Glossary for beginners: AI Women in Law

AI basics Working with Claude Files, folders and your computer GitHub: saving and sharing your work Publishing on the internet Data, accounts and security Workflow redesign

AI basics

Artificial intelligence (AI): Software that does tasks that normally need human judgement, such as reading text, answering questions or sorting documents.
Generative AI: AI that produces new content, such as text, code or images, in response to an instruction. Claude is generative AI.
Large language model (LLM): The type of AI behind tools such as Claude and ChatGPT. It has been trained on very large amounts of text and produces text by predicting what comes next.
Model: The trained AI system itself. Anthropic offers several Claude models; larger models are usually more capable and slower.
Prompt: The instruction or question you type into an AI tool. A clear prompt says what you want, who it is for and what a good answer looks like.
Token: The unit an AI model reads and writes in: a short piece of text, often part of a word. Usage limits and prices are counted in tokens.
Context window: How much text the model can take into account at one time, measured in tokens. Text beyond it is left out, which is one reason a project memory helps.
Hallucination: An answer from an AI tool that is stated confidently but is wrong or invented, such as a case that does not exist. Every output needs checking.
Training data: The text a model learned from before you used it. Some tools also use what you type to train future models unless you switch this off.
Retrieval (RAG): Short for retrieval-augmented generation: the tool first searches a set of documents, then writes its answer using what it found, so the answer can cite its sources.

Working with Claude

Claude: Anthropic's AI assistant. You use it in a web browser at claude.ai or in the Claude desktop app.
Claude Pro: The paid Claude plan the course uses. It includes Claude Code.
Claude Code: Anthropic's tool for building software with Claude. It can read and change the files in a folder on your computer and run commands. On the course you use it in the Code tab of the Claude desktop app.
Agent: An AI tool that can carry out several steps on its own to finish a task, such as reading files, making changes and checking the result, while you approve the important steps.
Session: One continuous conversation with Claude. When you start a new session, Claude does not remember the last one unless it is written down, which is what the memory folder is for.
Standing instructions (CLAUDE.md): A file in your project that Claude Code reads at the start of every session. It sets out how Claude should work with you, for example to plan first and to ask before deleting anything.
Memory folder: The folder of notes in your project where Claude records the problem, the redesign, each decision, its sources and a log of each session, so it does not forget or guess.
Plan mode: A way of working where Claude writes out what it intends to change, and you agree the plan, before it changes any file.
Permission mode: The setting that decides how much Claude Code may do without asking. On the course you use Manual, so Claude asks before each change.
Slash command: A shortcut you type in Claude Code that starts with /, such as /end-session, which runs a set routine.
Connector (MCP): A link that lets Claude use another app, such as Notion. MCP, the Model Context Protocol, is the standard these links are built on.

Files, folders and your computer

File: A single document saved on your computer, such as a Word document or a text file.
Folder: A container for files. Your course project is one folder, called my-first-tool.
Project folder: The folder that holds everything for one tool: its code, the standing instructions and the memory folder.
File extension: The letters after the dot in a file name, which show the type of file: .docx for Word, .md for Markdown, .html for a web page.
Markdown (.md): A plain text format that uses simple marks for headings and lists, such as # for a heading. CLAUDE.md and the memory notes are Markdown files.
Hidden file or folder: A file or folder your computer does not show by default, usually one whose name starts with a dot, such as .claude or .gitignore.
Terminal: A window where you type commands for the computer instead of clicking. On Windows it is called PowerShell; on a Mac it is called Terminal.
PowerShell: The terminal on Windows. To open it, press the Windows key, type PowerShell and click Windows PowerShell.
Command: An instruction typed into a terminal, such as gh auth login. Claude tells you exactly what to type when one is needed.
Install: Put a program on your computer so you can use it. Some work laptops do not allow this; use a personal computer if so.
Download: Copy a file from the internet to your computer, usually into the Downloads folder.
Zip file: One file that holds several files in compressed form. You extract (unzip) it to get the files out.

