AI Agents from Scratch
A practical, ground-up introduction to AI agents. Understand the ReAct loop, tools, orchestration, memory, and evaluation, then build real agents using NexusTrade's Aurora platform as your lab.
What You'll Learn
- Explain the difference between a bare language model, a chatbot, and an AI agent
- Describe the ReAct loop (Thought → Action → Observation) and why it makes agents agentic
- Design tool-calling workflows and understand why tools, function calling, and MCP servers are all the same core concept
- Use autonomy controls (whitelists, approvals) to safely run agents in production
- Connect memory, scheduling, and subagents into a full autonomous workflow
- Build traces and evaluators (including LLM judges) to measure agent quality
Course Content
Module 1: Welcome
- Who This Course Is For: What you'll learn, and why it matters
- Pre-Course Reflection: Where are you with AI today?
Module 2: What Is an AI Agent?
- ChatGPT Is Not an Agent: Language models vs. chatbots vs. agents
- Prompt Engineering: What it is, what a system prompt is made of, and why zero-shot isn't enough
- Exercise: Build a System Prompt: Compose a real system prompt and run it against Gemini 3 Flash.
- Prompt Engineering Trade-offs: The three levers every prompt trades: cost, latency, and accuracy
- Tools: How AI Actually Does Things: Function calling, APIs, and the structured-output trick
- Ask Aurora a Question: See a real tool call happen in Aurora
- Exercise: Walk Through the Classifier: See how NexusTrade routes your message to the right specialized prompt.
- Quiz: What Is an AI Agent?
- What Are MCP Servers?: Tools go by many names
Module 3: Orchestration and Autonomy
- The Orchestration Loop: How a plan turns into a while loop
- Autonomy and Permissions: Whitelist, blacklist, and the approval toggle
- Build a Strategy with Aurora: From one prompt to a full strategy
- Scheduling and Triggers: From manual start to autonomous execution
- Subagents: When one agent calls another
- Subagents and Scheduling in Practice: Hands-on walkthrough
- Quiz: Orchestration and Autonomy
Module 4: Memory
- Memory: From stateless models to vector databases and RAG
- Exploring Aurora's Memory: What does it remember about you?
- Quiz: Memory
Module 5: Evaluation and Observability
- Evaluation and Traces: How do you know your agent is doing what you want?
- Traces and Observability: What to capture and why
- Quiz: Evaluation
- Connecting Aurora's MCP Server: Use NexusTrade's tools from anywhere
Module 6: Capstone
- Final Recap: Everything you've learned, in 5 minutes
- Capstone: Build, Test, and Evaluate: An end-to-end autonomous workflow
- Course Reflection: Compare your answers to the pre-course reflection
- Get Your Certificate: Download your completion certificate