Connecting policy journalism to institutional information with Postilla AI
Public Policy, an Italian news agency specialising in political and parliamentary reporting, is building a platform that combines its live coverage of Italian and European institutions with official parliamentary documents and open political databases
By: Leopoldo Papi, Public Policy
In the newswire model, journalism is made of continuous updates. Reporters follow events as they happen, often from inside institutions, producing a large number of short, verified pieces based on direct sources, official documents and parliamentary proceedings.
At Public Policy, this is our daily work: covering political and institutional activity in real time. It is a form of journalism that depends on human presence, editorial judgement and source verification. For this reason, we do not see AI as a replacement for reporters. We see it as a way to support and strengthen their work.
This is the starting point of Postilla AI, a project designed to search and analyse institutional databases — beginning with the Italian Parliament and the European Parliament — starting from Public Policy articles. We are pleased to be developing it through the JournalismAI Innovation Challenge, supported by the Google News Initiative.
More than a news item
The idea is simple, but technically demanding. A newswire article can become more than a news item. It can also become a starting point for deeper research across official documentation: parliamentary questions, committee transcripts, legislative files, amendments, reports, plenary debates and supporting materials.
These public databases are extraordinarily rich. They record the activity of democratic institutions in great detail. In theory, they are open and available to everyone. In practice, they are often difficult to use. Information is scattered across thousands of pages, different formats, complex search interfaces and documents that no human being can reasonably read in full on a daily basis.
A new layer between journalism and documentation
This is where AI can create value. Not by producing journalism on its own, but by making public information easier to access, connect and understand.
Because newswire journalism already identifies the most relevant institutional developments, our articles can work as certified entry points into this wider documentary universe. Starting from a piece written by a journalist, the system can look for related official sources, retrieve relevant passages, summarise context and help users discover connections that would otherwise remain hidden.
In this sense, Postilla AI is not only a search tool. It is an attempt to build a new layer between verified journalism and institutional documentation: a way to navigate current affairs through official sources, guided by editorially selected information.
Two significant challenges
The project also comes with significant challenges.
The first challenge is building an AI agent capable of performing accurate and complete research. Institutional documents are not simple web pages. They have their own language, structure and metadata. A useful system must understand the information contained in an article, translate it into effective search strategies, identify the right databases, retrieve relevant documents and avoid superficial or misleading matches.
At the core of the project is the idea that institutional information should not be treated as a collection of isolated documents. Parliamentary activity is inherently relational: people, committees, legislative files, amendments, organisations and policy issues continuously interact with one another. Postilla AI therefore aims to organise information as a network of connections, building a graph of relationships that helps users move from a single document or article to the broader ecosystem in which it belongs. This graph-oriented approach makes it possible to surface patterns, recurring actors and hidden links that traditional document search often fails to reveal.
The second challenge is interface design. Finding documents is not enough. The information must be presented in a way that is useful for different users. Reporters may need detailed source trails, related acts and background material to enrich their coverage. Readers or professional subscribers may need advanced context, structured summaries and clear links to the original documentation. The goal is to create a UX/UI that highlights relevance, not just quantity.
The opportunity is substantial. Public institutions produce enormous amounts of information every day. Much of it is potentially relevant to journalists, companies, associations, institutions, researchers and citizens. Yet a large part remains underused simply because it is too complex, too fragmented or too time-consuming to explore.
Postilla AI is our attempt to address this gap. It aims to enrich human journalism by connecting it more effectively with official, public and open documentation. The ambition is not to automate reporting, but to make reporters — and the users of their work — better equipped to understand what institutions are doing, why it matters and where the evidence can be found.
_____
This article is part of a series providing updates from the second cohort of the JournalismAI Innovation Challenge, supported by the Google News Initiative. To read articles from our other grantees, click here.