From IT to AI
Client story at Produrable 2026
From IT to AI: How Covéa measured its sustainable digital journey
Covéa, France’s leading mutual insurance group, measured the carbon footprint of its information system in 16 weeks with Verdikt. It then extended the approach to artificial intelligence in just 6 weeks. A client story shared by Dorra Do Vale (COVEA – IT Sustainability Lead) and Hélène Levrault (VERDIKT – CPTO & co-founder) for Produrable 2026.
Key takeaway:
- IT footprint: a 16-week project covering workplace, infrastructure, cloud, SaaS and suppliers, co-sponsored by IT and CSR from day one.
- AI footprint: a 6-week extension covering 12 business use cases and Microsoft Copilot, with cost, emissions and business value assessed side by side.
- Result: measurement turned into decisions, such as a 12-ambition action plan, concrete equipment actions already in place, and a “keep, optimize or stop” verdict for each AI use case.
About Covéa: a multi-brand IT system
Covéa is the number one French mutual insurance group, with 21,000 employees and three brands: MAAF, MMA and GMF. Its information system is matrix-based: multi-site, multi-brand, and spread across the group IT department, brand IT teams and service providers.
The starting point: the IT system was a blind spot in an otherwise ambitious CSR commitment. Data was scattered, and the group had no way to measure the system’s environmental impact.
The project’s three goals:
- Measure the footprint of the IT system
- Identify priority areas for action
- Feed extra-financial reporting and the Sustainable Digital roadmap
Why Covéa chose Verdikt?
A multi-criteria CSR vision
A global reading of impact, not carbon alone.
Validated compliance
GHG Protocol, Bilan Carbone®, CSRD/ESRS-compliant, CyberVadis.
An exhaustive, granular approach
Devices, cloud services and applications, backed by 15,000 IT emission factors.
A solution to act and to steer
Continuous tracking over time rather than a one-off audit.
Step1 #IT footprint
IT: a 16-week project led by IT and CSR
| Phase | Timing | Content |
|---|---|---|
| 1Scoping | Weeks 1–3 | Scope, data, IT + CSR governance |
| 2Data collection | Weeks 4–9 | Device fleet, hosting, cloud, purchases |
| 3Calculation | Weeks 10–12 | Footprint by domain, review and validation |
| 4Analysis | Weeks 13–14 | Priority levers, presentation to the executive committee |
| 5CSR integration | Weeks 15–16 | Extra-financial reporting, annual re-measurement |
The scope was designed to assess the full impact of digital activities: workplace, infrastructure, cloud, suppliers, SaaS, network, sites and applications.
Step2 #AI footprint
AI: an extension that is both logical and urgent
| Phase | Timing | Content |
|---|---|---|
| 1Scoping | Week 1 | — |
| 2Data collection | Weeks 2–3 | Cloud, Microsoft and local infrastructure |
| 3Calculation | Week 4 | Per use case |
| 4Analysis | Week 5 | Priority levers and the Copilot trajectory |
| 5Presentation | Week 6 | To management |
The AI scope:
- Business AI services: historical AI (machine learning, voicebots), generative AI and LLMs, cloud hosting, internal data centers. This covers 12 business use cases.
- AI in collaboration tools: Microsoft Copilot, with two license types and usage patterns (Copilot Chat and Microsoft 365 Copilot).
The result is a unified view for decision-makers, without hiding the underlying complexity.
Result
Measurement turned into decisions
A 3-year 12-ambition action plan, based on concrete actions, quantified in terms of carbon reduction across the ESG pillars
Each domain has its own key indicator, dashboard and level of detail:
- Devices
- Data centers and cloud
- Purchases (SaaS and services)
Covéa built a vision and roadmap of 12 ambitions across the 3 pillars of CSR, each with 3–4 qualified, actionable levers.
Each action is quantified and prioritized to steer by impact. Actions combine two effects: a lower usage footprint and, above all, avoided emissions by not renewing equipment.
For each use case, the analysis crosses what it costs, what it emits and what it delivers, and gives an actionable verdict: keep, optimize, monitor or stop.
Each license carries a carbon cost, from chat to video-meeting transcripts. Initial usage was heterogeneous, moderate and constrained. The measurements give Covéa a comparison base to arbitrate future deployments by license and by usage.
- The most frugal technology by default
- The right infrastructure in the right place
- A framed rollout, not a generalized one
Sobriety is decided upfront, not corrected afterwards.
Key lessons for your organization
Compare to decide, before execution
Measure to act, during execution
Steer to progress, over the long term
Your AI and IT projects have their impact determined at the design stage.
FAQ
In the Covéa project, measuring the group’s IT footprint with Verdikt took 16 weeks, from scoping to CSR reporting integration.
Once the IT baseline is in place, extending the measurement to AI took Covéa 6 weeks. Verdikt offers an AI diagnostic in 4 to 6 weeks.
It covers business AI services (machine learning, voicebots, generative AI and LLMs), their hosting (cloud and internal data centers) and AI embedded in collaboration tools such as Microsoft Copilot.
By crossing three lines for each use case: cost, emissions and business value. This gives an actionable verdict: keep, optimize, monitor or stop.
Around 20–30% for digitalized companies, excluding downstream scope 3.
GHG Protocol, Bilan Carbone®, CSRD/ESRS and CyberVadis, with 15,000 IT emission factors.
AI diagnostic in 4 to 6 weeks
Mesure your AI footprint
The footprint, cost and value of each use case, followed by quantified arbitrations: to know what to keep, optimize or stop.
Book a 30 min reviewRead more