Artificial intelligence in government is entering a new phase.
For the past two years, much of the conversation has focused on generative AI pilots, policies and public perception. Now, attention is shifting toward something more operational: agentic AI. Systems that can take action, coordinate workflows and operate with a degree of autonomy.
The opportunity is significant. But so are the challenges.
A recent IBM Institute for Business Value report highlights a striking reality: While 76% of organizations are already exploring or deploying AI agents, only a fraction are realizing meaningful return on those investments.
That gap is not about ambition. It is about readiness of content.
The barrier isn’t adoption. It’s execution.
Government agencies are not standing still when it comes to AI. Many are already experimenting with use cases that range from document summarization to workflow automation. But as AI evolves from assistive tools to autonomous agents, the requirements change.
Agentic AI does not just generate content. It makes decisions, triggers actions and interacts with systems.
That level of capability exposes underlying weaknesses that were easier to ignore in earlier phases of AI adoption.
IBM’s research points to three consistent barriers: unstructured data, weak governance and fragmented approaches to automation.
None of these are new problems. But agentic AI amplifies them.
Unstructured, siloed information limits an agent’s ability to understand context. Without that context, outputs become unreliable. Governance gaps create risk when systems are making decisions at scale. And disconnected automation efforts lead to what IBM describes as “agent sprawl,” where multiple tools operate without alignment to a common outcome.
In other words, the issue is not whether agencies can deploy AI agents. It is whether those agents can operate in an environment that supports accurate, accountable and coordinated action.
Why agentic AI raises the stakes
Earlier forms of AI could be layered onto existing processes with relatively low risk. If a summary was imperfect or a recommendation missed nuance, a human could intervene before action was taken.
Agentic AI changes that dynamic.
These systems are designed to move work forward. They rely on access to information, an understanding of process and the ability to operate within defined rules. When those elements are inconsistent, the risks scale alongside the benefits.
This is why governance continues to emerge as the central theme in AI conversations across government. Not as a compliance exercise, but as an operational requirement.
As highlighted in previous discussions with public-sector leaders, trust in AI is built on trust in data. That principle becomes even more critical as AI systems move closer to execution rather than insight.
The role of structured information
If agentic AI depends on context, then context depends on how information is managed.
Most government agencies today operate across a complex landscape of shared drives, legacy systems, email and paper-based processes. Valuable information exists, but it is often fragmented, duplicated or difficult to access in a consistent way.
AI agents require high quality, well-structured data to function effectively. Without it, organizations see errors, inefficiencies and missed opportunities.
This is where the conversation needs to shift from AI to information management.
Before agencies can scale agentic AI, they need an environment where content is:
- Organized and classified as it enters the system
- Governed by consistent policies for access, retention and security
- Connected across processes, not isolated within departments
- Traceable, so decisions can be explained and audited
These are not new goals. But they take on new urgency in an agent-driven environment.
Another key insight from IBM’s report is the risk of focusing too narrowly on task-level automation.
Many organizations start by automating individual steps. While this can deliver incremental gains, it often results in disconnected systems that do not contribute to a larger outcome.
Agentic AI requires a different approach.
Instead of automating tasks in isolation, agencies need to think in terms of end-to-end processes. That means ensuring information flows seamlessly between systems, that decisions are made with full context and that governance applies consistently at every step.
In practical terms, this is less about deploying more AI tools and more about creating an environment where those tools can operate together.
An orchestration layer only works when the underlying information is structured, accessible and governed.
Preparing for what comes next
Agentic AI is not a distant concept. It is already being tested and deployed across industries, including the public sector.
The agencies that will benefit most are not necessarily the ones moving fastest to adopt it. They are the ones preparing their information environment to support it.
That preparation does not begin with algorithms.
It begins with content.
When information is properly managed, classified and governed, agencies gain something more valuable than efficiency. They gain confidence. Confidence that systems are operating on trusted data. Confidence that decisions can be explained. Confidence that innovation will strengthen, rather than erode, public trust.
Agentic AI has the potential to transform how government work gets done, but it will only deliver on that promise if the data behind it is ready.
And for most agencies, that is where the real work begins.
Andy MacIsaac is a senior marketing leader at Laserfiche, where he drives go-to-market strategy and thought leadership for AI-powered content management, process automation, and data governance in the public sector. With more than two decades of experience partnering with government agencies and education institutions, he helps organizations modernize operations while maintaining security, compliance, and trust. Andy has led industry marketing, demand generation, and sales enablement initiatives across leading software and consulting organizations, translating complex technologies into practical outcomes. As a trusted advisor to CIOs and agency leaders, he is passionate about responsible innovation that improves efficiency, transparency, and service delivery.
Leave a Reply
You must be logged in to post a comment.