Tier 0: AI Fundamentals for Managers
Build the conceptual foundation to evaluate AI opportunities, communicate with technical teams, and make informed decisions about AI investments.
What You'll Learn
This tier provides a non-technical introduction to modern AI. You'll understand the terminology, architecture patterns, and business applications of AI systems without writing code. By the end, you'll speak the language of AI teams and identify where AI adds genuine business value versus hype.
Lesson 1: The AI Landscape
Map the evolution from basic algorithms to Generative AI. You'll understand how AI, Machine Learning, Deep Learning, and Foundation Models nest inside each other—not as buzzwords, but as a historical progression of capability. We'll cover the three learning paradigms (Supervised, Unsupervised, Reinforcement Learning) and explain why Foundation Models represent an economic shift: train once, adapt everywhere. This lesson establishes the hierarchy that frames all subsequent discussions.
Lesson 2: What is a Large Language Model?
Deconstruct the engine behind ChatGPT and Claude. You'll learn that LLMs don't "know" facts—they're probabilistic engines trained on next-token prediction. We'll explain tokenization (why pricing is measured in tokens, not words), the Transformer architecture (how the Attention Mechanism maintains context over long conversations), and fine-tuning (why base models need instruction tuning to follow commands). This lesson demystifies how these systems work so you can set realistic expectations.
Lesson 3: What is a Vector Database?
Understand how AI systems "remember" your business context. You'll learn that traditional databases search for exact matches while vector databases search by semantic similarity—finding "refund policy" even when you ask about "money back guarantees." We'll cover embeddings (how text becomes math), similarity search (cosine distance), and why vector databases are the foundation of every RAG system. This lesson explains why AI can answer questions about your proprietary documents.
Lesson 4: Getting Better Answers
Master the practical techniques that separate good AI outputs from mediocre ones. You'll learn specificity (detailed prompts outperform vague requests), persona assignment (instructing the AI to "think like a CFO"), and output formatting (requesting structured responses like tables or JSON). We'll cover iteration strategies—how to refine prompts based on initial results—and when to break complex tasks into sequential steps. This lesson gives you the skills to get production-quality results from day one.
Lesson 5: Context Engineering
Learn why context is the difference between generic AI and AI that understands your business. You'll understand the context window (how much the AI can "remember" in a conversation), context priority (models pay more attention to the beginning and end), and RAG (Retrieval-Augmented Generation) as a pattern for injecting relevant documents. We'll explain why "context stuffing" fails and how to design prompts that guide the model's attention. This lesson teaches you to work with the AI's memory constraints, not against them.
Lesson 6: AI Seven Terms
Define the seven critical concepts that frame AI conversations: Prompts (instructions), Context (memory), Parameters (generation controls like temperature), Tokens (cost units), Embeddings (semantic search), Fine-tuning (specialization), and Inference (runtime execution). Each term connects to a business decision—context windows affect how much data you can analyze per request, parameters control creativity versus consistency, tokens drive API costs. This lesson creates a shared vocabulary for technical and business stakeholders.
Lesson 7: Generative vs. Agentic AI
Understand the architectural shift from content creation to autonomous action. Generative AI is reactive—you request text, it generates text. Agentic AI is proactive—you define a goal, it executes a multi-step plan using tools. We'll contrast linear workflows (Prompt → Response) with circular loops (Perceive → Reason → Act → Observe). The key insight: in Generative AI, the LLM is the mouth; in Agentic AI, the LLM is the brain. This lesson prepares you for the next wave of AI products that don't just advise—they execute.
Lesson 8: What are AI Agents?
Move from "stuck in the box" models to systems with hands. You'll learn about Compound AI Systems—surrounding the LLM with Tools (APIs), Memory (vector databases), and Verifiers (safety checks). We'll explain the shift from hard-coded logic to probabilistic reasoning: instead of writing "if weather then call API," you give the agent a goal and it decides which tools to use. The ReACT pattern (Reason → Act → Observe) becomes the cognitive loop. This lesson shows how agents transform AI from advisor to executor.
Lesson 9: Anatomy of AI Agents
Deconstruct production agent architecture into five components: Perception (how the agent interprets user requests), Planning (how it breaks goals into steps), Action (tool execution), Memory (short-term context plus long-term storage), and Orchestration (the control loop). You'll understand why agents need both episodic memory (this conversation) and semantic memory (knowledge base). We'll cover failure modes—what happens when tool calls fail—and recovery patterns. This lesson provides the blueprint for evaluating vendor solutions or scoping custom builds.
Lesson 10: Agent-to-Agent Protocol
Explore how multiple AI agents coordinate to solve complex problems. You'll learn the A2A (Agent-to-Agent) communication standard that enables one agent to delegate tasks to another. We'll cover role specialization (why you'd build separate agents for Research, Planning, and Execution rather than one mega-agent) and handoff protocols (how agents pass context without losing critical details). This lesson introduces the infrastructure layer that makes multi-agent systems practical.
Lesson 11: Multi-Agent Systems
Scale beyond single-agent workflows to orchestrated teams. You'll learn three collaboration patterns: Sequential (agent A finishes, then hands off to agent B), Parallel (multiple agents work simultaneously, results merge), and Hierarchical (a manager agent coordinates specialist agents). We'll cover when complexity justifies multi-agent architecture versus when a single well-prompted agent suffices. The frameworks (AutoGen, CrewAI, LangGraph) provide the orchestration layer. This lesson helps you evaluate when your use case needs multi-agent coordination.
Lesson 12: The AI Tech Stack
Map the five layers that transform models into products. Layer 1 (Infrastructure) provides compute—cloud APIs, on-premise GPUs, or edge deployment. Layer 2 (Models) offers intelligence—proprietary models like GPT-4 or open models like Llama. Layer 3 (Data) adds business context via RAG and vector databases. Layer 4 (Orchestration) manages the agent loop with frameworks like LangChain or AutoGen. Layer 5 (Application) creates the user interface. The vertical Observability layer monitors everything with tools like Arize Phoenix. This lesson shows you where different vendors fit and which layers you need to build versus buy.
Lesson 13: The Hard Limits of AI
Separate hype from reality by understanding what AI cannot yet do reliably. You'll learn about the Reliability Gap (hallucinations and the challenge of 100% accuracy), Resource Walls (data scarcity and energy constraints), and Cognitive Limits (the gap between simulating emotion and feeling it). We'll discuss why 99% accuracy fails for mission-critical systems and why AGI (general intelligence across all domains) remains distant. The new division of labor: AI handles execution and scale, humans provide intent and judgment. This lesson sets realistic expectations for AI initiatives.
Why This Matters
These thirteen lessons transform how you think about AI strategy. You'll stop asking "Can we add AI?" and start asking "Where does AI eliminate manual execution?" You'll evaluate vendor pitches with technical literacy. You'll set realistic timelines based on architectural constraints, not marketing promises.
Master these fundamentals before moving to Tier 1's hands-on implementation.
Complete these lessons to build the conceptual foundation for AI literacy.