AI Mayak Academy Glossary: 2026

300+ terms: AI, Claude Code, programming, business, running things in production

How to use it: press Cmd+F (or Ctrl+F) and type any term. Each term links to the course lessons where it's covered in detail.

Where a beginner should start. Look up these words first: Prompt (your request to the AI), LLM (large language model), Token, Context window, Hallucination, Agent, Claude, ChatGPT. You'll need the rest later.

Terms are listed alphabetically. Coverage: AI basics, Claude Code, programming, monetization, infrastructure, security.

2FA / Two-Factor Authentication

In plain words: Two-factor authentication: signing in takes two independent confirmations.
Definition: Login requires two independent factors (password + SMS code/app/key).
Picture this: A door with two locks: one key won't open it.
In the course: Permissions and security, Security and secrets

A/B test

In plain words: Compare two versions of a page or email and see which one works better.
Definition: Comparing two versions (A and B) to see which performs better on a metric.
Picture this: Two shop windows side by side: which one draws more people?
In the course: Product analytics, AI advertising

AAA / Authentication, Authorization, Accounting

In plain words: The security trio: who you are, what you're allowed to do, what you did.
Definition: Security framework: who you are, what you can do, what you did.
Picture this: Guard at the entrance → floor pass → visitor log.
In the course: Permissions and security, Security and secrets

AEO / Answer Engine Optimization

In plain words: Optimizing so AI assistants quote your content in their answers.
Definition: Optimizing content to appear in AI answer engines (ChatGPT, Perplexity, Claude).
Picture this: SEO for an era when search became a conversation instead of a list of links.
In the course: The SEO machine

Agent

In plain words: AI that acts on its own: it plans, calls tools, adapts.
Definition: An AI system that takes actions autonomously toward a goal: plans, calls tools, adapts.
Picture this: An employee you give a goal to, not step-by-step instructions.
In the course: The agentic AI market, Subagents, Agent teams, Multi-agent orchestration

Agent SDK

In plain words: Anthropic's library for building agents in your own code.
Definition: Anthropic library for building Claude-powered agents in code.
Picture this: A LEGO set for assembling your own AI helpers.
In the course: Subagents, Hardware AI

Agent teams

In plain words: A team of agents works on a task, each with its own role.
Definition: Multiple agents collaborating on a task with defined roles.
Picture this: A construction site: architect + foreman + workers, not one jack-of-all-trades.
In the course: Agent teams, Multi-agent orchestration

AGI / Artificial General Intelligence

In plain words: Hypothetical AI that matches humans at every task.
Definition: Hypothetical AI matching human cognitive ability across all tasks.
Picture this: It doesn't exist yet: a horizon, not a real product.
In the course: AI Roadmap 2027-2030

AI Alignment

In plain words: The field working to make AI do what we actually want, not just the letter of the task.
Definition: Research field ensuring AI systems act according to human values.
Picture this: A genie grants your wish literally, when you needed it granted in spirit.
In the course: AI Ethics & Safety

API key rotation

In plain words: Replacing API keys regularly, in case one leaks.
Definition: Periodically replacing API keys to limit damage from leaks.
Picture this: Changing your apartment lock every six months instead of waiting for the key to be stolen.
In the course: API keys and .env, Security and secrets

AI Safety

In plain words: Keeping AI from causing harm, on purpose or by accident.
Definition: Field focused on preventing harmful AI behavior.
Picture this: Traffic laws for AI: rules of the road so there are no crashes.
In the course: AI Ethics & Safety, AI Roadmap 2027-2030

Alert

In plain words: An automatic notification about a problem or a limit being exceeded.
Definition: Automated notification of a system issue or threshold breach.
Picture this: A warning light on the dashboard: something needs your attention.
In the course: Managing agents, Production Observability

ANI / Artificial Narrow Intelligence

In plain words: AI built for a single task: chess, translation, recognition.
Definition: AI specialized for one task (chess, image recognition, translation).
Picture this: A hammer: great for nails, useless as a saw.
In the course: AI Roadmap 2027-2030

Anchoring

In plain words: The first piece of information sets the frame for every judgment that follows.
Definition: Cognitive bias where the first information disproportionately influences a decision; in prompts: setting context that biases the output.
Picture this: A ship's anchor: it decides where the ship stays.
In the course: Prompting fundamentals

Anthropic

In plain words: The makers of Claude. Their focus is safe AI.
Definition: The company building Claude, founded in 2021, focused on AI safety.
Picture this: The manufacturer of your main tool.
In the course: Default Shift

API / Application Programming Interface

In plain words: The way programs talk to each other: a standard "language."
Definition: A contract that lets two programs talk to each other.
Picture this: A wall outlet: any device's plug fits if it meets the standard.
In the course: API keys and .env, APIs and integrations

