☕ Study tool

AI flashcards.

Memorize the vocabulary of AI and modern development with spaced repetition. Flip a card, rate how well you knew it, and the deck schedules it to come back right before you would forget. 51 cards across 5 decks, free, no account, all in your browser.

Browse every card

All 51 cards in plain text, pulled from the dictionary. Study above, or just read through.

AI fundamentals (12)

LLM: An LLM (large language model) is an AI trained on huge amounts of text to predict the next word, which is how it writes and reasons.
Token: A token is a small chunk of text an AI reads at once, usually a short word or a piece of one.
Context window: A context window is how much text an AI can keep in mind at once, its short-term memory, measured in tokens.
Prompt: A prompt is simply what you ask an AI to do: the instructions or question you type in.
Hallucination: A hallucination is when an AI states something false with total confidence, inventing facts, sources, or quotes.
Model: A model is the trained AI itself, the thing that actually generates answers. GPT-5, Claude, and Gemini are all models.
Temperature: Temperature is a setting that controls how random or focused an AI's answers are.
Inference: Inference is the moment an AI model actually runs to produce an answer, as opposed to being trained.
Embedding: An embedding turns text (or images) into a list of numbers that captures meaning, so a computer can compare ideas.
Vector database: A vector database stores embeddings so an AI can quickly find the most similar pieces of information by meaning.
RAG: RAG (retrieval-augmented generation) lets an AI look things up before answering, so it can use your documents instead of only its training.
Fine-tuning: Fine-tuning is taking a general model and training it a bit more on your own examples so it specializes.

AI agents & MCP (8)

AI agent: An AI agent is an AI that can take actions on its own: not just answer, but use tools, run steps, and work toward a goal.
Tool calling: Tool calling is when an AI decides to use an external tool (search, a calculator, your API) instead of answering from memory.
MCP: MCP (Model Context Protocol) is a standard way to plug tools and data into an AI, so any model can use any connected tool.
Coding agent: A coding agent is an AI tool that does not just suggest code but reads your files, makes changes, runs commands, and fixes its own errors in a loop.
Structured outputs: Structured outputs force an AI to answer in a strict format like JSON, so your code can use the result reliably.
Multimodal: Multimodal AI can handle more than one kind of input or output: text, images, sound, even video.
Context rot: Context rot is when an AI coding assistant slowly loses track of earlier details in a long session and starts contradicting itself or inventing things.
Conversational AI: Conversational AI is AI you talk to in plain language, back and forth, like ChatGPT or a voice assistant.

Coding with AI (10)

Vibe coding: Vibe coding is building software mostly by describing what you want in plain English and letting an AI write the actual code.
Training cutoff: A model's training cutoff is the date its training data stops, so it knows nothing that happened after it.
System prompt: A system prompt is a standing instruction set at the start of a chat that tells the model how to behave, before you ask anything.
Sandbox: A sandbox is an isolated, locked-down space where code can run without touching your files, your network, or the rest of your system.
API key: An API key is a secret password that lets your app use an outside service, like an AI model, proving the request comes from your account.
Dev server: A dev server is a program on your computer that shows a live, instantly-updating preview of your app while you build it.
Localhost: Localhost is a private web address that shows your app only on your own computer, so you can preview it before anyone else can.
Scaffolding: Scaffolding is the empty starter skeleton of an app, the folders and blank files, that the AI sets up first before any real features.
Refactor: To refactor is to clean up and reorganize existing code without changing what the app actually does.
Rollback: A rollback is undoing a change by returning your app to an earlier version that was working.

Web & coding basics (12)

API: An API is a way for two pieces of software to talk: one asks for something, the other answers, over the web.
Git: Git is the tool that tracks every change to your code, so you can rewind, branch off, and work with others without chaos.
Commit: A commit is a saved checkpoint of your code: a snapshot, with a short note, that you can always return to.
Deploy: Deploy means to put your code live on the internet so other people can actually use it.
JSON: JSON (JavaScript Object Notation) is a simple text format for structured data, built from keys and values that both people and programs can read.
HTTP: HTTP is the language browsers and servers use to talk. Every page, image, and API call is an HTTP request and response.
DNS: DNS (Domain Name System) is the internet's phonebook: it turns a name like kbcafe.com into the numeric address computers use.
Domain: A domain is your website's human-friendly address, like kbcafe.com, that you rent and point at a server.
Environment variable: An environment variable is a setting stored outside your code, used for secrets and config like API keys.
Repository: A repository (or "repo") is the folder that holds all your project's code and its full history of changes.
Build: A build is the step where the computer packages all your code and files into a finished version ready to put online.
Terminal: A terminal is the window where you type commands to your computer, the place a CLI lives.

Startup & SaaS (9)

MVP: An MVP (minimum viable product) is the smallest version of your idea that is still useful enough to launch and learn from.
SaaS: SaaS (software as a service) is software you pay for by subscription and use in your browser, instead of buying and installing it.
ARR: ARR (annual recurring revenue) is the predictable subscription revenue a business earns in a year.
Churn: Churn is the rate at which customers cancel and leave. High churn quietly drains a business.
LTV: LTV (lifetime value) is how much money one customer brings in, on average, over the whole time they stay.
CAC: CAC (customer acquisition cost) is how much you spend, on average, to get one new paying customer.
Freemium: Freemium is a model where the basic product is free and you pay to unlock more, the most common way software grows.
Conversion rate: A conversion rate is the percentage of people who take the action you want, like signing up or paying.
Product-market fit: Product-market fit is the moment a product clearly solves a real need, so people adopt it and tell others on their own.