Building the foundation for autonomous operations

A federal agency can have an AI system that identifies a supply shortage in seconds and an operating process that takes days to respond.

The model may perform exactly as intended. The mission still waits.

For years, government modernization focused on improving visibility. AI is accelerating that progress while exposing a different problem: In many organizations, information is moving faster than the institution itself.

An acquisition risk can be identified immediately. A maintenance issue can be flagged automatically. A supply disruption can be predicted before it occurs. Yet, the response still depends on disconnected systems, manual coordination and processes that were never designed for that speed.

As federal agencies and their industry partners build toward autonomous operations, success depends not on generating intelligence, but on turning insights into action.

Context: understand what the mission requires

Consider a delayed critical part. Identifying the shortage is the easy part. Leaders then need to know which equipment depends on it, what missions are affected, whether another source is available and what funding can support an alternative. Answers to those questions rarely live in one place.

That is often the first obstacle to acting on AI-generated insight: understanding the operational context around the problem. A shortage means something different depending on the mission, the resources available, the funding attached to it and the consequences of delay.

Information becomes useful when leaders can trust it and connect it to the decisions in front of them. That trust comes from authoritative records, consistent definitions and clear ownership.

Federal agencies are already putting AI to work. GSA’s inventory lists deployed acquisition analytics that classify transactions to inform multi-agency spending decisions. Agencies can build on that progress through connected workflows.

A decision is only as useful as the operational picture behind it. Leaders need to understand their resources and options while there is still time to act.

Integration: Connect insight to execution

Federal missions cross organizational boundaries. The systems supporting them need to work across those boundaries, too.

When inventory, procurement, maintenance and finance operate separately, people become the integration layer. They copy information, reconcile records and chase approvals. Each hand-off adds time between recognizing a problem and resolving it.

The Defense Logistics Agency’s (DLA) recent warehouse modernization effort illustrates the foundational work needed to change that. This summer, DLA completed the rollout of its SAP-based Warehouse Management System across all 24 distribution locations. At DLA Distribution Hill, Utah, the final site to transition, leaders described real-time inventory tracking and more efficient workflows supporting greater speed and accuracy.

Connected operations make it possible for AI to move beyond analysis and into execution. At Corteva Agriscience, AI extracts information from emailed documents and automates sales-order creation, reducing manual effort by 90 percent. The value comes from connecting information directly to a business process and removing friction.

In an autonomous enterprise, people and intelligent agents work together inside connected processes. An agent can identify an exception, assemble relevant information and coordinate the next authorized step. People remain responsible for consequential decisions, but they spend less time gathering information and more time exercising judgment.

Governance: Make authority explicit

An agency would not assign an employee a consequential responsibility without defining the role, permissions and expectations. Intelligent agents require that same discipline.

Federal agencies should treat AI agents not as a substitute for public servants, but as tools that help government deliver services with greater speed, consistency and reach. Properly designed agents can handle routine work, synthesize large volumes of public comments and regulatory data, identify emerging risks and give employees faster access to relevant policies and guidance. The objective is not to replace judgment, but to create more capacity for it.

Autonomy should reflect the consequence of the action. Routine, reversible tasks may permit greater autonomy. Higher-consequence decisions require explicit human authority and appropriate controls.

Agencies must also be able to trace the information used, the action taken and the approval behind it. Governance belongs inside the workflow, where it guides execution and handles exceptions.

Speed and accountability must advance together.

Build for outcomes

Start with a few consequential workflows where friction carries a visible cost. Simplify the process, establish a baseline and assign an accountable owner. Involve the employees who understand the dependencies and exceptions. Expand when performance, controls and workforce adoption support it.

AI is shortening the time it takes to identify problems. The next challenge is shortening the time it takes to respond to them.

Outpace. Out-sustain. Out-decide.

The mission gains little from an answer delivered in seconds if the organization still needs days or weeks to act.

Jamison Braun leads U.S. Public Services at SAP.