For the last months I have been working analysing the public procurement of artificial intelligence solutions in Public Administration. The results have now been published as a playbook: Meaningful Stakeholder Engagement in Public Procurement for Artificial Intelligence. A Mission-Oriented Playbook.
The general goal of the research, fostered and commissioned by ParticipationAI, is whether AI is just a regular technology that can be purchased yet as another commodity, or is it “something else”. Our thesis is that, effectively, it is much more than something else. And that, at least, two crucial aspects should be taken into account:
- Purpose, aimed at the public interest.
- Impact, as it has a high transforming potential.
In both cases, we believe that the meaningful concurrence of a plural diversity of actors is strictly necessary at all stages of the life cycle of AI. Thus, Ai procurement needs quite a different framework from that of traditional procurement:
- Deal with it like the investment in a digital public infrastructure.
- Think of the concurrence of many different actors, and have a strategy on how to best engage them.
- Organise governance of AI procurement as a mission-oriented policy.
Resources:
Abstract
Artificial Intelligence is rapidly becoming a general-purpose technology with significant potential for transformation at all levels. The public sector is increasingly adopting it for a range of purposes – from improving public service delivery to designing and implementing policies – including promoting and shaping the technology itself in the public interest.
However, this potential does not come without challenges. The deployment of artificial intelligence has revealed serious issues, notably the quantity and quality of data required, the difficulty in training algorithms and understanding their inner workings, and the need to ensure compliance with administrative procedures and human rights more broadly. Ultimately, there is a persistent struggle to control its entire lifecycle.
As a result, the procurement of AI has diverged significantly from the conventional acquisition of standard technologies, as new complexities emerge in defining its purpose, shaping its delivery, and ensuring the transparency, predictability, accountability, and measurability of its impact.
To gain insight into this shift, we conduct expert interviews and a thorough literature review on the public procurement of artificial intelligence solutions and carry out interviews with key actors directly involved in the process.
Preliminary findings suggest that a comprehensive model is still lacking, although it draws heavily on previous experiences in data governance and digital privacy, particularly in relation to delivery.
However, the road ahead – especially beyond the strictly technical domain, in terms of purpose and impact – has been explored but remains largely uncharted, although there appears to be emerging consensus around the importance of meaningful stakeholder engagement, both quantitatively and qualitatively: the field is too vast and dynamic to be managed solely by public administrations or their contractors, and too complex to be addressed without the involvement of a diverse range of actors and publics, spanning various approaches, frameworks, disciplines, and roles.
In this playbook, we propose a future direction that identifies the key stages at which, and the ways in which, stakeholder engagement can add value to the entire process of public procurement of artificial intelligence solutions
Executive summary
Artificial Intelligence (AI) is reshaping economies, institutions, and societies at unprecedented speed and scale. As a general-purpose technology, AI holds significant potential for transformation – yet it also poses complex challenges, particularly in the context of democratic governance and the public sector.
Public administrations are increasingly adopting AI to enhance service delivery, optimise internal operations, and inform public policy. However, the integration of AI into public decision-making processes introduces new risks around transparency, accountability, equity, and public trust, among others. These challenges are particularly pronounced in the procurement and deployment of AI systems within the public sector.
A complex transformation
Unlike traditional technologies, AI systems evolve, interact with large-scale datasets, and often operate as opaque decision-making tools. Public procurement processes – typically designed for static, well-defined goods or services – struggle to accommodate the dynamic, systemic, and high-risk nature of AI systems. Existing procurement guidelines often do not capture the full lifecycle of AI or account for its institutional, ethical, and societal implications.
In addition, the governance of AI in the public sector is frequently confined to technical or legal compliance frameworks. While such frameworks are necessary, they are insufficient on their own to ensure that AI systems align with democratic values and deliver public value. What is needed is a broader governance perspective that includes not only rules and risks but also public purpose, institutional change, and meaningful engagement with affected communities and stakeholders.
Moreover, AI is coming — it’s coming fast, perhaps too fast for existing systems to keep up. Its development spans so many fronts that no single institution or sector can hope to stop it, let alone contain it. The pace and scope of this technological wave exceed the capacities of public administrations acting alone, just as they do those of the private sector. Navigating this complexity requires a collective effort: all actors — governments, industry, academia, and civil society — must come together, coordinate their actions, and share responsibility in shaping an AI future that serves the public good.
