Book a demo

AI diagnostic

Your AI. What does it emit, cost, and deliver?

The AI Diagnostic measures the footprint, cost and value of each of your use cases. In 4 to 6 weeks, you know what to keep, optimize or stop.

Book a 30 min reviewRun my self-assessment

The problem

AI use cases keep piling up. Nobody has the full list.

Copilots, assistants, agents, in-house models: IT sees a cost, leadership expects value, and the CSR team has to quantify a footprint that dashboards don't break down.

No shared list


Use cases get added one by one, across multiple vendors, with no shared inventory.

Cost without value


IT sees the spend. Leadership expects ROI. No one connects the two.

Invisible footprint


The CSR team has to quantify a footprint that vendor reports don't break down.
Logo Covea
“We measured the impact of our IT as a whole, from the group down to our three brands. AI is now part of the equation — we are assessing its effect on our carbon trajectory.”
Dorra Do Vale - Sustainability Lead, Group IT (DSIN), Covéa · 21,000 employees · 3 brands
From IT to AI: How Covéa Measures Its Digital Footprint?
Client Story at Produrable 2026
Read more

Also steering with Verdikt: Saint-Gobain, Arkema, RATP

The method

Every use case gets a verdict.

The diagnostic starts by building a single inventory. For each use case, it assigns an owner, a footprint, a cost and an estimated value — with a confidence level shown: measured, estimated or declared.

Use caseEmissions / yearCost / yearBusiness metricVerdikt
AI voicebot conversations23,317 kgCO2e€4,663k10.1M callsOptimize
AI document analysis206 kgCO2e€3.5k44k documentsStop
AI chatbot conversations1,052 kgCO2e€210k1.4M conversationsKeep
AI email processing11 kgCO2e€1.4k1.6M emailsMonitor
Copilot chat3,180 kgCO2e€850k12k licensesOptimize

The process

4 to 6 weeks. A deliverable at every stage.

Week 1
Scoping
Inventory of use cases, identification of owners and available data.
An initial register: one line per use case, with owner and status.
Weeks 2–3
Data collection
Collection of cost and consumption data, footprint calculation for each use case.
Footprint, cost and value quantified — with a confidence level for each data point.
Weeks 3–4
Scoring
Scoring of each use case on footprint, cost, value and risk.
A verdict for each use case: Keep, Optimize, Monitor or Stop.
Week 5
Trade-offs
Scenarios and quantified trade-offs — 3 to 5 decisions with the expected gain.
3 to 5 trade-offs in euros and tCO₂e, ready to present.
Week 6
Readout
Presentation to the executive committee, with a 12-month roadmap.
A ready-to-use executive committee deck + 3 months of platform access.
Two or three people are enough: IT or FinOps for the data, CSR for the carbon trajectory, ideally finance for the value.

FAQ

What we're often asked.

We review your AI use cases, the data available and where to start. You leave with a clear picture of the scope. The verdict comes from the diagnostic, after measurement.

No. The diagnostic starts from what you have: Excel exports, cloud billing (AWS, Azure, GCP), Microsoft 365, ServiceNow, EasyVista. Anything missing is flagged, with a confidence level shown.

A FinOps tool tracks spend, and an ESG tool gives a macro view. The diagnostic goes down to the level of each use case, using an auditable method — Bilan Carbone®, GHG Protocol. It complements those tools; it doesn’t replace them.

You leave with your trade-offs and three months of platform access to reassess on your own. Arkema does it twice a year.

Yes. Your customers, banks and partners may keep asking for comparable data. The diagnostic produces structured data you can provide either way.

30 minutes to take stock of your AI use cases.

 

Describe your situation: we'll tell you what the diagnostic can measure and where to start.

Book my time slot