Introduction to AI
14 modules, hands-on: local models, agents, automation and more. Built for Brett's 4090 PC.
Recommended start: Module 4 (Local LLMs) and Module 5 (AI agents).
Module 1 · How AI & LLMs work
- 1.1 What AI is (and isn't) coming soon
- 1.2 How a model learns coming soon
- 1.3 Tokens, next-word prediction & context windows coming soon
- 1.4 Strengths, limits & hallucinations coming soon
- 1.5 The model landscape in 2026 coming soon
Module 2 · Prompt & context engineering
- 2.1 Clear, specific prompts coming soon
- 2.2 Examples, structure & roles coming soon
- 2.3 Thinking step by step & breaking tasks down coming soon
- 2.4 Context engineering: what goes in the window coming soon
- 2.5 Templates & a personal prompt library coming soon
Module 3 · Using chat assistants well (Claude/ChatGPT/Gemini/Grok)
- 3.1 Picking the right assistant coming soon
- 3.2 Files, images & research modes coming soon
- 3.3 Projects, memory & custom instructions coming soon
- 3.4 Privacy, settings & good habits coming soon
Module 4 · Local LLMs
Module 5 · AI Agents
Module 6 · Tools, automation & MCP
- 6.1 How tool calling works coming soon
- 6.2 MCP: clients, servers & transports coming soon
- 6.3 Build your own MCP server coming soon
- 6.4 No-code & scheduled automation coming soon
- 6.5 Tool safety & permissions coming soon
Module 7 · Harnesses & coding agents
- 7.1 Anatomy of a harness coming soon
- 7.2 Build a tiny harness yourself coming soon
- 7.3 Coding agents on a real repo coming soon
- 7.4 Customising the harness: skills, subagent configs, hooks coming soon
- 7.5 Running agents unattended coming soon
Module 8 · Agent swarms / multi-agent
- 8.1 When multi-agent is worth it coming soon
- 8.2 Orchestrator-worker by hand coming soon
- 8.3 Frameworks: CrewAI, LangGraph & friends coming soon
- 8.4 Coordination, failure & cost control coming soon
- 8.5 Capstone: your personal agent team coming soon
Module 9 · RAG & knowledge bases
- 9.1 RAG concepts & when to use it coming soon
- 9.2 Chunking, embeddings & retrieval quality coming soon
- 9.3 Answers with citations coming soon
- 9.4 A personal knowledge base coming soon
Module 10 · Generative media (image/video/voice)
- 10.1 How image models work & prompting them coming soon
- 10.2 Local image generation on the 4090 coming soon
- 10.3 Video generation coming soon
- 10.4 Voice: speech-to-text & text-to-speech coming soon
- 10.5 Ethics, rights & labelling coming soon
Module 11 · Evals & reliability
- 11.1 Why evals; building a test set coming soon
- 11.2 Automated grading coming soon
- 11.3 Evaluating agents & traces coming soon
- 11.4 Regression testing & monitoring coming soon
Module 12 · Fine-tuning & customisation
- 12.1 When to fine-tune (and when not) coming soon
- 12.2 Preparing a dataset coming soon
- 12.3 LoRA fine-tune on the 4090 coming soon
- 12.4 Evaluate & compare coming soon
Module 13 · Safety, privacy & cost
- 13.1 Responsible AI & bias coming soon
- 13.2 Privacy & data handling coming soon
- 13.3 Understanding and controlling cost coming soon
- 13.4 Choosing cheaper & faster coming soon
Module 14 · Basic introduction to cybersecurity
- 14.1 Security basics & your threat model coming soon
- 14.2 Accounts: passwords, 2FA & phishing coming soon
- 14.3 Networks, firewalls & exposed ports coming soon
- 14.4 Patching & the Essential Eight coming soon
- 14.5 Secrets & API-key hygiene coming soon
- 14.6 AI-specific risks: prompt injection & agents coming soon
- 14.7 Backups & incident response coming soon