AI Foundations AI learns from examples, not fixed rules, like a spam filter. Large Language Models (LLMs) are advanced autocomplete, predicting tokens (text chunks) word by word. Training involves reading vast text, guessing missing words, and learning from human feedback. AI can "hallucinate" by confidently stating incorrect information; always double-check facts. Treat AI like a brilliant intern: give clear instructions, context, and review its work. AI Buzzwords Explained AI is the broad term for machines doing human-smart tasks. Machine Learning (ML) is AI learning from examples, not rules. Deep Learning uses neural networks with many layers for complex tasks. Generative AI creates new content (text, images, music). LLMs are generative AI specialists for language. AI Agents go beyond chat; they plan steps and use tools to complete goals. Prompting for Better Results The AI can't read your mind, only your prompt. Use the 5-part formula: Role, Task, Context, Format, and Examples. Be specific: "Write a product description" vs. "Write a 60-word description for a handmade brown leather wallet for men who like classic style, with a warm tone, fitting eight cards." Treat prompting as a conversation; refine with follow-up questions. Mistakes: vagueness, cramming too much, forgetting audience, accepting the first answer, and including sensitive data. AI Agents: Doing Real Work Chatbots answer; agents achieve goals using tools. Agents follow a loop: goal, plan, use tools, check results, deliver. Tools like web search, file access, and code execution give agents hands. Delegate research, comparisons, drafting, but keep humans in charge of money and critical decisions. Provide agents access only to needed data and ask for confirmation before important actions. Creating Images with AI Image generators turn words into pictures by cleaning random noise based on training data. Prompt recipe: Subject, Style, Setting, Mood/Lighting, Composition, Colors, optional Text. Match prompts to the image's use (e.g., simple logos for small sizes). Tools: AI assistants (ChatGPT, Gemini) for quick edits, Canva for layouts, Midjourney for artistic visuals, Idogram for text, Adobe Firefly for commercial use. Use AI for drafts and brainstorming; refine important assets like logos with human designers. AI for Slides and Documents Workflow: Notes -> Outline -> Draft -> Edit -> Export. AI turns messy notes into structured outlines and designed drafts quickly. Edit carefully: check facts, use real images, cut text, and add your personality. Tools: Gamma, Canva, Google Slides with Copilot. Export as PDF for sending, editable files for collaboration, or links for quick viewing. AI for Audio: Voice and Music Text-to-Speech (TTS) predicts how human speech sounds, with expressive AI voices. Scripts should be ear-friendly: short sentences, spelled-out symbols, use punctuation for pauses. Voice cloning requires explicit consent from the voice owner due to ethical and legal risks. AI music tools generate songs from text descriptions; prompt for genre, mood, instruments, and length. Check licensing terms for commercial use and keep proof of your plan/source. AI Video Generation Text-to-video creates clips from descriptions; image-to-video animates existing images. Most tools generate short clips (5-10 seconds); combine them in an editor. Video prompt formula: Subject, Action, Setting, Camera, Style. Keep one main action per clip. Maintain character consistency by reusing reference images and exact descriptions. Use ethically: disclose realistic AI content, get consent for real people, and avoid copyright infringement. AI Video Production Pipeline Four stages: Script, Voice, Visuals, Captions. Plan scripts with ~150 words per minute of narration. Generate audio scene by scene for easier editing. Animate for explanations ("why"), use real footage for demonstrations ("how"). Autogenerate captions but always proofread them. Add chapter timestamps for long videos and do a final review. Faceless YouTube Channels & AI Rules Faceless channels use narration over visuals (animations, footage, slides) or AI avatars. Workflow: Niche -> Research -> Video Production -> Publish & Analyze. AI avatars can present videos but require consent if using a digital twin. YouTube requires disclosure for realistic AI content. Monetization requires original content that adds value; avoid low-effort, mass-produced videos. AI Co-workers: Claude Cowork & Skills Claude Cowork handles multi-step file-based tasks on your computer. Delegate repetitive tasks: renaming files, creating spreadsheets, summarizing reports. Share only a dedicated folder for Claude to access. Describe the desired outcome clearly; Claude plans and asks for permission. Skills allow teaching Claude a task once, ensuring consistent results. Building with Claude Code Claude Code creates files, tools, and websites from plain English descriptions. Works in a terminal; requires a project folder. Use commands like `mk\u00a0dir`, `cd`, and `claude\u00a0run`. Describe the goal, not the code. Use the `claude.md` file for project instructions. Safety: review actions, keep backups, never paste sensitive data. Vibe Coding: Describe, Build, Test, Fix Vibe coding: describe desired outcome in plain language; AI writes code. The loop: Describe -> Build -> Test -> Fix -> Repeat. Describe with goal, audience, purpose, and style. Build in small, testable steps. Test thoroughly: click everything, check on mobile, look for errors. Fixing: Report bugs specifically (what you did, expected, happened). Ask questions, use error messages. Building Your Personal Website Key elements: Headline, About section, Work examples, Contact info. Gather content first (text file). Prompt Claude Code to build a one-page site using your content and preferred style. Refine using the loop: make specific, small changes (e.g., "make headline bigger," "add white space"). Final checks: spelling, links, mobile view, and user feedback. Building a Business Website Focus on customer needs and a clear call to action. Structure: Benefit headline, CTA button, Services/Prices, Reviews, Contact info. Use