The word lobbying often evokes images of secrecy, backroom deals and a lack of accountability. And clearly, across its long history stretching back all the way to the beginnings of organised decision-making, there has been no shortage of opacity. At its worst, lobbying obfuscates and obstructs decision-making, driven by narrow interests misaligned with the broader political mandate.

Yet at its best, lobbying contributes to a legitimate democratic function. When diverse stakeholders represent a broad range of views, they provide structured input—arguments, expertise and evidence—that helps lawmakers navigate complexity, reconcile different viewpoints or choose a decisive standpoint. This is how complex societies translate competing interests into workable rules through an accessible and transparent political process.

For the purposes of this analysis, lobbying is used as a neutral, technical term: it comprises any form of targeted advocacy aimed at informing, influencing or nudging policy-makers and legislators—whether conducted by multinational corporations, industry alliances, civil society organisations or NGOs. It does not discuss AI's role in relation to broader advocacy campaigns aimed at swaying public opinion.

As modern societies have grown more complex, and governance has become more interdependent across sectors, borders and levels, political decision-making itself has evolved into an increasingly elaborate process involving multiple actors, competing interests and intricate procedural stages. Advocacy has evolved in tandem.

Nowhere is this clearer than at the European level. A community of 27 Member States brings inevitable complexities and interdependencies—not necessarily by bureaucratic design, but by the very nature of multinational governance. These are deeply embedded in every stage of EU decision-making. The EU’s policy machinery comprises dozens of institutions, thousands of actors and multiple layers of procedure — from public consultations and impact assessments to committee deliberations and delegated acts — amounting to what certainly is one of the most intricate governance system in the world.

Yet contrary to common perception, the EU law-making system is in many respects more transparent and accessible than those of many national governments—at least from a formal procedural perspective. Public records and proceedings, official documents, stakeholder contributions and expert opinions are extensively documented on the EU's europa.eu domain, where the Union publishes detailed information about the objectives, research, stakeholder input and influences behind its policies and laws. And there is a real push from EU institutions for broad legitimacy by allowing multiple inroads for participation, such as public consultations, calls for evidence and impact assessments.

It may not be perfect, but consider for example the European Parliament's committees, where much substantive policy work occurs: their sessions are publicly accessible via live stream, whilst the respective committees of national parliaments often meet behind closed doors—even in some of the most transparent member states.

But here comes the catch: according to the Financial Times, during the 2019–2024 electoral period alone, the EU adopted 13,942 legal acts—with major legislative proposals undergoing elaborate proceedings involving consultations, impact assessments, hearings and multiple rounds of amendments. Such steps serve the important purposes of stakeholder consultation, public participation and broad legitimacy.

Complexity challenges transparency

Paradoxically, the multitude of granular steps also compounds the inherent complexity of decision-making in the world's most extensive supranational entity. When procedural intricacy becomes an obstacle to clarity—even if intended to ensure accessibility and participation—the issue is no longer a lack of transparency but the sheer volume and complexity of available information.

As a result of increasing formalization and public oversight, effective advocacy has become ever more diligent, technical and traceable, demanding exactly the right kind of input, in the right format, to the right actor, at the right stage of the process. Being out of step with the formal process risks missing critical and legitimate windows for contribution, meaning that even the most compelling arguments are lost. The value of highly specialized procedural expertise has thus grown dramatically, driving advocacy toward professionalization and higher professional standards.

In parallel, public sensitivity and interest in the legitimacy of lobbying has soared. The EU itself requires interest representatives to disclose their interests and activities in the Transparency Register, binding them to a code of conduct before they may undertake lobbying or obtain access to institutional premises.

As past cases have shown, it would be naïve to suggest that these advancements have eradicated all problems or that questionable practices are entirely disappearing. Yet they have significantly narrowed the manoeuvring space of the stereotypical rainmaker figure operating informally through networks and backrooms.

This shift toward transparency and professionalization is critically important for the legitimacy and societal acceptance of advocacy as a professional activity in democratic societies.