API key

In plain words: A secret password for accessing an API.
Definition: A secret token authorizing API requests; like a password.
Picture this: The key to the staff entrance: whoever finds it gets in, and you get the bill.
In the course: API keys and .env, Security and secrets

ARR / Annual Recurring Revenue

In plain words: Yearly recurring revenue from subscriptions.
Definition: Predictable yearly subscription revenue (MRR × 12).
Picture this: A harvest you bring in every year without replanting.
In the course: First clients, Unit Economics

async / asynchronous

In plain words: The code doesn't wait for a response: it starts something and moves on.
Definition: Code that runs without blocking: start a task, do other things, get the result later.
Picture this: You put the kettle on, and while it boils you slice the bread.
In the course: APIs and integrations

Attention mechanism

In plain words: The "attention" mechanism: the model figures out which parts of the input matter.
Definition: Transformer technique to weigh which input tokens matter most for each output token.
Picture this: Reading a letter, you pick out the key phrases and the rest becomes background.
In the course: How an LLM works

Attribution

In plain words: Where a customer came from: which channel worked.
Definition: Determining which marketing touchpoint led to a conversion.
Picture this: A detective's investigation: trace the path back to the moment of purchase.
In the course: AI advertising

Audit log

In plain words: A log of actions: entries only get added, nothing is ever deleted.
Definition: Append-only record of significant actions for security and compliance.
Picture this: The sign-in book at the front desk: who came and went, and when; no pages get torn out.
In the course: Permissions and security, Managing agents, Production Observability

Authentication

In plain words: Proving "I am who I say I am."
Definition: Verifying who you are: login, biometrics, key.
Picture this: Showing your passport at the border.
In the course: Permissions and security, Security and secrets

Authorization

In plain words: What you're allowed to do once you're signed in.
Definition: Determining what you're allowed to do once authenticated.
Picture this: A visa in your passport: you can enter the country, but not every area.
In the course: Permissions and security, Security and secrets

AWS / Amazon Web Services

In plain words: Amazon's cloud platform: servers, databases and AI, all for rent.
Definition: Amazon's cloud computing platform, the largest in the world.
Picture this: The world's biggest warehouse of servers, rented by the hour.
In the course: Deployment

B2B / Business-to-Business

In plain words: You sell to companies, not to individuals.
Definition: Selling to other companies rather than consumers.
Picture this: A flour supplier selling to the bakery, not to the people who buy the rolls.
In the course: First clients, Monetization case studies

B2C / Business-to-Consumer

In plain words: You sell directly to individuals.
Definition: Selling directly to end consumers.
Picture this: A coffee shop for people walking by.
In the course: First clients

Backend

In plain words: The "behind the scenes" part of an app: business logic, databases.
Definition: Server-side code: business logic, databases, APIs hidden from the user.
Picture this: A restaurant kitchen: the customer doesn't see it, but without it there's no meal.
In the course: Websites and web apps, MCP

Backpropagation

In plain words: A way to train neural networks: the error travels "backward" and the weights get adjusted.
Definition: Algorithm for training neural networks by adjusting weights backward from the error.
Picture this: A coach after the game: "here's where you went wrong, fix that move."
In the course: How an LLM works

BANT / Budget Authority Need Timeline

In plain words: Four criteria for qualifying a lead: budget, decision-making authority, need, timeline.
Definition: Sales qualification framework: does the prospect have all four?
Picture this: A checklist for "is this client worth my time?"
In the course: Cold outreach, Sales AI

Batch API

In plain words: Batch processing: 50% off, with answers within 24 hours.
Definition: Process many requests at once at a 50% discount, with results within 24h.
Picture this: A laundromat: one load for the whole batch vs. washing each shirt separately.
In the course: Prompt Caching and the Batch API, Cost Engineering

bcrypt

In plain words: A password hashing algorithm that's deliberately slow.
Definition: Slow hash function for passwords; resists brute force.
Picture this: A lock where even trying every key takes years.
In the course: Security and secrets

Benchmark

In plain words: A standard test for comparing models.
Definition: Standardized test comparing AI models on common tasks (MMLU, HumanEval).
Picture this: The Presidential Fitness Test: the same exercises for everyone.
In the course: Comparing models

Bias

In plain words: A systematic error in a model caused by skewed training data.
Definition: Systematic error from skewed training data; AI reproduces societal biases.
Picture this: A teacher who learned from a single textbook and doesn't know any other points of view.
In the course: AI Ethics & Safety

