Capgemini: Strong public sector interest in agentic AI over next 2-3 years, but data readiness a challenge

• 90% planning to explore, pilot Agentic AI over next 2-3 years
• 21% say they have the requisite data to train and fine-tune AI models

Capgemini Research Institute published a report last month, ‘Data foundations for government – From AI ambition to execution,’ based on surveys done in Dec ’24 and Jan ‘25, which found that two thirds of the public sector organizations it surveyed are already exploring or actively using generative AI (Gen AI) initiatives to aid the provision of public services. 

Public sector organizations are also preparing to embrace agentic AI, with 90% of survey respondents planning to explore, pilot, or implement the technology within the next 2-3 years. 

However, Capgemini cautioned that these organizations lag in crucial data readiness, hindering their ability to leverage the full potential of AI. Currently, they face significant challenges with trust, compliance, data management and data sharing.

[Report Methodology: In Dec 2024 and Jan 2025, the Capgemini Research Institute conducted a survey of executives from 350 public sector organizations in 13 countries, 10 developed and three developing nations. Two respondents from each organization – one from the IT/data function and one from a line of business were interviewed via an online questionaire. They represented orgs across six public sector segments: public administration, tax and customs, welfare, defense, security, and healthcare. They operated at various levels of govt, including national, state, local, and international, and were located across North America, Europe, APAC, and the Middle East. The surveys were complemented by interviews with more than 15 public sector leaders.]

Organizations have high expectations for AI

With governments seeking to boost efficiency, improve public services, and address complex societal challenges, public sector organizations have high expectations for AI. According to the new report, within the next 2-3 years, 39% of public sector organizations aim to evaluate the feasibility of agentic AI, 45% intend to explore pilot programs, and 6% plan to scale their existing agentic AI initiatives. Attitudes towards agentic AI adoption are mostly consistent across segments, levels of government, and organizational sizes. Nearly two-thirds (64%) of organizations have progressed to pilots and scaled deployments, or are exploring Gen AI, with this number rising to 82% in defense agencies, 75% in healthcare, and 70% in security.

“With rising citizen demands and stretched resources, public sector organizations recognize 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 (pic), Public Sector Global Industry Leader at Capgemini. “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. However, to reach this future, governments need to focus on building the right data infrastructure and governance frameworks.”

Organizations struggle with AI adoption due to data and trust issues

Despite ambitions to embrace and scale AI use, public sector executives cite data security issues (79%) and limited trust in AI-generated outputs (74%) as primary barriers to widespread adoption. In the EU, organizations report a significant gap in confidence when it comes to complying with the EU AI Act[1], with less than four in ten (36%) prepared to meet these requirements. 

To progress their Gen AI adoption, public sector organizations require better data mastery, with the public sector showing limited progress in key areas of data management and utilization since 2020. The report finds that only 12% of organizations consider themselves very mature in activating data, while 7% report being very mature in nurturing data and AI-related skills. Only a fifth (21%) of public sector organizations surveyed have the required data to train and fine-tune AI models, including Gen AI models.

Data sharing concerns and the rise of the Chief Data Officer

Data sharing is vital for AI adoption as it boosts the volume and diversity of data to enhance AI model performance and optimize decision making. But data sharing initiatives are further complicated by concerns about data, cloud, and AI sovereignty. Despite all public sector organizations surveyed either having or planning to have data sharing initiatives, they are not yet mature; most organizations (65%) worldwide are still in the planning or pilot stages.

Governments are increasingly recognizing the critical role of harnessing data in the public sector, and this is reflected in the growing prominence of Chief Data Officers (CDO) and Chief AI Officers (CAIO). As many as 64% of public sector organizations already have a CDO, while 24% plan to appoint one, showing a willingness to invest in dedicated leadership for data-driven governance. Furthermore, the increasing strategic value of AI has resulted in over a quarter (27%) of public sector organizations appointing a Chief AI Officer, over a quarter (27%) already having one and 41% planning to introduce this new C-level role.