Industry Spotlight | September 25, 2024

AI agents power the public sector  

Government agencies often struggle with a myriad of repetitive and time-consuming tasks, mountains of data that need to be managed, and inefficient processes that can drain resources and hinder productivity. 

Meanwhile, citizens are often looking for more transparency from governments about the work they do; a more efficient way of engaging; and easier ways to access public services without getting lost in a maze of online forms, processes and confusing websites. 

Autonomous decision-making is here to help 

The good news is that AI can be a powerful force in helping government organisations make decisions, improve services and get work done. And we’re starting to see examples of how AI agents, in particular, are being used by governments to analyse vast datasets at incredible speeds, provide public sector bodies with insights they can put to work immediately, automate hundreds of processes, and enhance the services offered to citizens.

What we see coming next are multi-agent systems composed of multiple interacting intelligent agents, and a new approach called Agentic AI where systems will be able to augment human work further by understanding and interpreting complex information and goals and then take actions based on continuous learning and analyses of extensive datasets.

The Agentic AI approach is particularly well suited to addressing the needs of public sector organisations in the following three ways:

1. Improved efficiency 

Think of an AI multi-agent system as your team of expert co-workers, available 24x7, with the knowledge you need to collaborate on complex tasks.

Decisions are made in the context of specific use cases. Patterns and trends that traditional methods may have missed are identified, and complex workflows are adjusted in real-time as circumstances evolve. This allows human workers to focus on higher-value tasks, representing a significant shift towards augmenting workers with intelligence tailored to their roles.

Agentic AI can improve efficiency for public sector workers in these ways:

• Knowledge retrieval agent: Retrieves relevant information from a data source of tax regulations, policies and procedures to provide real-time answers to agents' queries.  

• Sentiment analysis agent: Monitors the tone and sentiment of the caller in real time, alerting the human agent if the caller is becoming frustrated or upset, and suggesting appropriate responses to de-escalate the situation.  

• Follow-up agent: Schedules and manages follow-up actions, such as sending emails or scheduling callbacks, ensuring that no task is overlooked.

Simplifying the tax process in Brussels

Identifying the most suitable business process opportunities is a critical starting point. And DXC’s work with the Brussels Tax Department is a great example of how AI agents can help simplify the entire tax administration process.

This was in response to a desire to address increased citizen demands; streamline administrative work; and remove backlogs of work to free up capacity to introduce new services and improve citizen experiences.

2. Improved government outcomes

With careful implementation, the multi-agent approach has the potential to deliver transformative outcomes that go beyond improvements in efficiency and engagement methods to include:   

• Faster decision making:  Examples include determining eligibility for grants, allowances, or support services and potentially for tax breaks.  

• Improved data interpretation: Agents can be used to improve fraud detection, resulting in increased revenues to spend on public services, among other examples. 

• Data-driven insights: Agentic AI can process and analyse large volumes of data to provide actionable insights for policymakers and service providers. This can lead to more informed decisions and a better allocation of resources. 

   

3. Better citizen experiences 

Beyond the ability to increase capacity, Agentic AI has the potential to improve experiences offered to citizens in multiple ways:   

• Personalisation: Agentic AI can gather and analyse individual citizen data to provide personalised recommendations for services or benefits, ensuring that citizens receive the information most relevant to them.  

• Customised communications: AI can adapt communication styles and language based on citizen preferences, making interactions more user-friendly and accessible.  

• Early intervention: AI can identify patterns that indicate potential issues for citizens (e.g., financial distress, health risks) and alert relevant authorities to intervene early.  

• Improved access to services: By simplifying interfaces and providing contextual understanding, Agentic AI opens the door to improving access and use of the data sources available across multiple government departments. 

Unlocking a treasure trove of cultural data in Italy

Sifting through decades of bibliographic information can be laborious and time-consuming for researchers, academics, students and others. That was the challenge the Italian Ministry of Cultural Heritage wanted to solve for users of its portal Alphabetica, which connects millions of records and digital materials from over 6,500 libraries in Italy’s National Library Service.

What makes DXC different?

DXC's consulting and engineering experts can help public and private organisations get started on their AI initiatives by:  

• Assessing an organisation’s readiness to start exploring AI use cases   

• Conducting a consultative workshop to explore potential use cases   

• Providing services for AI use case design, implementation and operation  

DXC is currently developing a multi-agent framework for assessing, designing, deploying and improving multi-agent AI, which can be applied to almost any government process.