Brand voice

In plain words: A brand's recognizable tone: it sounds like one person.
Definition: Consistent personality and tone in all brand communications.
Picture this: A friend's voice: you recognize it after a few words.
In the course: AI copywriting, Personal Brand AI

Breach

In plain words: A leak: someone got access to data without permission.
Definition: Security incident where data is accessed without authorization.
Picture this: A hole in the fence, and the cow got out.
In the course: Permissions and security, Security and secrets

CAC / Customer Acquisition Cost

In plain words: How much you spent to get one paying customer.
Definition: Money spent acquiring one paying customer (marketing + sales).
Picture this: The cost of the rod and bait it took to catch one fish.
In the course: First clients, Unit Economics

CAC payback

In plain words: How many months it takes a customer to pay back what you spent acquiring them.
Definition: Months for a new customer's revenue to recoup the acquisition cost.
Picture this: When the fishing rod pays for itself in fish.
In the course: Unit Economics

Cache

In plain words: Storage for recent results, so a repeat answer comes back faster.
Definition: Storage of recent results to avoid recomputing: a speed boost.
Picture this: A cheat sheet so you don't have to calculate the same thing twice.
In the course: Context Rot, Prompt Caching and the Batch API

Canary deploy

In plain words: First 5% of users, you watch, then everyone.
Definition: Roll out a new version to a small % of users first, monitor, then expand.
Picture this: The canary in the coal mine: if it's alive, everyone can go in.
In the course: Deployment, Production Observability

CDN / Content Delivery Network

In plain words: A network of servers around the world that serves your site from close to the visitor.
Definition: Network of servers worldwide caching content close to users.
Picture this: Amazon warehouses in every city: faster delivery.
In the course: Deployment

Chain-of-thought / CoT

In plain words: Asking the model to think out loud, step by step.
Definition: Prompting the model to reason step by step before the final answer.
Picture this: A student solving a word problem: first "given," then "solution," then "answer."
In the course: Prompting fundamentals, Extended Thinking

ChatGPT

In plain words: OpenAI's AI assistant that kicked off the mass wave.
Definition: OpenAI's consumer AI assistant, launched in 2022, which popularized LLMs.
Picture this: AI's iPhone moment: there was a before and an after.
In the course: Default Shift, Comparing models

Churn

In plain words: The percentage of customers who dropped the product during a period.
Definition: % of customers canceling per period.
Picture this: A hole in the bucket: you pour water in and it leaks out.
In the course: Delivery and Retention, Unit Economics

CI/CD / Continuous Integration / Continuous Deployment

In plain words: An automatic assembly line from commit to production.
Definition: Automated pipeline: code commit → tests → build → deploy.
Picture this: A factory assembly line: you put the part on and the rest happens on its own.
In the course: Headless and CI/CD

Classification

In plain words: The classification task: sorting things onto shelves.
Definition: ML task: assign input to one of a fixed set of categories.
Picture this: A mail sorter: work / personal / spam.
In the course: How an LLM works

Claude

In plain words: Anthropic's AI models. The family: Haiku (fast), Sonnet (balanced), Opus (powerful), Fable (the most capable, for complex tasks). The lineup changes: see What's current.
Definition: Anthropic's family of AI models: Haiku, Sonnet, Opus, Fable (current lineup: see the What's current page).
Picture this: Toyota's lineup: Yaris, Camry, Land Cruiser.
In the course: Comparing models, Default Shift

Claude Code

In plain words: Anthropic's AI agent for programming: Claude that can see and change your files. It runs in the terminal, VS Code, JetBrains, the Claude app, the browser and on your phone; it's included in the paid plans.
Definition: Anthropic's agentic coding tool: Claude with file system access; runs in the terminal, VS Code, JetBrains, the Claude desktop app, the browser and on the phone.
Picture this: Not a chat, but an employee sitting at your computer.
In the course: Default Shift, Installation and setup

CLAUDE.md

In plain words: An instruction file for Claude in your project; it reads it every time.
Definition: Project file Claude Code auto-reads as context: coding rules, structure, conventions.
Picture this: A note for a new employee: "here's how we do things."
In the course: CLAUDE.md

claude-mem

In plain words: Long-term memory for Claude between sessions.
Definition: Long-term memory tool for Claude that persists across sessions.
Picture this: A notebook the AI doesn't lose.
In the course: Obsidian

.claude folder

In plain words: A hidden project folder: agents, commands, hooks, skills.
Definition: Hidden folder storing agents, commands, hooks and skills per project.
Picture this: A tool closet right next to your workstation.
In the course: CLAUDE.md, The WAT framework

.claudeignore

In plain words: A list of paths Claude doesn't read: saves tokens.
Definition: File listing paths Claude Code should ignore (like .gitignore).
Picture this: A "keep out" sign on the door.
In the course: Context Rot

