## Notes from 28 May 2025 [[2025-05-27|← Previous note]] ┃ [[2025-05-29|Next note →]] I recently came across two insightful posts by [Arnaud Denoix](https://www.linkedin.com/in/arnaud-denoix/), the director of the [[Public Sector Reform in France#Public Interest Groups (a french quango model|Groupement d'Intérêt Public)]] "[Plateforme de l'Inclusion](https://inclusion.gouv.fr/)". They addressed two themes that are rarely discussed with such clarity in public administration: stopping public products and designing agile public institutions. The **[first post](https://www.linkedin.com/posts/arnaud-denoix_arr%C3%AAter-un-produit-cest-compliqu%C3%A9-surtout-activity-7333009151331082241-vKSI/?utm_source=share&utm_medium=member_ios&rcm=ACoAABJfvIUBLMnEUSA-EbpMEBKTnb3K9do__SQ)** was about the decision to sunset _Carnet de Bord_, a digital tool developed for tracking the progress of job seekers. The tool had been in use for two years, but with low user retention, limited measurable value, and a design that, despite good intentions, ended up increasing data entry work for frontline professionals. The honesty of the reflection was striking. He described how difficult it is—in the public sector especially—to stop a product, even when the impact is clearly insufficient. There’s always a lingering hope that a minor feature tweak will turn things around. But more than that, there’s institutional inertia, unresolved policy problems, and well-meaning sponsors who want to “de-complexify” service delivery. Reading someone from inside government say, essentially, “this didn’t work, and here’s what we learned,” felt rare and important. It was also encouraging to see that the decision to end the tool wasn’t seen as a failure, but rather as a normal part of an iterative public service strategy. The **[second post](https://www.linkedin.com/posts/arnaud-denoix_la-suppression-des-op%C3%A9rateurs-detat-pour-activity-7330842325893410816-WvXt/)** led me to a deeper discovery: the legal structure of the organization itself. I hadn’t paid attention before, but the _[Plateforme de l'Inclusion](https://inclusion.gouv.fr/)_ operates as a **GIP - [[Public Sector Reform in France#Public Interest Groups (a french quango model|Groupement d'Intérêt Public)]]**, a specific category of public law entity in France. These groupings bring together public and private actors under a shared mission of general interest. They are not agencies or ministries, but legally autonomous, purpose-driven structures. What makes it even more compelling is that many of these GIPs originate as spin-offs from France's public entrepreneur-in-residence programme, the Entrepreneurs d'intérêt général, which is run by [beta.gouv.fr](http://beta.gouv.fr/), the French government's digital services incubator. Unlike traditional bodies, GIPs can hire more flexibly - Arnaud points out that they were able to contract around 30 freelance digital professionals under fixed-term public contracts, avoiding the rigidity of the civil service hiring process. This particular GIP supports nearly 120,000 social and employment caseworkers with digital tools, and its team of 45 employees and 45 freelancers builds on the design practices of _[beta.gouv.fr](https://beta.gouv.fr/)_—the French State’s public digital service incubator. Projects often emerge from user needs and iterate quickly. Some succeed, others don’t. But the structure allows for movement. Personally, I think this GIP format is a compelling institutional innovation. I would argue that all GIPs should be required to include _sunset clauses_ (time-bound limits or mandatory reviews) to avoid quiet institutionalization. But even as it stands, it’s a remarkable mechanism: limited in scope, designed around a public mission, and open enough to recruit talent and iterate on services without starting from scratch every time. Fascinating to see this level of design thinking embedded in law and practice.