GitHub: saving and sharing your work

GitHub: A website that stores projects and keeps every saved version, so you can see what changed and go back. The course saves each participant's tool there.
Repository (repo): A project folder stored on GitHub, with its full history. Your tool lives in a repository, which can be private or public.
Private and public repository: A private repository can be seen only by you and people you invite. A public repository can be seen by anyone. Directory entries link to public ones.
Git: The program that records versions of your files. GitHub is a website built around it.
Commit: A saved version of your project with a short message saying what changed.
Push: Send your saved versions from your computer to GitHub.
Clone: Copy a repository from GitHub onto your computer.
Branch: A separate line of changes in a repository, so you can try something without affecting the main version.
.gitignore: A file that lists which files should not be saved to GitHub, such as files holding passwords or keys.
README: The front page of a repository: a file that explains what the project is and how to use it.
Licence: The terms that say how others may use your code. Open-source licences such as MIT allow reuse; with no licence, others may not reuse it.
Open source: Software whose code is public and licensed so that anyone may use, change and share it.
Two-factor authentication (2FA): A second check when you sign in, usually a code from an authenticator app on your phone. GitHub requires it.

Publishing on the internet

Website and web app: A website mainly shows information. A web app lets people do something, such as fill in a form and get a result.
Front end: The part of a tool people see and click in the browser.
Back end: The part that runs on a server, out of sight, such as saving form answers to a database.
Server: A computer that runs all the time and serves your tool to anyone who visits.
Deploy: Put your tool on the internet so people can open it with a link.
Vercel: A service that deploys websites and web apps. The course uses its free Hobby plan, which is for personal, non-commercial projects.
URL: A web address, such as https://aiwomeninlaw.com/course/.
Domain: The name part of a web address, such as aiwomeninlaw.com. You buy it from a domain registrar.
HTML, CSS and JavaScript: The three languages of web pages: HTML for content, CSS for appearance, JavaScript for what happens when you click.
Python: A programming language often used for data and AI work. Claude writes it for you if your tool needs it.
Browser: The program you use to open websites, such as Chrome, Edge or Safari.

Data, accounts and security

Database: Organised storage for information a tool collects, such as form answers, so it can be found and updated later.
Sign-in (authentication): Checking who someone is before letting them in, for example with Google or an emailed link.
API: A way for one program to talk to another, for example a tool sending a request to Claude and getting an answer back.
API key: A secret code that lets a program use a paid service such as Claude through its API. Treat it like a password: do not paste it into chats or save it to GitHub.
Environment variable: A setting stored outside your code, often used to hold secrets such as API keys, so they are not saved with the code.
Personal data: Information about an identifiable person. Privacy laws limit how it may be collected, used and shared.
Confidential information: Information you have a duty to keep private, such as client matters. It stays out of public AI tools unless your organisation has approved the tool for it.
Fictional data: Invented names, matters and details used to build and test a tool without using real client information.
Enterprise version: A paid business version of an AI tool, usually with contract terms that it will not train on your inputs and with stronger data controls.

Workflow redesign

Workflow: The sequence of steps, people and handovers that turns a request into finished work, such as how a contract request is received, reviewed and answered.
Process map: A diagram of a workflow, step by step, showing who does what and where work waits.
As-is and to-be: The as-is map shows how the work runs today. The to-be map shows how it should run after the redesign.
Handover: A point where work passes from one person or team to another. Handovers are often where delays happen.
Bottleneck: The step where work piles up and waits, which sets how fast the whole workflow can go.
Baseline: A measure of how the workflow performs before you change it, such as average turnaround time, so you can show what changed afterwards.
Use case: A specific task a tool will do for specific people, such as "check incoming NDAs against our standard positions for the commercial team".
First version (MVP): The smallest working version of a tool that does the main job well enough to test with real users. MVP stands for minimum viable product.
Prototype: An early version built to test an idea, which may be thrown away.
Human in the loop: A step where a person reviews or approves what the tool produced before it is used.
BPMN: Business Process Model and Notation: a standard set of shapes for drawing process maps, so anyone can read them.