The directory brings together data publishers, APIs and AI guidance. Select sources by definition, coverage and reuse conditions, then test their usefulness.
Choose by need, then compare evidence
Find territorial or sector datasets for analysis.
Integrate an API with access conditions and error recovery.
Prepare an AI use with evaluation, human review and provenance.
What to compare for each kind of work
Need
Questions to examine
Downloadable data
Licence, period, fields, coverage and update frequency.
API
Authentication, quotas, schema, pagination and service terms.
AI system
Use case, evaluation set, errors, human oversight and total cost.
Results and cases without an answer
Prepare cases with expected results, incomplete data and out-of-scope requests. Compare errors, corrections and human handover using the same inputs. A successful supplier-selected demonstration does not by itself measure performance on your cases.
Deliverables to request
Source record: publisher, definitions, coverage, licence and date.
Checks for completeness, consistency, duplicates and freshness.
Repeatable evaluation with difficult cases and an abstention procedure.
A catalogue guarantees neither access to every database nor rights for every use. Health data, personal data and quota-controlled feeds need specific checks with the publisher.
Is a public source always freely reusable?
Check the specific dataset’s licence, terms and restrictions. A catalogue entry and accessible documentation do not mean that all records are open or suitable for your purpose.
FRENCH DIGITAL SOCIETY only publishes identifiable organisations and official resources. Inclusion is not a certification or ranking; check references, availability, legal information and contractual terms before appointing a provider.
1 resources shown
Keep a resource shortlist
Select resources in the list to compare their role and scope. Compare your shortlist on screen with your team. Selection stays on this page, with no account or browser storage; filtering does not clear it.
Inclusion does not establish quality or availability. Check identity, scope, source and relevant evidence before using a resource.
Institutional resource · France
Acceslibre
Collaborative database and API describing the accessibility of public-facing establishments for different disability needs.
Use it for: Find a venue’s accessibility information, then confirm specific requirements.
Public service for consulting public legal information about French businesses, associations and public bodies, useful for identifying an organisation before a professional conversation.
Use it for: Find an organisation’s identifiers and public information; identifying it does not establish the quality of its services.
Record documenting the API mechanism for submitting essential public-procurement data and its publication destinations. Distinguish submission needs from consulting already published data.
Use it for: Assess essential procurement data with its schema and coverage limitations.
Open platform for French public data. Examine each dataset’s producer, files and licence; the portal does not provide a uniform freshness guarantee for every dataset.
Use it for: Before reusing a dataset, check its publisher, licence, update date and geographic scope.
API covering social and professional inclusion services. The official record indicates restricted access: verify authorisation, schemas and reuse conditions, particularly for contact details available through the API.
Use it for: Prepare an authorised access request for inclusion services, with their audience, area and reuse conditions.
The official record states that the API is currently no longer exposed on pole-emploi.io while distribution and reuse arrangements are reviewed. Read the notice before planning integration; this documentation record does not establish operational access.
Use it for: Read the official interruption notice before planning integration or a fallback.
Access to Géorisques maps, datasets and API. When consulted on 10 October 2026, the publisher displayed a technical incident and geographic access restrictions affecting several features; check the notice before use.
Use it for: Find risks associated with an area and read the data limitations.
A directory entry is a starting point. Match the resource to a defined need, check the original source and compare approaches before committing a budget.
1. Describe the outcome
Write down the audience, the problem, the expected result and the deadline. A website redesign, a security review and a data project call for different evidence and skills.
2. Check the source
Open the organisation’s own site. Check its legal identity, the scope of its services, the date of its guidance and how to contact it. For public data, review the publisher, licence and update frequency.
3. Compare like with like
Ask providers about relevant work, delivery method, responsibilities, accessibility, handover and support. Compare proposals against the same brief and note what is excluded.
A public resource may help you learn or benchmark; a provider may carry out a mission. Inclusion here does not imply endorsement, affiliation or current availability. Read the expert selection guide for a fuller checklist.
Browse with context
What each directory category can help you do.
The filters group different types of resources. These notes explain the intended use and the checks to make before acting.
Choose this category when the need concerns a website, application, integration or long-term technical maintenance. Distinguish a delivery provider from a public guide explaining how to prepare the project.
Before you decide: Ask about ownership of the domain and code, accessibility checks, security updates, documentation and the support plan.
Use this category for brand messages, visual identity, editorial production and channels. A communication plan should name its audience and the action it is meant to support.
Before you decide: Compare examples of published work, approval steps, usage rights and how consistency will be maintained across formats.
Explore this category for infrastructure, connectivity, cybersecurity and operational support. Start with an inventory of essential services and who can access them.
Before you decide: Confirm response arrangements, backups, recovery responsibilities, subcontractors and the boundary between support and security advice.
Find public datasets, APIs and guidance for analysis or a controlled AI experiment. A catalogue entry describes a source; it does not automatically make the data suitable for a decision.
Before you decide: Check licence, update frequency, coverage, missing values, access conditions and whether a human can review consequential outputs.
Look here for search visibility, campaigns, email and measurement guidance. Begin with a target audience and a meaningful outcome before selecting a channel.
Before you decide: Ask which baseline will be used, how consent and tracking work, and who will maintain content after a campaign.
Use networks to discover organisations, training or support. An entry in a network is a lead for further checking, not proof of availability or quality.
Before you decide: Verify membership criteria where relevant, the person who will carry out the work and the terms of any proposed introduction.
The date records an automated availability check, not a guarantee of data freshness, service quality or eligibility. Some publishers block automated requests. For each dataset or API, check the actual licence, update date, coverage and access terms before reuse.
Select a suitable digital resource
Define the expected outcome and constraints before choosing a service or provider.
Scroll horizontally to read all columns.
Decision guide
Criterion
What to check
Practical implication
Need
Specify users, task, data processed and expected result.
A well-known tool can still be unsuitable for your team or data.
Evidence
Request a representative demonstration with the access rights future users will have.
An administrator demonstration does not establish usability for everyone.
Acceptance criteria
Describe a complete task, its inputs, expected outcome and the person who will validate it.
A delivered screen does not prove the journey works; agree on error handling and support requests as well.
Data scope
List essential fields and connect each to a service purpose.
Broader collection complicates management without necessarily improving outcomes.
Data minimisation
For a defined purpose, collection of personal data should be limited to what is adequate, relevant and necessary. A convenient feature alone does not justify collecting fields with no established use. Source: CNIL — Minimisation
Worked example
Fictional case: a charity needs a booking tool. It tests a scenario with one user, a cancellation and an export, then checks who can see the data and retrieve the calendar. This exposes terms to settle before wider use.
Where should I start when several digital tools look suitable?
Describe a real task, required data, who will access them and the expected result. A limited trial against that scenario reveals more than a context-free feature list.