/clear

In plain words: A command that wipes the conversation context in Claude Code.
Definition: Slash command resetting the Claude Code conversation context.
Picture this: Wipe the whiteboard and start fresh.
In the course: Built-in commands, Context Rot

CLI / Command Line Interface

In plain words: The terminal: you type commands instead of clicking.
Definition: Text-based interface for typing commands instead of clicking.
Picture this: A typewriter vs. a drawing tablet: more austere, but faster.
In the course: Installation and setup, Claude Code CLI

Closing

In plain words: The final stage of a sale: closing the deal.
Definition: Final stage of sales: getting the prospect to commit and sign.
Picture this: The last hammer blow on a nail: it's in and it holds.
In the course: Closing Techniques

Cloud

In plain words: Someone else's servers, rented over the internet.
Definition: Computing resources rented from a provider over the internet.
Picture this: Electricity from the outlet: you don't build your own power plant.
In the course: Deployment

Cloudflare

In plain words: A platform at the edge of the network: hosting, protection, serverless functions.
Definition: Edge platform for hosting, security, CDN and Workers (serverless).
Picture this: The Swiss Army knife of infrastructure: many tools in one.
In the course: Deployment

Clustering

In plain words: Grouping similar items without categories defined in advance.
Definition: ML task: group similar items without predefined categories.
Picture this: Sorting a pile of nuts by size just by looking, without a ruler.
In the course: How an LLM works

CMS / Content Management System

In plain words: A system for managing a website's content.
Definition: Software for creating and managing digital content (WordPress, Notion CMS).
Picture this: Store shelves: you change the products, the shelves stay the same.
In the course: AI copywriting, The SEO machine

Cold outreach

In plain words: Cold contact: you write to strangers with an offer.
Definition: Reaching out to prospects with no prior relationship.
Picture this: Knocking on strangers' doors with an offer.
In the course: Cold Outreach, Cold outreach

/compact

In plain words: A command that compresses the conversation history while keeping the essentials.
Definition: Slash command compressing the Claude Code conversation history.
Picture this: Lecture notes: fewer words, same ideas.
In the course: Built-in commands, Context Rot

Completion

In plain words: What an LLM generates in response to a prompt.
Definition: Model output finishing the prompt: what the LLM generates.
Picture this: "Complete the sentence," the school exercise.
In the course: How an LLM works

Computer Use

In plain words: Claude controls a computer: sees the screen, clicks, types.
Definition: Claude's ability to control a computer via screenshots and clicks.
Picture this: A remote intern sitting at your computer.
In the course: Computer Use, Computer Use Advanced

Computer use tier

In plain words: The Anthropic plan level needed for Computer Use (as of October 2026 the feature is included in paid plans).
Definition: Anthropic plan level required for the Computer Use feature (included in paid plans as of October 2026).
Picture this: A driver's license class: to drive a big rig you need a CDL, not a regular license.
In the course: Computer Use, Plans and access levels

Container

In plain words: An isolated package with code and its dependencies.
Definition: Isolated package with code, dependencies and runtime that runs anywhere.
Picture this: A shipping container: anything can be inside, and the ships are all the same.
In the course: Deployment

/context

In plain words: A command that shows what's currently in Claude Code's context.
Definition: Slash command showing the current Claude Code context contents.
Picture this: "Show me what's in your head right now."
In the course: Built-in commands

Context window

In plain words: How many tokens the model can see at once: its "working memory."
Definition: Max tokens an LLM can read at once (input + output).
Picture this: The size of your desk: whatever doesn't fit doesn't get used.
In the course: Tokens and context, Context Rot

Conversion

In plain words: A visitor took the action you wanted: signed up, bought.
Definition: Visitor performing a desired action (signup, purchase, demo).
Picture this: A passerby became a customer.
In the course: First clients, Funnels for AI products

Conversion rate

In plain words: The percentage of visitors who took the target action.
Definition: % of visitors who convert.
Picture this: Out of 100 passersby, how many walked into the store.
In the course: Funnels for AI products, Product analytics

Convolutional / CNN

In plain words: Convolutional networks: for images, they look for local patterns.
Definition: Neural network type specialized for images: filters scan local patterns.
Picture this: A magnifying glass sliding across a picture, noting what it sees.
In the course: How an LLM works

Copilot

In plain words: GitHub's AI assistant, right in your code editor.
Definition: GitHub's AI coding assistant in the IDE.
Picture this: A copilot: doesn't make the decisions, but gives you pointers.
In the course: Default Shift, The AI tools market

