Impact & Public sector

By starting with what costs the most today rather than what demonstrates best. Three levers recur: sharing a component across entities instead of building it several times, preferring open source where licensing funds nothing useful, and delivering domain by domain so every stage can be stopped. We respond to public tenders and work under framework agreements. One candid point: a modernisation does not fund itself from savings in the first year, and claiming otherwise simply defers the difficulty.

Through automated checks built into the delivery chain, complemented by manual audits. That combination is not a precaution: a significant share of the criteria in the French accessibility framework cannot be verified automatically, because they concern meaning, journey coherence and the relevance of alternatives. A site that passes every automated check can still be unusable with a screen reader. The accessibility statement and the multi-year plan remain obligations of your organisation; we prepare them with you.

On European sovereign cloud, or on an offering qualified by the French cybersecurity agency where sensitivity requires it. AI use cases rely on open models operated in that same perimeter, which avoids sending social or health data to an external service. No data leaves the European Union without an explicit processing agreement, and the architecture is designed taking extraterritorial laws into account rather than ignoring them.

It depends on the state of the existing estate, and we would rather say so than publish a savings percentage that supposedly holds for everyone. What can be measured honestly is the maintenance cost of one domain before and after migration, the time to deliver a change of comparable size, and the number of components reused across entities. All three can be read from the first migrated domain onwards, which lets you stop or continue on figures rather than on a projection made at the outset.

By starting from your theory of change rather than from the data that is easy to extract. A beneficiary count is an activity indicator, not an impact one: what funders ask for today is what changed for those people, and over what period. We then automate collection of the chosen indicators and make them legible in dashboards open to your supervisory bodies. Existing frameworks, social return on investment or indicator catalogues, provide a common language with your funders rather than a substitute for that definition work.

Mutual insurers, charities, social economy groups, territorial civil service training bodies and foundations. Three projects are presented on this page: securing the information system of the French Mutuality Federation with an EBIOS Risk Manager analysis, building the information system of the ZUPDECO charity, and the cloud-native migration of Groupe UP. Other references exist but remain covered by confidentiality agreements.