It is in this context that AI enters the scene—with the potential to accelerate the technical sophistication and specialization of modern advocacy. Importantly, AI could also level the playing field between actors with vastly diverging resources: by augmenting both the quality and volume of work at relatively low cost, it enables organizations of any size to operate more effectively.

But AI is also a technology that demands particularly transparent and ethical use, especially in a sector so sensitive to questions of legitimacy and influence.

What role could AI ultimately have in this field? To answer that meaningfully, we first need to understand what AI can and cannot do—the concrete use cases that determine its broader impact. Given the twin challenges of overwhelming procedural complexity and information overflow, AI's entry into the sector shouldn't be about replacing human expertise or relationships. If employed correctly, it can make navigable what is already public but practically buried in overwhelming complexity, ensuring that decision-makers have access to the best possible knowledge for making the best possible decisions.

Typical use cases

Much of the work involved in lobbying is operational, time-sensitive and document-heavy—the type of tasks where well-designed AI tools can provide immediate support. Below are some of the most promising near-term use cases where AI can assist in day-to-day advocacy work, augmenting rather than replacing human judgment or interaction. In the next parts of this series, we'll examine more use cases, also where AI struggles or fails to deliver reliable contributions, as well as the important questions of quality assurance, transparency, ethics and legal compliance.

1. Targeted intelligence and alerting

AI systems can be trained to scan institutional portals, parliamentary agendas, policy announcements, court registries, acts, delegated acts, and a myriad of similar open sources—aiming to detect early signs of developments across files and committees. While this may sound straightforward, it is anything but—especially if the goal is to reliably identify what is relevant for each user, avoid information overload, and prevent redundant reporting of the same signals.

Crucially, this is not about insider knowledge that policy professionals acquire through trusted personal contact—the kind no AI in the world can, or should, replace. It's about navigating publicly available information that is, in principle, accessible to all—but so vast, fragmented and procedurally complex that even seasoned policy professionals struggle to stay on top of it. And precisely because this information is public, no stakeholder can afford to miss it.

This work goes far beyond basic keyword monitoring. Isolated keywords alone cannot capture the procedural weight, contextual significance, or interdependencies that define EU policy signals. To surface meaningful developments, systems must understand where in the process a document sits, its surrounding context, and how it relates to a user's interest profile.

Ensuring relevance, accuracy, and completeness across a vast range of sources and formats is technically demanding. Reliable performance requires robust filtering, continuous source alignment, and an understanding of procedural context. When done well, however, such systems enable a shift from reactive tracking to anticipatory insight — helping ensure that no procedural milestone is overlooked.

2. Background information and context

AI tools can assemble background knowledge on complex policy files — from tracing the evolution of a legislative proposal to identifying related initiatives, stakeholder positions, and procedural timelines. This may involve retrieving and connecting documents such as explanatory memoranda, delegated and implementing acts, consultation responses, impact assessments, and legal opinions across multiple sources and formats.

But while the goal sounds intuitive, the technical challenge is considerable. Policy documents are often lengthy, inconsistently structured, and highly nuanced. Contextualizing them and presenting them meaningfully to users requires an understanding of institutional vocabulary, legal hierarchy, procedural dependencies, and cross-references that stretch across the legislative architecture of a given governance system—be it the EU or a national jurisdiction.

Achieving this requires finely tuned models and domain-specific knowledge integration. When successful, such systems don’t just retrieve documents — they help professionals quickly build a coherent picture of the policy landscape, reducing research time and enabling better-informed interventions.

3. Supporting content production

AI can assist with some of the most time-intensive aspects of advocacy: understanding complex policy texts and producing tailored responses. Lobbying often relies on clear, timely communication: position papers, stakeholder letters, consultation responses, press releases, and internal briefings. AI can help with drafting, summarising, or adapting content to specific audiences and formats — particularly under tight time constraints; of course, always requiring human oversight and discretion.

This offers major efficiency gains. With AI assistance, professionals can respond faster to emerging developments, articulate their positions more clearly, and iterate drafts more quickly across languages and levels of technicality. This will give an agenda-setting advantage to those who can communicate their position more rapidly and convincingly.

