The Grimoire · So You Don't Pull a Cher
Every AI term, mechanism, and "what does that actually do?" — fully explained, in plain English. No vests, no maxxing, no condescension.
What it is: AI that creates a new draft, image, summary, plan, or bit of code from your instructions.
Where it comes up: ChatGPT, Claude, Gemini, image tools, and most "make this for me" AI features.
Analogy: Not Google. More like an assistant making the first messy version so you are not staring at a blank page.
What it is: The instruction you give an AI tool.
Where it comes up: Anytime you type into ChatGPT, ask for an image, or tell Codex what to change.
Analogy: A brief with standards: audience, task, tone, context, and what good looks like.
What it is: When AI gives an answer that sounds polished but is wrong or made up.
Where it comes up: Names, dates, laws, links, quotes, sources, numbers, or anything high-stakes.
Analogy: Regina George confidence, zero receipts. Verify before you forward.
What it is: The place a claim comes from, such as an official page, policy, contract, meeting note, report, or person who owns the decision.
Where it comes up: When AI gives you a fact, date, price, quote, policy summary, or recommendation you may rely on.
Analogy: The receipt, not the story about the receipt.
What it is: A pointer to the source behind a claim. It might be a link, document name, page number, or quoted reference.
Where it comes up: Research summaries, vendor comparisons, policy notes, legal-sounding claims, and anything AI says came from somewhere specific.
Analogy: A name tag for the claim. Helpful, but you still check whether the person is actually on the list.
What it is: Checking whether an AI claim is accurate, current, relevant, and supported before you use it.
Where it comes up: Before sending AI output in an email, deck, client note, policy question, budget discussion, or anything high-stakes.
Analogy: Elle Woods checking the evidence before anyone gets to act confident.
What it is: Something AI filled in because the prompt or source material did not explicitly say it.
Where it comes up: When AI turns a possibility into a decision, guesses context, blends similar companies, or answers like it knows your workplace.
Analogy: The part of the Burn Book entry that sounds plausible until someone asks who actually said that.
What it is: The trained system inside an AI product that predicts and generates the response.
Where it comes up: When people say GPT-5, Claude Sonnet, Gemini, or "which model are you using?"
Analogy: The app is the venue. The model is the person doing the work in the back room.
What it is: The background information AI needs to understand your situation.
Where it comes up: When the first answer is generic and you need to add audience, goal, constraints, examples, or documents.
Analogy: Telling your stylist the event, weather, dress code, and shoes before asking for the outfit.
What it is: A chunk of text an AI model reads and predicts with. A token can be a word, part of a word, punctuation, or a weird little text fragment.
Where it comes up: Token limits, pricing, long documents, context windows, and the viral "how many Rs are in strawberry?" problem.
Analogy: The model may not see strawberry as nine neat letters. It sees chunks. That is why an LLM can confidently say there are two Rs when there are three.
What it is: AI that can take multiple steps toward a goal, sometimes using tools along the way.
Where it comes up: Coding assistants, research agents, workflow builders, and tools that can plan, act, check, and revise.
Analogy: An assistant with a checklist. Useful, but still needs boundaries and review.
What it is: The app you use to open websites.
Where it comes up: Visiting ChatGPT, reading docs, testing your site, or checking whether a link works.
Analogy: Chrome, Safari, and Edge are the front doors. The website is what is inside.
What it is: The structure of a webpage: headings, paragraphs, links, forms, buttons, and images.
Where it comes up: When you want AI to create a page, fix newsletter HTML, or understand what is on a site.
Analogy: The bones of the outfit before hair, makeup, and accessories.
What it is: The styling rules for a webpage: layout, colors, fonts, spacing, and responsive behavior.
Where it comes up: When text spills out of boxes, buttons look wrong, or a page needs to work on mobile.
Analogy: Same content, better blazer. CSS is what makes it look intentional.
What it is: Code that makes a webpage interactive after it loads.
Where it comes up: Quote generators, filters, dropdowns, forms, calculators, and buttons that change what you see.
Analogy: HTML is the room, CSS is the decor, JavaScript is someone turning the lights on and moving the furniture.
What it is: A text-based place to talk to your computer by typing instructions instead of clicking app buttons.
Where it comes up: Starting a website preview, installing tools, running a script, or following developer setup instructions.
Analogy: The Starbucks counter. Apps are the printed menu. The terminal is where the off-menu options live.
What it is: A Windows terminal shell: the thing that understands Windows-style commands and passes them to the right tool.
Where it comes up: On Windows when instructions say to run something, list files, start a server, or check a folder.
Analogy: The Windows barista/register system. It takes the order, checks what ingredients exist, and either makes the drink or tells you why it cannot.
What it is: A typed instruction you give your computer in Terminal or PowerShell.
Where it comes up: Copying setup instructions, running tests, starting a local site, or asking AI what a command does before using it.
Analogy: The secret drink-menu order. Not listed on the friendly app menu, but if you say it exactly right, you get precisely what you wanted.
What it is: The exact location of a file or folder on your computer.
Where it comes up: When you send AI a file location, upload assets, fix image links, or open a local HTML page.
Analogy: The full address, not "that cute place downtown."
What it is: The ending of a filename that tells your computer what kind of file it is.
Where it comes up: HTML newsletters, images, Markdown docs, PDFs, CSVs, and anything AI asks you to save or upload.
Analogy: The label on the garment bag: .html, .png, .md, .pdf.
What it is: A small piece of code that places another service inside your webpage.
Where it comes up: Comments, signup forms, videos, maps, calendars, surveys, and other tools you do not want to build from scratch.
Analogy: Borrowing the professional photo booth instead of constructing one in your living room.