Copy

In plain words: Marketing and sales text aimed at getting someone to act.
Definition: Marketing/sales text aimed at conversion.
Picture this: Not literature, but persuasion with a purpose.
In the course: AI copywriting

CPC / Cost Per Click

In plain words: The average cost of one ad click.
Definition: Average ad cost per click.
Picture this: The price of getting a person to look at your shop window.
In the course: AI advertising

Cron

In plain words: A scheduler that runs tasks on a timetable.
Definition: Unix scheduler running jobs at time intervals.
Picture this: An alarm clock for scripts: "run this every day at 9 a.m."
In the course: Loop and Scheduled Tasks

CRM / Customer Relationship Management

In plain words: A system for tracking customers, deals and communications.
Definition: Software tracking customers, deals and communications (Salesforce, HubSpot).
Picture this: A salesperson's address book on steroids.
In the course: CRM on autopilot, Sales AI

CSRF / Cross-Site Request Forgery

In plain words: An attack that exploits a user's active session from another website.
Definition: Attack tricking an authenticated user's browser into an unwanted action.
Picture this: Someone slipped a stranger's order under your signature while you weren't looking.
In the course: Security and secrets, Hook-Deny-By-Design

CTR / Click-Through Rate

In plain words: The share of ad impressions that led to a click.
Definition: % of impressions resulting in clicks.
Picture this: Out of 100 people walking past the shop window, how many stopped.
In the course: AI advertising

Cursor

In plain words: An AI-first code editor, a competitor to VS Code + Copilot.
Definition: AI-first code editor competing with VS Code + Copilot.
Picture this: VS Code as if it had been built around AI from scratch.
In the course: Default Shift, The AI tools market

Customer success

In plain words: The team that makes sure customers get value and don't leave.
Definition: Team helping customers get value, reducing churn.
Picture this: A trainer who stays with the client for the whole program instead of disappearing after payment.
In the course: Delivery and Retention

Custom command

In plain words: Your own slash command: a shortcut to your own routine. Commands have now been merged with skills (.claude/skills/<name>/SKILL.md); old files in .claude/commands/ keep working.
Definition: User-defined slash command for Claude Code (.claude/commands/X.md); now merged with skills, old files keep working.
Picture this: A hotkey for an action you use a lot.
In the course: Built-in commands

Database

In plain words: Where an application keeps its data, in an organized form.
Definition: Structured storage for application data: SQL or NoSQL.
Picture this: An archive with cabinets and shelves, not a pile on the floor.
In the course: Websites and web apps, APIs and integrations

Deck

In plain words: A slide presentation: for sales, investors, training.
Definition: Presentation slides for sales, investors or training.
Picture this: A comic-book version of a longer story.
In the course: AI presentations

Decoder

In plain words: The part of a transformer that produces the answer token by token.
Definition: Transformer part generating output tokens one by one.
Picture this: A simultaneous interpreter: word by word, following the meaning.
In the course: How an LLM works

DeepSeek

In plain words: A Chinese AI lab. Its current lineup is DeepSeek V4 with open weights; the service's data is stored in China.
Definition: Chinese AI lab; current line is DeepSeek V4 with open weights (MIT license).
Picture this: A quality niche manufacturer, underrated in the West.
In the course: Comparing models, Local models

Deep learning

In plain words: Machine learning with many-layered networks: the foundation of modern AI.
Definition: ML with multi-layer neural networks; the basis of modern AI.
Picture this: An assembly line with dozens of processing stations.
In the course: How an LLM works

Demo

In plain words: Showing the product to a potential customer.
Definition: Live or recorded product demonstration for prospects.
Picture this: A test drive.
In the course: Cold outreach, Sales AI

Dependency

In plain words: An outside library your code depends on.
Definition: External library/package your code uses.
Picture this: Parts from another factory: without them the car won't run.
In the course: API keys and .env

Deploy / Deployment

In plain words: Rolling code out to the live environment.
Definition: Pushing code to the production environment.
Picture this: The play is finally on stage, not in rehearsal.
In the course: Deployment

Discovery call

In plain words: The first call: figuring out what the client wants.
Definition: First call with a prospect to understand their needs.
Picture this: A doctor asking questions before making a diagnosis.
In the course: Cold outreach, Sales AI

Distributed system

In plain words: A system made of many machines that coordinate with each other.
Definition: System where components run on multiple machines, coordinating.
Picture this: A symphony orchestra vs. a solo pianist.
In the course: Deployment, Production Observability

Docker

In plain words: The most popular container platform.
Definition: Most popular container platform.
Picture this: The standard shipping container, but for software.
In the course: Deployment