real content and photos; honesty builds trust. Forms need a helper service (Netlify Forms recommended). Basic SEO: clear titles, meta descriptions, alt text, and mobile-first design. Publishing Your Website Live GitHub: Online storage for project files, tracks versions (commits). Netlify: Hosts your website, automatically updates from GitHub. Connect GitHub and Netlify accounts. Claude Code can automate GitHub repo creation and pushing changes. Deploy your site via Netlify, which provides a free `.app` domain and HTTPS. Domain Names, DNS, and HTTPS Domain: Your website's name (e.g., yourname.com), rented yearly. Registrars sell domains; check renewal prices and turn on privacy protection. DNS: The internet's contact list, mapping names to IP addresses. Use 'A' records (name to IP) and 'CNAME' records (name to name) to point your domain to Netlify. HTTPS provides security (padlock icon) via free certificates, crucial for user trust. Automation: Triggers, Actions, and Data Flow Automate repetitive, predictable tasks to save time. Rule: "When this happens (trigger), do that (action)." Data flow: Information (sender's name, email) passes between steps. Start with simple automations like email replies or form-to-spreadsheet. Tools: Built-in email responders or platforms like NA10, Zapier, Make. NA10 Workflows: Nodes and Data NA10 (Nodemation): Visual tool to build workflows with connected nodes (boxes). Nodes perform specific jobs (e.g., read form, send email). Workflows start with a trigger node and have multiple action nodes. Data travels between nodes as labeled items ("lunchboxes"). Debugging involves checking input/output data at each node. AI Chatbots with Memory and Rules AI chatbots understand natural language and handle unexpected questions. Memory nodes allow bots to remember previous conversation parts. System messages provide rules, personality, and facts to guide the AI's responses. Test bots rigorously with tricky questions and edge cases. Improve bots by refining system messages and testing consistently. RAG: AI that Reads Your Documents Retrieval Augmented Generation (RAG) allows AI to search documents before answering. Documents are split into "chunks." "Embeddings" convert text chunks into numbers representing meaning, allowing semantic search. A vector database stores these chunks and finds the closest matches to a query. Build two workflows: one to load documents into the vector store, another for the chatbot to search and answer. SMS and WhatsApp Automation SMS is suitable for short, expected updates (confirmations, reminders). WhatsApp offers richer media but has stricter business rules and requires approval. Twilio connects NA10 to the phone network for SMS. Official WhatsApp Business API is needed for automated WhatsApp messages. Strict rules apply: get consent, make opt-out easy, send useful info, and never buy phone lists. Telegram Bots for Answers and Alerts Telegram bots are free, easy to create via BotFather, and welcomed by the platform. Answer bots reply to customer questions (can use AI/RAG). Alert bots notify you of events (new orders, form submissions). Use the Telegram trigger and AI agent nodes in NA10. Keep bot tokens secret and use chat IDs to direct messages. Social Media Automation (Facebook) Automate scheduled posts and draft replies to common comments. Use a Google Sheet content calendar and the Facebook Graph API. Connect Facebook via Meta's developer portal, get a page access token. For comments, AI can draft replies to simple questions; escalate complaints to humans. Balance automation with personal interaction; avoid spammy or fake engagement. Voice Agents: Talking AI Assistants Voice agents use speech-to-text, an AI brain (LLM), and text-to-speech. They can perform actions via tools (calendar, email). Plan: Define job, personality, facts, tools, and human handoff points. Use clear, short language and test for robustness. Ethical use: be honest about AI, get consent for recordings, and never use for mass cold calls. Self-Hosting NA10 with Docker and Caddy Self-hosting offers control and potentially lower costs but requires responsibility for updates and security. Use a VPS (Virtual Private Server) or a mini PC. Docker packages NA10 and its dependencies into containers for easy deployment. Caddy provides automatic HTTPS (padlock) and domain routing. Essential steps: Rent VPS, configure DNS, install Docker, start NA10/Caddy, set up backups. Running AI Models Offline (Local AI) Local AI runs on your computer, offering privacy, offline use, and no per-message costs. Requires sufficient RAM (8GB+ recommended). Alama: A free app to download and run various local language models. Whisper: A model for transcribing audio into text offline. Combine local models for a private pipeline (e.g., Whisper for transcription, Alama for summarization). AI for Faster Learning Use AI for active learning, not passive consumption. Create study plans with clear goals and practice projects. Use "explain like I'm five" and analogies to grasp concepts. Quiz yourself regularly and review weak spots. AI can role-play, teach back concepts, and summarize long videos/documents. Always double-check facts, especially for important information. Earning Money with AI Skills Focus on solving business problems (e.g., saving time, improving customer service) rather than selling "AI." Offer specific services: websites, automations, chatbots, content creation, or teaching. Price for value, research market rates, offer packages, and consider monthly retainers. Find clients through personal networks, local businesses, and freelance platforms. Build samples, personalize outreach, and always get testimonials. Responsible AI Use Checklist Privacy: Remove private data before pasting into AI. Deep Fakes: Be honest about AI content; never fake real people without consent. Use secret codes for urgent requests. Copyright: Read tool terms, avoid copying protected work, get permission for uploads. Bias: Notice and question biased AI outputs; ensure human oversight for critical decisions. Responsibility: Always be the final decision-maker and use AI ethically.