Gartner estimates that agentic AI will autonomously resolve 80% of customer service issues without human intervention by 2029. Doing so could lead to a 30% reduction in operational costs, the research and advisory firm said.
“Agentic AI is poised to revolutionize the way service interactions are conducted,” Gartner said in a recent press release.
That could be a game-changer for government agencies, which deal with large volumes of data and face mounting pressure to increase efficiency with limited resources.
During Government Executive’s Generative AI Workshop, Hyland Vice President of Content Intelligence Rohan Vaidyanathan said agentic AI is a “very exciting approach to solving data and automation problems.”
Unlike previous AI models, agentic AI can act autonomously to complete tasks.
“With agentic AI, it's really goal driven,” Vaidyanathan said. “You have a goal that an agent or an autonomous or a semi-autonomous system is trying to achieve. And it works toward those goals by being able to retain context about that data, being able to plan and sense ahead, being able to also reason with the analysis that it is doing.”
The Hyland VP added that “not every decision or decision path needs to be preprogrammed.” This can provide more flexibility and result in AI agents “making the right choice based on the context of the data and the situation.”
> Watch the workshop | The evolution of agentic AI in government
With skyrocketing volumes of complex, unstructured data, many government agencies are turning to content intelligence and AI-powered automation. This report from Government Technology and Hyland explores how agencies can unlock the full power of their content to deliver critical insights and fuel innovation.
Yes, humans are still in control.
“I think the biggest myth or misunderstanding I hear is because it is agentic AI, it is completely autonomous. And there is this notion of multiple agents working together, and somehow AI is going to take it all over and we can’t control it,” Vaidyanathan said.
While it’s true that agentic AI can be autonomous, that’s usually for “low-hanging fruits, low-impact actions” and collecting data to answer questions, Vaidyanathan said. It’s “completely possible,” he added, to put checks and balances in place to make AI agents pause and wait for human intervention before a decision is made.
The idea of an AI model that actively pursues goals, continuously evaluates progress and adjusts to the context of incoming data is intriguing for government agencies. This is especially true for agencies that have long been bogged down by large volumes of unstructured data that’s scattered across siloed legacy systems.
From public records requests to licensing and permitting workflows, agentic AI can help government agencies extract insights, make preliminary decisions, improve operational efficiency and reduce manual burden.
An example cited by Vaidyanathan during the workshop: A city council processing a Freedom of Information Act request could use agentic AI to scan council minutes, cross-reference public records and decide what information to release — and have a human review sensitive information when needed.
> Learn more | AI-powered IDP is the future of intelligent enterprise
— Rohan Vaidyanathan, Vice President of Content Intelligence, Hyland
It’s all about the data — where it is and how it’s being used to inform decisions.
First, agencies need to unify their data.
With Hyland’s content intelligence capabilities, agencies can transform unstructured data into AI-ready content. This enables agencies to deploy AI across departments.
“Business processes are not reliant on just data from one system or the other,” Vaidyanathan said. “They have to look at the entire spectrum in a very unified manner and in a very contextual manner.”
Once agencies bring different pieces of data together, the next approach should be looking for ways to better understand and utilize the information.
“Unify your data. Start breaking down those silos and start enriching your context so that the right kind of information is available while making the decision and reflecting and learning from that decision,” Vaidyanathan said.
This starts with being transparent and being able to hand over control to the users when it’s requested and needed. This is especially important in many government use cases, Vaidyanathan noted.
— Rohan Vaidyanathan, Vice President of Content Intelligence, Hyland
Hyland is “very clear that data belongs to you, the insights from it belong to you, and we are not cross-replicating this data and training it with someone else’s data,” Vaidyanathan said.
“Every customer owns their data, their insights, and being clear about it is super important,” the Hyland VP added.
To build trust, Hyland provides service cards to customers to let them know what AI models they’re using, and how data is being collected and utilized.
“Where did you get answers to a question from? What exact parts of the document did you combine to answer a specific question? Being able to explain that decision and being able to visually show that goes a long way in reinforcing trust,” Vaidyanathan said.
This remains true with more advanced models of AI: Organizations need to maintain a log of how data flows through the system.
Was it completely automated? Were humans involved in the process? Those are questions you need to be able to answer.
Early use cases of agentic AI have involved reducing mistakes, cutting costs and making processes more efficient, Vaidyanathan said. But, he stressed, the possibilities are endless.
“One of the first things that government agencies can do is take stock of what their content looks like and which parts they need to unify in order to make their vision a reality,” Vaidyanathan said. “How do you take advantage of technologies like the Hyland Content Innovation Cloud™ in order to bring and unify that data together?”
Soon after, agencies can deploy autonomous AI agents to drive more complex workflows and unlock greater efficiency.
To fully realize the potential of agentic AI, government agencies need a robust content intelligence system that transforms content and unstructured data into actionable, AI-ready information.
Hyland’s AI-powered content, process and application intelligence products can deliver new levels of value at every touchpoint — via applications you use every day.
Are you interested in learning more about our Content Intelligence product line? Schedule a conversation to discuss next steps.