Docker compose

In plain words: Tying several containers together with one config file.
Definition: Tool defining multi-container apps in one YAML file.
Picture this: A recipe's ingredient list: one file with everything you need.
In the course: Deployment

Downtime

In plain words: The time when a service isn't working.
Definition: Period when a service is unavailable.
Picture this: The store is closed, and customers go to the competition.
In the course: Production Observability

Dwell time

In plain words: How long a visitor spent on a page.
Definition: How long a visitor stays on a page; an SEO signal.
Picture this: The guest stayed for coffee, so they must be interested.
In the course: The SEO machine

Edge

In plain words: Computing close to the user, at the "edge" of the network.
Definition: Computing close to the user (Cloudflare, AWS Lambda@Edge).
Picture this: A coffee shop on your block, not out of town.
In the course: Deployment

ElevenLabs

In plain words: One of the leading TTS platforms: voice generation, voice cloning.
Definition: Leading voice synthesis platform: text-to-speech, voice cloning.
Picture this: A dubbing studio in one click.
In the course: Voice tools, Voiceover

Embedding

In plain words: A numeric representation of meaning: similar texts produce similar vectors.
Definition: Numeric vector representing the meaning of text/images; similar things have similar vectors.
Picture this: Coordinates on a map of meanings: "king" and "queen" sit close together.
In the course: RAG, How an LLM works

Emergent capabilities

In plain words: Abilities that show up in large models on their own, without anyone teaching them.
Definition: Skills that appear in large models without explicit training.
Picture this: A child starts making jokes, though nobody ever explained jokes to them.
In the course: The evolution of AI, AI Roadmap 2027-2030

Encoder

In plain words: The part of a transformer that turns input into an internal representation.
Definition: Transformer part converting input into an internal representation.
Picture this: A court stenographer turning speech into records for the archive.
In the course: How an LLM works

Encryption at rest

In plain words: Encrypting data while it's stored, on disk.
Definition: Data encrypted when stored on disk.
Picture this: A safe for your papers when you leave the office.
In the course: Permissions and security, Security and secrets

Encryption in transit

In plain words: Encrypting data while it travels over the network.
Definition: Data encrypted while moving over the network (TLS).
Picture this: An armored truck for the cash, not walking it over in a bag.
In the course: Permissions and security, Security and secrets

Endpoint

In plain words: A specific API URL you send requests to.
Definition: Specific URL where an API service receives requests.
Picture this: The exact window at the DMV for one particular service.
In the course: API keys and .env, APIs and integrations

Engagement

In plain words: How involved your audience is: likes, comments, shares.
Definition: Audience interaction with content (likes, comments, shares).
Picture this: The audience reacts instead of sitting in silence.
In the course: Social media automation, Personal Brand AI

Environment / dev/staging/prod

In plain words: Environments: dev (development), staging (testing), prod (real customers).
Definition: Separate copies of a system for development, testing and real users.
Picture this: Garage → test track → highway.
In the course: Deployment

Environment variable

In plain words: A value set in the environment rather than in the code; often a secret.
Definition: Configuration value set outside the code (often secrets).
Picture this: The key isn't hanging on the door, it's in your pocket.
In the course: API keys and .env

Error budget

In plain words: The allowed amount of downtime: a balance between stability and speed.
Definition: Allowed downtime per period: a balance of reliability vs. speed of change.
Picture this: "We can afford to be down an hour a month," and no more.
In the course: Production Observability

Evaluation / eval

In plain words: A systematic check of the quality of a model's answers.
Definition: Systematic testing of model output quality.
Picture this: The SAT for a model: a standard set of questions and scores.
In the course: Evals

eval.json

In plain words: A skill's file of test cases for checking it.
Definition: Anthropic Skill file defining test cases for skill validation.
Picture this: A sheet of test assignments: will the intern pass?
In the course: What Skills are, Evals

Extended thinking

In plain words: A mode where Claude thinks longer before answering hard problems. In the newest models it's been replaced by adaptive thinking: the model decides how much to think, and you set the effort level.
Definition: Claude mode using more compute on harder problems before answering; newer models use adaptive thinking steered by effort.
Picture this: A chess player thinking hard before a move instead of grabbing the first one that comes to mind.
In the course: Extended Thinking

fal.ai

In plain words: A hosting platform for image and video AI models.
Definition: Hosting platform for image/video AI models.
Picture this: Turnkey GPU model hosting.
In the course: Image generation, AI video generation

Feature flag

In plain words: A switch for a feature that doesn't require redeploying.
Definition: Switch turning features on/off without redeploying.
Picture this: A light switch on the wall: the lamp is already wired in.
In the course: Deployment, Production Observability

