The FROOT Framework — 17 Modules for AI Mastery

Learning AI is overwhelming. There are thousands of tutorials, courses, and blog posts — but no structured path from "what is a token?" to "how do I deploy a multi-agent system with guardrails in production?"

FROOT fixes that.

The 5 Layers

FROOT is an acronym for the five layers of AI competency:

🌱 F — Foundations (F1-F4)

Start here. Understand the building blocks.

- F1: GenAI Foundations — What are tokens, embeddings, attention? How do LLMs actually work? - F2: LLM Deep Dive — Model families, context windows, temperature, top-p, structured output - F3: AI Glossary — 200+ terms defined with context (not just definitions) - F4: .github Agentic OS — The 7 primitives and how they wire together

🧠 R — Reasoning (R1-R3)

Learn how to make AI think better.

- R1: Prompt Engineering — Techniques, patterns, anti-patterns, versioning - R2: RAG Patterns — Chunking, indexing, retrieval, reranking, citation - R3: Deterministic AI — Zero-temperature, seed pinning, structured output, reproducibility

⚙️ O — Orchestration (O1-O6)

Build systems, not just prompts.

- O1: Semantic Kernel — Microsoft's AI orchestration SDK - O2: Agent Patterns — Single agent, multi-agent, supervisor, swarm - O3: MCP & Tools — Model Context Protocol, tool design, server implementation - O4: Azure AI Services — OpenAI, AI Search, Document Intelligence, Content Safety - O5: Infrastructure — AKS, Container Apps, GPU sizing, networking - O6: GitHub Copilot — Extensions, agents, custom instructions

🔧 O — Operations (implied in the T layer)

Run AI in production reliably.

🔄 T — Transformation (T1-T3)

Advanced topics for scaling.

- T1: Fine-Tuning — When to fine-tune vs prompt, LoRA, data preparation - T2: Responsible AI — Fairness, safety, transparency, compliance - T3: Production AI — Monitoring, drift detection, evaluation pipelines

How to Learn

Self-paced

All 17 modules are available at docs.frootai.dev/learning. Each module averages 750+ lines of content with code examples, diagrams, and practical exercises.

With Copilot

Ask the MCP server: ``


What does module F1 cover?
Search knowledge for "RAG chunking strategies"

Workshops

We offer guided workshops for teams. See docs.frootai.dev/workshops.

Why "FROOT"?

Because great AI systems need strong roots. You can't build a reliable RAG pipeline without understanding embeddings (F1). You can't deploy agents without knowing orchestration patterns (O2). You can't go to production without responsible AI guardrails (T2).

FROOT ensures you build from the ground up — no gaps, no blind spots.

Start learning at docs.frootai.dev/learning/f1-genai-foundations