IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Making the Most of AI: How Agencies Can Get Their Data Ready

As government agencies consider the potential of new AI technology across the enterprise, they keep coming up against the same question: How do they prepare the data needed to deploy these solutions successfully?

Hands typing on a laptop with colorful graphs emerging from the screen. Dark background.
It’s no secret that the public sector has often lagged in technology adoption. This trend continues with the adoption of AI-driven solutions. Many states have been aggressive, while others have been in more “wait and see” mode.

AI platforms are designed to improve processes and reduce operating costs. They help ensure efficiency in supply chain operations and delivering customer service despite workforce shortages. They can also help employees by prepopulating digital forms with data from Word documents, then automating reminders and follow-up communications, summarizing meetings and drafting paperwork. In addition, generative AI can help find answers in volumes of existing policies, summarizing large documents or suggesting content for review.

Ultimately, AI can help agencies free up people to be more focused on strategic activities and automate redundant tasks. But only if the documents and answers they provide are readily available.

Positively, we see more proofs of concept and trials ramping up at government agencies. Leaders are looking for low-risk areas where they can get their feet wet and become comfortable working with new technologies such as AI. Many are upgrading to scalable, cloud-based technology powered by AI. They’re also launching modernization efforts, including chatbots, self-service tools and mobile apps.


The problem many agencies face when leveraging these capabilities is the foundational data is often siloed and duplicated. Related data is stored in different platforms, which are not in sync. There is also a significant amount of documentation and non-relational data that does not reside in structured systems. It could be in someone’s PC or hard drive, or even in someone’s head as institutional knowledge.

So, the primary challenge agencies must tackle is how to approach organizing this data to make it useful for employees and constituents alike — how to get the information out of the head and into the AI model.

The first step is to identify the question you’re trying to answer and the insights you seek to find in the data. This will help you determine the information you need to make available.

When locating and organizing your data, think about the easiest problems to fix and the places where we can help move the needle with the greatest impact. Think of this process as how you would approach spring cleaning your house. You can’t clean the whole house at once, so perhaps start with a bathroom — or even the closet in the bathroom.

In the case of organizing data, identifying small areas to initiate your “clean out” like data silos is a good place to start. Also, target areas that can be addressed with a one-time update that corrects problems or improves efficiency.

Now that you have organized the items in the bathroom closet, identify the quality of the contents. When looking at your data ask yourself: Does it have redundancies? Is it secure? What is the source? Is it reliable?

Once you have the data organized and redundancies eliminated, you must assess if the data set is complete to answer the questions you’ve identified. If not, what are the gaps and what other sources can you go to to fill them in? Then repeat the process to locate and streamline the data to create a comprehensive baseline.


Governments are leveraging AI to improve public service across various use cases, including in health care, transportation, the environment and benefits delivery. AI offers agencies untapped potential, and, moving forward, it will become foundational.

But for most organizations, using data and AI requires a significant data management overhaul. That includes identifying and assessing the value of existing data, designing a scalable data platform and developing a long-term data strategy to help the organization achieve impact at scale.

It also takes sustained commitment from management and a willingness to make the up-front investment needed, combined with skilled advisers with the experience and technical resources to help organizations design and implement their programs. Agencies that make this kind of investment will be the ones who are best able to serve their constituents in the AI era.

Celeste O'Dea is vice president of federal capture and engagement at Oracle.