Few-shot

In plain words: A prompt with a few examples before the task.
Definition: Prompting with several examples before the task.
Picture this: "Like this, like this and like this. Now you try."
In the course: Prompting fundamentals

Fine-tuning

In plain words: Extra training of a ready-made model on specialized data.
Definition: Additional training of a pretrained model on specialized data.
Picture this: A continuing-education course for someone who's already a trained professional.
In the course: Fine-tuning

Flux

In plain words: Image generation models from Black Forest Labs (for the current model family, check their website).
Definition: Black Forest Labs image generation models (current family: see their website).
Picture this: A Midjourney competitor, with some models available as open weights.
In the course: Image generation

Freemium

In plain words: A free basic version + paid advanced features.
Definition: Free basic tier + paid upgrades for power features.
Picture this: The free demo laptop: more memory costs extra.
In the course: Pricing, Packaging

Frontend

In plain words: The part of an app the user sees.
Definition: User-facing code: UI, browser logic.
Picture this: The shop window.
In the course: Websites and web apps

Function

In plain words: A reusable block of code with one job.
Definition: Reusable block of code performing one task.
Picture this: One tool in a set: for one job.
In the course: APIs and integrations

Function calling

In plain words: The LLM calls your functions with the right parameters.
Definition: An LLM's ability to call functions you define with structured arguments.
Picture this: An employee calls the right number with the right request.
In the course: APIs and integrations, MCP

Funnel

In plain words: The funnel that turns a visitor into a customer.
Definition: Sequence of steps converting visitor → customer.
Picture this: A funnel that gradually narrows: lots at the top, few at the bottom.
In the course: Funnels for AI products

GAN / Generative Adversarial Network

In plain words: Two networks: one creates, the other critiques, and the competition makes both better.
Definition: Two networks: a generator creates, a discriminator critiques; the competition improves quality.
Picture this: An artist and a critic in one loop: both get better.
In the course: How an LLM works

GDPR

In plain words: The European law on personal data.
Definition: EU privacy law: consent, right to deletion, breach notification.
Picture this: The rules of the road for handling data in the EU.
In the course: AI Regulation & Compliance

Gemini

In plain words: Google's family of AI models, which are multimodal.
Definition: Google's flagship multimodal AI family.
Picture this: Google's answer to ChatGPT/Claude.
In the course: Comparing models

Generative

In plain words: Generative AI creates something new instead of classifying what already exists.
Definition: AI that creates new content: text, images, audio, video.
Picture this: An artist vs. an archivist: one creates, the other files things away.
In the course: The evolution of AI

Git

In plain words: A version control system: every change gets recorded.
Definition: Version control system tracking code changes.
Picture this: A time machine for code: roll back to any moment.
In the course: GitHub Worktrees

GitHub

In plain words: The biggest host for Git repositories, plus social features.
Definition: Largest hosting service for Git repositories, with collaboration features.
Picture this: A social network for developers and a code warehouse in one.
In the course: GitHub Worktrees, Headless and CI/CD

GitHub Actions

In plain words: Automatic builds and tests right inside GitHub.
Definition: CI/CD automation built into GitHub.
Picture this: An assembly line inside the programmers' social network.
In the course: Headless and CI/CD

GitOps

In plain words: An approach where Git is the source of truth about configuration.
Definition: Operations workflow where Git is the source of truth for infrastructure.
Picture this: The house blueprint lives in Git: whatever's in Git is what gets built.
In the course: Headless and CI/CD, Production Observability

GPU / Graphics Processing Unit

In plain words: A chip for parallel computing: the backbone of AI.
Definition: Chip optimized for parallel math; the backbone of AI training and inference.
Picture this: An army of workers doing one task vs. one generalist.
In the course: Hardware AI, Local models

Gradient descent

In plain words: An algorithm that finds the minimum error by taking steps "downhill."
Definition: Algorithm finding the minimum of an error function by going downhill step by step.
Picture this: Coming down a mountain in fog, feeling with your foot for which way is lower.
In the course: How an LLM works

Gross margin

In plain words: The share of revenue left after direct costs.
Definition: Revenue minus direct costs, as a % of revenue.
Picture this: Out of a $100 sale, $60 is yours after ingredients.
In the course: Unit Economics

Ground truth

In plain words: The reference correct answer used to measure accuracy.
Definition: Verified correct answer; the reference for measuring model accuracy.
Picture this: The official reference meter kept at the bureau of standards.
In the course: Evals

GTM / Go-To-Market

In plain words: The strategy for bringing a product to market.
Definition: Strategy for launching a product to market: channel, audience, message.
Picture this: A flight plan from takeoff to first revenue.
In the course: Packaging

