Last update: 04/08/2026
AI4DRPM
Digital-ready policymakingDigital governmentCollaboration
Artificial intelligence (AI) is becoming part of everyday life and public administrations are no exception. In this context, institutions across Europe are testing how it can support administrative, legal and policymaking tasks. This article explains what this means in practice: from analysing draft legislation and supporting impact assessments, helping policy officers draft documents to citizens navigating public services.
Key AI terms
-
Artificial intelligence (AI): refers to systems that display intelligent behaviour by analysing their environment and taking actions – with some degree of autonomy – to achieve specific goals.
-
Machine learning (ML): AI that enables computers to learn patterns and make predictions or decisions without needing every rule to be manually programmed.
-
Generative AI (GenAI): AI technologies designed to produce original content, such as written text, images, music, software code, or video.
-
Large Language Models (LLMs): AI models trained on very large text collections so they can process, understand, and generate language in a way that resembles human communication.
-
Retrieval-Augmented Generation (RAG): a method that improves the quality of language model responses by enabling the model to retrieve relevant, current information from external sources before generating an answer.
How can AI support policymaking?
Artificial intelligence is increasingly being used to support public administrations throughout the policymaking cycle. According to the 2025 OECD study on AI in regulatory design and delivery, AI can help analyse large volumes of legislation and policy documents, identify patterns and inconsistencies, support impact assessments and legislative drafting, and facilitate the processing of stakeholder input. By automating repetitive and time-consuming tasks, AI allows policymakers to devote more time to strategic analysis, evidence-based decision-making and expert judgement.
Across Europe, administrations are already experimenting with AI-powered tools to improve the efficiency, quality and consistency of policy design and implementation. The examples below illustrate how AI can support digital-ready policymaking (DRPM) from legislative drafting and regulatory assessments to research, citizen engagement and access to public services.
AI4DRPM
Since January 2025, interoperability assessments became mandatory under the Interoperable Europe Act. To comply with this requirement, the European Commission introduced the Digital Statement (DS) to help assess the digital implications of new legislative proposals.
Preparing the DS requires policy officers to identify provisions with digital implications, classify requirements and describe aspects such as data flows and IT needs. As this process is burdensome and time-consuming, the European Commission started testing AI4DRPM in 2025 to support policy officers in their tasks.
AI4DRPM is a system that combines machine learning classification with GenAI to analyse draft legislation, identify digitally relevant provisions and classify them into categories such as data, digital solutions or process automation. It can thus help prepare a first draft of the DS, reduce the time spent screening legislation and improve consistency between assessments, allowing policy officers to focus on more substantive analysis.
LEOS
Legislative drafting involves many contributors, comments and versions, making collaboration and version control difficult. To ease this process, the European Commission released LEOS (Legislation Editing Open Software) already in 2015.
It is an open source legal drafting platform where users can work on the same legislative text, add comments and suggestions, compare versions and co-edit provisions. What’s more, LEOS stores legal texts using Akoma Ntoso - a standard for structured legal documents. This makes legal text easier for IT systems to process and reuse, supporting interoperability.
While LEOS is not itself an AI tool, the Commission has explored how it can integrate AI. The 2021 study on Drafting legislation in the era of AI and digitalisation identified potential functions including drafting support, legal consistency checks and legal analysis.
L'Assistant
Public officials increasingly use GenAI for writing and research. The challenge is to provide these capabilities securely without creating fragmented landscape of tools. Starting from 2025, France’s Interministerial Directorate for Digital Affairs (DINUM) developed l’Assistant, a browser-based AI virtual assistant that allows public officials to draft and rephrase text, summarise documents, analyse information and ask questions.
Users can also upload documents and create “projects”, where the same instructions and reference material can be reused across different conversations. For longer document collections, the platform uses RAG to retrieve relevant passages before generating the response. The service relies on Mistral models and the Albert API on SecNumCloud-qualified infrastructure.
L’Assistant demonstrates how a shared platform can provide a common secure environment and lower the need for each administration to procure or develop separate AI tools.
Practical Guide to Justice
Citizens often know what problem they have but not which public service can solve it. Portugal’s Practical Guide to Justice addresses this through a conversational assistant. The Directorate-General for Justice Policy (DGPJ) coordinates the project, designed with Genesis.Studio and supported by Microsoft.
Users describe their situation in natural language and the tool provides guidance based on information from Portuguese justice services. The current version also includes a feedback mechanism to support human control and continuous improvement.
The tool supports more user-centred public services by allowing citizens to express their needs in natural language rather than navigating multiple websites, saving time and improving access to information.
PLLuM
In Poland, Regulatory Impact Assessments (RIAs) are mandatory for all normative acts. They require public servants to find relevant information, follow detailed guidance and apply consistent analytical standards, creating substantial administrative workload.
PLLuM, the Polish Large Language Model, offers a possible solution. The project started in 2024 through a consortium of Polish research institutions and was funded by the Ministry of Digital Affairs. PLLuM is a family of open language models designed for the Polish language and public sector.
The RIA Department of the Chancellery of the Prime Minister subsequently developed a pilot RIA Virtual Assistant. The chatbot combines PLLuM with RAG to retrieve regulatory information and guidance before generating an answer, helping officials navigate RIA requirements and prepare assessments.
What does this mean for DRPM?
Digital-ready policymaking aims to make legislation and public services more efficient, user-centred, transparent, interoperable and future-proof. The examples presented above demonstrate that AI can play an important role in achieving these objectives by supporting tasks such as legislative drafting, regulatory screening, research, impact assessment and citizen interaction.
While AI is not a substitute for policy expertise or democratic accountability, it can significantly reduce administrative burden, improve consistency and help public administrations make better use of available evidence. As AI technologies continue to mature, they offer a valuable opportunity to strengthen DRPM practices and support the delivery of more responsive, effective and digitaly-ready public policies, with public officials remaining firmly responsible for final decisions and oversight.
Login or create an account to comment.