the first day to the last.

DURATION

2 weeks to 100 days

From business case to 100-day plan

DELIVERY

Signed by a partner

The same partner, first day to last

LANGUAGES

FR · EN

Native delivery

DELIVERABLE

A defensible case

Evidence a board can act on

INDUSTRIES SERVED

Where we operate.

We focus on sectors where AI is repricing competitive advantage. Each industry brings its own economics, data dynamics and strategic battlegrounds, and our work is grounded in those specifics rather than in generic frameworks.

Enterprise software

Repricing value under LLM commoditization.

Healthcare software

AI as the next pricing battleground.

Private equity · Venture

Pre-LOI strategic diligence on AI-native targets.

Digital services

Rebuilding the economics of the billable day.

Read the research note

AI FINOPS PRACTICE BRIEF · PDF

Download the practice brief

Engage

The right AI decision, made earlier.

Thirty minutes with a partner, to understand your situation and tell you plainly whether Odyssey is the right firm to help. We take on a limited number of mandates each quarter.

Speak with a partner

Read our thesis

Typical engagement: 4–12 weeks

Independent of any vendor

Bilingual FR / EN

Engage

The right AI decision, made earlier.

Thirty minutes with a partner, to understand your situation and tell you plainly whether Odyssey is the right firm to help. We take on a limited number of mandates each quarter.

Speak with a partner

Read our thesis

Typical engagement: 4–12 weeks

Independent of any vendor

Bilingual FR / EN

Engage

The right AI decision, made earlier.

Thirty minutes with a partner, to understand your situation and tell you plainly whether Odyssey is the right firm to help. We take on a limited number of mandates each quarter.

Speak with a partner

Read our thesis

Typical engagement: 4–12 weeks

Independent of any vendor

Bilingual FR / EN