Agentic AI is set to witness increasing utilisation in public sector organisations in the near future. A report by the Capgemini Research Institute reveals that 90% of public sector organisations plan to explore, pilot, or implement agentic AI over the next 2-3 years.
However, despite this growing demand, the sector continues to grapple with data readiness challenges, which hinder the effective adoption of artificial intelligence technologies.
Titled ‘Data foundations for government – From AI ambition to execution’, the report highlights that nearly two-thirds (64%) of public sector organisations are already exploring or actively deploying generative AI (Gen AI) to support public service delivery.
The growing interest in agentic AI — systems capable of making autonomous decisions — reflects a broader move by governments worldwide to improve decision-making, enhance service efficiency, and address complex societal challenges.
However, the report also finds that only 21% of organisations currently have the requisite data needed to train and fine-tune AI models, including Gen AI. Data remains a fundamental barrier to AI execution, with many public sector bodies still in early stages of developing strong data foundations.
“With rising citizen demands and stretched resources, public sector organisations recognise the ways in which AI can help them do more with less. However, the ability to deploy Gen AI and agentic AI depends on having rock-solid data foundations,” said Marc Reinhardt, Public Sector Global Industry Leader at Capgemini.
The report shows that within the next two to three years, 39% of public sector organisations plan to evaluate the feasibility of agentic AI, 45% aim to explore pilot programmes, and 6% intend to scale up existing initiatives. The adoption outlook appears consistent across departments, levels of government, and organisation sizes. Sectors such as defence (82%), healthcare (75%), and security (70%) are especially active in exploring Gen AI.
Despite this momentum, concerns around data security and trust in AI outputs persist. As many as 79% of public sector executives identified data security as a primary challenge, while 74% expressed limited trust in the results produced by AI systems.
“To reach this future, governments need to focus on building the right data infrastructure and governance frameworks,” Reinhardt added. “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.”
The study also points to slow progress in building organisational capabilities around data. Only 12% of respondents consider themselves very mature in activating data, while just 7% claim maturity in fostering data and AI-related skills.
Data sharing is another key area facing hurdles. Although all surveyed organisations reported having or planning to have data sharing initiatives, most (65%) remain at the planning or pilot stage. Concerns about cloud, AI sovereignty, and data privacy continue to slow down these efforts, despite the role data sharing plays in boosting AI model performance and enabling better decision-making.
In response, governments are placing greater emphasis on leadership roles dedicated to data and AI. The report finds that 64% of public sector organisations already have a Chief Data Officer (CDO), while another 24% plan to appoint one. Similarly, 27% of organisations have a Chief AI Officer (CAIO), with 41% intending to introduce this role in the future.