Purpose of this playbook
This playbook addresses a key governance gap in current practice: the limited integration of stakeholder engagement into the public procurement of AI. It seeks to support governments in designing more inclusive, anticipatory, and mission-oriented approaches to AI governance.
Its core argument is twofold. First, in the diagnostic dimension, the playbook contends that AI systems deployed by public administrations function as digital public infrastructure (DPI). These systems are not merely technological tools but foundational, enabling structures that underpin the delivery of public services, reorganise institutional workflows, and generate far-reaching societal effects through network externalities and data flows. Second, in terms of governance and solution design, the playbook advances the adoption of a mission-oriented policy approach as the most appropriate framework for steering the development and use of AI as DPI. This approach enables public institutions to define collective objectives, mobilise cross-sectoral resources, and embed public values such as transparency, inclusion, and accountability into AI governance. From this dual perspective, the playbook explores how stakeholder engagement can be systematically integrated across the full lifecycle of AI procurement—from problem framing and needs assessment to deployment, monitoring, and evaluation.
Approach
This playbook builds on:
- A comprehensive review of literature on AI governance, digital public infrastructure, and public procurement in Public Administrations.
- A qualitative analysis of 10 expert interviews with practitioners from public administration, the private sector, international organizations, academia and civil society.
- Conceptual frameworks in the field of public governance, public innovation, policy design and stakeholder engagement.
The report is structured around three interrelated pillars:
- Lifecycle framing of AI procurement, split in two parts: the state of the question of AI procurement according to the experts; and how an optimal, comprehensive approach could be organised into three phases: Purpose, Delivery, and Impact. And the proposition to deal with the procurement of AI solutions as an investment in digital public infrastructure.
- Stakeholder ecosystem mapping, disaggregating roles and functions across government, civil society, academia, and industry.
- Governance tools, applying a mission-oriented policy model, and including canvases and engagement instruments, tailored to support public institutions in implementing mission-oriented AI strategies.
Key insights
- AI in public administration is more than a tool; it is infrastructure.
Its systemic effects – on internal processes, interdepartmental coordination, and societal norms – require strategic governance beyond standard procurement procedures. - Current procurement models are not fully aligned with AI’s characteristics.
Most existing guidelines emphasise legal compliance and cost-efficiency, but fall short to address the public interest, complex environments, evolving risks, ethical trade-offs, or the long-term impacts on institutional capacity and public trust. - Stakeholder engagement is underdeveloped and inconsistently applied.
Despite frequent references in strategy documents, engagement practices are often ad hoc, generic, or limited to consultation phases without meaningful influence on decision-making. - A mission-oriented approach provides a robust governance framework.
Missions allow public institutions to align AI adoption with long-term societal goals, coordinate actors across sectors, and embed inclusion, transparency, and accountability by design. - Governments need new instruments to operationalise engagement.
These include:- Actor maps to identify relevant stakeholders and their roles;
- A toolbox of participatory mechanisms, categorised by degree of institutionalisation and democratic function;
- Governance canvases for aligning purpose, design, delivery, and evaluation phases with engagement practices.
Policy implications
To ensure that AI serves the public interest and strengthens democratic governance, public institutions are encouraged to:
- Reframe AI procurement as public policy, not just technology acquisition.
Position AI systems as components of digital public infrastructure with strategic, ethical, and organisational implications. - Adopt lifecycle-based governance models.
Move beyond narrow procurement windows and embed oversight, auditability, and stakeholder feedback mechanisms throughout the AI system’s lifespan. - Invest in stakeholder engagement capabilities.
Build internal capacity for mapping, involving, and co-creating with diverse actors, including underrepresented communities, domain experts, and civil society organisations. - Use missions to structure cross-sectoral coordination.
Apply mission-oriented innovation frameworks to define clear objectives, share responsibilities, and monitor outcomes in an adaptive and participatory manner. - Develop engagement standards and frameworks.
Establish benchmarks for inclusive, meaningful, and proportionate stakeholder engagement in AI procurement, drawing on best practices from other environments.
Bibliography
The 157 references used to pen this report/playbook can be found at https://ictlogy.net/bibliography/reports/bibliographies.php?idb=158
Downloads
Peña-López, I. (2025). Meaningful Stakeholder Engagement in Public Procurement for Artificial Intelligence. A Mission-Oriented Playbook. Barcelona: Participation AI.
Peña-López, I. (2025). Meaningful Stakeholder Engagement in Public Procurement for Artificial Intelligence. A Mission-Oriented Playbook. Barcelona: Participation AI.
Share:
