A new report has revealed growing interest among public sector organisations in adopting agentic AI technologies. According to a study from the Capgemini Research Institute, around 90% of the surveyed entities plan to explore, pilot, or implement agentic AI over the next two to three years. However, despite this enthusiasm, the report also suggests significant impediments to the practical realisation of agentic AI strategies in public institutions, particularly in data readiness.
According to the study, within the next two to three years, 39% of public sector organisations aim to evaluate the feasibility of agentic AI. Around 45% intend to explore pilot programmes, while 6% plan to scale their existing agentic AI initiatives within the same timeframe.
Majority in public sector pursuing Gen AI
The report also indicates varied progress across different sectors. About 64% of public sector organisations have moved to pilot and scaled deployments or are actively exploring generative AI (Gen AI) initiatives. This engagement is particularly pronounced in defence (82%), healthcare (75%), and security (70%). However, the sector overall has shown limited advancement in data management and utilisation since 2020. Only 12% of organisations consider themselves highly mature in activating data, and a mere 7% report being advanced in cultivating data and AI-related skills.
“Looking ahead, governments can be more agile and effective as AI augments the work of government employees to source information, conduct policy analysis, make decisions, and answer citizen queries,” said Capgemini Public Sector Global Industry Leader executive vice president Marc Reinhardt. However, to reach this future, governments need to focus on building the right data infrastructure and governance frameworks.”
Organisations face data and trust issues in AI adoption
Currently, only 21% of public sector organisations possess the necessary data to effectively train and fine-tune AI models. Data security concerns (79%) and limited trust in AI-generated outputs (74%) were found to be the major barriers. Compliance with the EU AI Act also presents challenges, with only 36% of European organisations feeling prepared to meet these regulatory requirements.
Data sharing is critical for enhancing AI model performance and decision-making, yet concerns about data, cloud, and AI sovereignty complicate initiatives. Although all surveyed organisations have data-sharing initiatives or plans, 65% remain in the planning or pilot stages.
Reflecting the strategic importance of data, the roles of Chief Data Officers (CDOs) and Chief AI Officers (CAIOs) are becoming more prominent. Currently, 64% of organisations have a CDO, and 24% plan to appoint one. Similarly, 27% have a CAIO, and 41% aim to introduce this role.
The report’s insights are based on a survey conducted in December 2024 and January 2025, involving 350 public sector organisations across public administration, tax and customs, welfare, defence, security, and healthcare segments. Executives from IT/data functions and lines of business participated, representing organisations across various government levels and regions, including North America, Europe, APAC, and the Middle East.
Read more: Enterprise AI adoption accelerates, but ROI remains elusive, survey finds
