Artificial Intelligence in Government: Agentic AI
Classroom, Virtual
Who Should Attend
This course is designed for government, military, and contractor professionals who are ready to move beyond using AI one prompt at a time and explore the next evolution of artificial intelligence: AI that can pursue goals, complete multi-step work, and take action within established boundaries.
The course is appropriate for professionals across agencies, occupations, and grade levels who conduct research and analysis, prepare reports and briefings, manage programs or projects, coordinate recurring processes, review information, support customers or stakeholders, or want to identify new ways AI can improve how government work gets done.
No programming or technical background is required.
If you can identify a process that takes too many steps, requires too much repetitive effort, or routinely consumes time that could be spent on higher-value work, this course is for you.
Why You Should Attend
Generative AI changed how we create information. Agentic AI is changing how work gets done.
Most professionals currently use AI one prompt at a time: ask a question, receive a response, review the result, and provide another instruction.
Agentic AI introduces something fundamentally different.
Instead of simply responding to individual prompts, an AI agent can be given a goal, determine what needs to happen, work through multiple steps, use available information and tools, evaluate results, and determine what should happen next—while operating within boundaries established by people.
Imagine moving from asking AI to draft a report to creating an agent that can gather information, organize it, identify missing elements, prepare a draft, check the work against established requirements, flag potential problems, and prepare the product for human review.
That is the move from AI assistance to AI-enabled action.
As organizations begin adopting these capabilities, professionals who understand agentic AI will be better prepared to identify opportunities, redesign inefficient processes, participate in AI initiatives, evaluate proposed solutions, and determine where human judgment and oversight must remain firmly in control.
And you won't just learn about agents in this course.
You will build one.
Our AI courses emphasize platform-neutral, transferable skills. Hands-on exercises may use selected commercially available or government-approved AI platforms to provide practical experience applying those skills.
Course Overview
This highly interactive one-day course takes participants beyond traditional generative AI and into the emerging world of agentic AI.
Participants will learn what makes an AI system agentic and explore the essential components that allow an agent to pursue a goal, work through multiple steps, use information and tools, evaluate results, and operate within defined boundaries.
Through demonstrations and realistic government scenarios, participants will examine the building blocks of an effective agent—including its goal, instructions, knowledge, tools and actions, decision points, guardrails, validation requirements, and human oversight.
But this course does not stop at understanding the technology.
Participants will learn how to identify government processes that may be appropriate for agentic AI, break those processes into manageable steps, determine what the agent should and should not be permitted to do, and establish appropriate human checkpoints.
Then they will put those concepts into practice.
The course culminates in a hands-on Agent Build Lab, where participants will design, build, test, and refine a functional AI agent for a defined workplace task using an accessible agent-building environment.
By the end of the day, participants will have moved through the complete agent development cycle:
Define → Design → Build → Test → Evaluate → Refine
Participants leave with more than an understanding of agentic AI. They leave having experienced how to build one, how to test one, and how to determine whether an agent should be trusted to perform a particular task.
Course Objectives
Upon completion of this course, participants will be able to:
- Distinguish agentic AI from generative AI, traditional automation, and standard prompt-based interactions
- Explain how goals, instructions, knowledge, tools, actions, decision logic, feedback, and human oversight work together within an AI agent
- Identify government tasks and processes that are—and are not—appropriate candidates for agentic AI
- Break a multi-step business process into goals, tasks, inputs, decisions, actions, and approval points
- Design an agentic workflow that defines the role of AI and the role of the human
- Build and configure a functional AI agent for a defined workplace task using an accessible agent-building environment
- Test an AI agent using expected, unexpected, and boundary scenarios and refine the agent based on observed performance
- Establish appropriate guardrails, human approval points, escalation criteria, and validation requirements
- Recognize key risks involving accuracy, privacy, security, authorization, accountability, and unintended actions
What You Will Learn
From Generative AI to Agentic AI
Understand the evolution from AI that responds to individual instructions to AI systems capable of pursuing goals and coordinating multi-step work.
What Makes an AI Agent an Agent?
Explore the essential components of an AI agent, including goals, instructions, knowledge, tools, actions, decision-making, feedback, guardrails, and human oversight.
Agents, AI Workflows, and Automation
Learn how agentic AI differs from traditional automation and generative AI workflows—and why not every process needs an agent.
Finding the Right Government Use Cases
Learn to identify recurring, multi-step work where an agent may improve efficiency, consistency, or quality, while recognizing tasks where risk, complexity, or required human judgment makes agentic AI inappropriate.
Designing an Agentic Workflow
Break an existing process into objectives, tasks, inputs, decision points, dependencies, actions, outputs, and human approval requirements before building the agent.
Keeping Humans in Control
Determine what an agent may do independently, what requires human review or approval, when an agent must stop, and when an issue must be escalated.
Guardrails, Risk, and Responsible Use
Examine what changes when AI moves from generating information to potentially taking action. Participants will consider accuracy, privacy, cybersecurity, authorization, accountability, unintended consequences, and appropriate limits on agent autonomy.
Build Your Own AI Agent
Move from concept to creation in the hands-on Agent Build Lab.
Participants will define an agent's purpose, establish its instructions and boundaries, identify the information and resources it needs, configure its workflow, establish human oversight points, and build a functional agent using an accessible agent-building environment.
Test, Challenge, and Improve Your Agent
Building an agent is only the beginning.
Participants will test their agents against normal situations, unexpected conditions, incomplete information, and boundary scenarios. They will examine how the agent responds, identify weaknesses or unintended behaviors, and refine its instructions, workflow, and guardrails to improve performance.
Course Format
This is not a lecture-only introduction to agentic AI.
The course is highly interactive, demonstration-rich, and lab-driven. Participants progress throughout the day from understanding agentic AI to analyzing workflows, designing an agent, building it, testing it, and improving it.
Learning activities include instructor demonstrations, government use-case challenges, workflow-mapping exercises, human-versus-agent decision scenarios, risk and guardrail exercises, structured testing, and the culminating Agent Build Lab.
Throughout the course, participants will repeatedly confront three questions that are central to responsible agentic AI:
What can the AI do?
What should the AI do?
Where must the human remain in control?
By the end of the day, participants will understand the answers—and will have applied them to an agent they built themselves.
Don't just learn what AI agents can do. Build one.