But real-world applications are far from plug-and-play. Reliable AI-assisted drafting in a policy context requires models to understand sector-specific language, legal nuance, and procedural implications. Inputs must be carefully structured and outputs rigorously verified, while the role of human judgement remains—and should remain—irreplaceable, particularly when it comes to strategic positioning, legal accuracy, or reputational tone.

Still, in practice, the ability to accelerate content preparation — without sacrificing quality — is already proving valuable and will become a standard feature of modern advocacy workflows.

4. Strategic horizon scanning

Another use case for advanced AI in lobbying lies in identifying emerging signals of relevance — not just within official procedures, but across the wider virtual environment. These might include early mentions of a policy concept or nascent ideas for new legislation in speeches, parliamentary debates, press releases, stakeholder events, industry statements, or media reporting. AI can help surface such developments before they consolidate into formal initiatives, enabling forward-looking positioning and thematic awareness.

Of course, such a strategic functionality goes far beyond the confines of advocacy and lobbying, but it supports them, too. The goal is not simply to track known files, but to detect signals that indicate shifts in the political or regulatory landscape — long before they show up in official registers. When aligned with a user’s priorities, this offers a valuable shift from reactive monitoring to strategic anticipation.

Such use cases are conceptually powerful — but also technologically demanding. Delivering them requires filtering and contextualising vast volumes of fragmented material obtained in diverse formats, connecting disparate sources, and calibrating notifications to reflect the evolving relevance of signals over time. Done well, it can support more agile, informed and timely engagement — especially in environments where timing, framing and first-mover positioning are decisive.

5. Stakeholder landscaping

Here, AI use becomes more nuanced when considering real-world requirements. Effective policy work requires more than understanding procedures or texts — it demands clarity about the people, organisations and networks shaping those texts. AI can assist in building and updating stakeholder maps by scanning official and open-source material to identify relevant actors, institutional roles, and public positions.

In formalised environments, this is relatively straightforward. When a legislator is appointed as a rapporteur, submits amendments, or files a written question, these actions leave visible, structured traces in public records. Tracking such formal signals — including committee affiliations, speaking engagements, or published consultation responses — is largely a matter of technical integration and retrieval, and does not necessarily require advanced AI.

However, real-world influence does not always follow formal structures. Individuals without official roles may exert disproportionate impact through informal leadership, behind-the-scenes negotiation, or strategic alliances. Detecting these dynamics requires broader monitoring across the virtual landscape — including press statements, event participation, social media presence, network effects and thematic clustering. This is where more advanced AI systems can contribute, by identifying patterns and associations across multiple unstructured sources.

Still, even the most sophisticated systems face clear limitations. Stakeholder maps produced solely by AI will rarely be complete or fully accurate. No algorithm has access to internal knowledge of policy bodies, personal relationships or informal groupings that experienced policy professionals navigate daily. Each professional brings their own knowledge of relevant contacts, internal stakeholders and strategic priorities — dimensions of influence that cannot be inferred from open data alone.

Therefore, AI-generated stakeholder maps can offer a useful starting point or first approximation of the playing field — especially for newcomers or in rapidly shifting environments. But without targeted human input, validation and refinement, they are not a reliable basis for planning advocacy activities. As of today, stakeholder intelligence remains a hybrid domain: AI can assist in organising and extending awareness, but only in tandem with the contextual judgement, memory and relationships that human professionals bring to the table.

Looking ahead

In the next part of this article, we'll look at more use cases—including complex and sensitive applications—exploring questions of quality assurance, transparency, ethics, and legal compliance in an environment where influence and integrity must coexist.

Follow Thembi on LinkedIn to be informed when the next article becomes available.

Stay in the loop with Thembi

We’re building a smarter way to track and understand EU policy, combining AI-powered monitoring with intuitive insights that deliver fast, comprehensive and actionable intelligence.

Thembi helps you stop chasing data and start driving policy work.

Sign up to be the first to know when we launch.