Hallucination

In plain words: An LLM confidently makes up things that don't exist.
Definition: LLM confidently outputs false information.
Picture this: A student at an exam who doesn't know the answer but bluffs beautifully.
In the course: AI Ethics & Safety, RAG

Hash

In plain words: A one-way function: text → a short string.
Definition: One-way function mapping data to a fixed-size string.
Picture this: Ground beef: you can't turn it back into a steak.
In the course: Security and secrets

Headless

In plain words: No interface: runs automatically on a server.
Definition: Running without a UI: automated, server-side.
Picture this: A driverless car: controlled by commands.
In the course: Headless and CI/CD

Hook

In plain words: A script that fires on Claude Code events.
Definition: Claude Code script triggered on events (PreToolUse, PostToolUse, SessionStart).
Picture this: A motion sensor in a hallway: something happens and an action kicks off.
In the course: Hooks, Hooks LIVE, Hook-Deny-By-Design

Hook (marketing)

In plain words: The grabber in the first line or first second, so people don't scroll past.
Definition: Opening line/visual that grabs attention in the first seconds.
Picture this: The hook on a fishing line: without it the fish swims away.
In the course: Personal Brand AI

Hosting

In plain words: A service that runs your site or code on servers on the internet.
Definition: Service running your code/site on servers accessible via the internet.
Picture this: A rented spot on a store shelf.
In the course: Deployment

HTTP / Hypertext Transfer Protocol

In plain words: The standard for exchanging data between a browser and a server.
Definition: Standard for web communication between client and server.
Picture this: The postal rules of the internet.
In the course: APIs and integrations

HTTPS

In plain words: HTTP plus encryption: the security standard.
Definition: HTTP with TLS encryption, required for the modern web.
Picture this: The same letter, but in a sealed envelope.
In the course: APIs and integrations, Security and secrets

Hugging Face

In plain words: GitHub for AI models and datasets.
Definition: Hub for open-source ML models and datasets.
Picture this: The world's main warehouse of open-source models.
In the course: Local models

IAM / Identity and Access Management

In plain words: Managing users, roles and permissions in the cloud.
Definition: System managing users, roles and permissions in the cloud.
Picture this: The badge office at a secure facility.
In the course: Permissions and security, Security and secrets

ICP / Ideal Customer Profile

In plain words: A description of your ideal customer: who the product suits best.
Definition: Description of the perfect customer: biggest fit, value, willingness to pay.
Picture this: A portrait of the ideal shopper for your store.
In the course: First clients, Choosing a niche

IDE / Integrated Development Environment

In plain words: A code editor with tools, all in one window.
Definition: Code editor with built-in tools (VS Code, JetBrains, Cursor).
Picture this: A developer's Swiss Army knife.
In the course: Installation and setup

In-context learning

In plain words: The model learns right in the prompt, with no training.
Definition: An LLM learns a task from examples in the prompt, no training.
Picture this: An intern watches and repeats instead of taking a course.
In the course: Prompting fundamentals

Incident

In plain words: An unplanned outage that needs a response.
Definition: Unplanned service disruption requiring a response.
Picture this: A wreck on the highway: it needs to be cleared fast.
In the course: Production Observability

Inference

In plain words: Using a trained model: generating an answer.
Definition: Using a trained model to generate output for a given input.
Picture this: A trained cook following the recipe: not learning, just cooking.
In the course: How an LLM works

Infrastructure as code / IaC

In plain words: Describing infrastructure in code files.
Definition: Defining infrastructure via files (Terraform, Pulumi).
Picture this: The house blueprint in git: construction from it is repeatable.
In the course: Deployment, Production Observability

/init

In plain words: A command that generates CLAUDE.md by analyzing the current folder.
Definition: Claude Code command initializing CLAUDE.md from codebase analysis.
Picture this: "Look around and make a cheat sheet for this project."
In the course: CLAUDE.md

Instruction tuning

In plain words: Extra training so a model follows instructions well.
Definition: Fine-tuning a model to follow instructions well.
Picture this: Teaching an employee not theory, but how to take on assignments.
In the course: How an LLM works

JSON / JavaScript Object Notation

In plain words: The standard data exchange format: text with "key-value" pairs.
Definition: Standard data exchange format: readable text with key-value pairs.
Picture this: A form: each field has a name and a value.
In the course: API keys and .env

JWT / JSON Web Token

In plain words: A signed authorization token carrying data about the user.
Definition: Signed token for authentication carrying user claims.
Picture this: A stamped pass: can't be forged, and the expiration date is visible.
In the course: Permissions and security, Security and secrets

Keyword

In plain words: What users type into search: the foundation of SEO.
Definition: Search term users type; central to SEO.
Picture this: The words on your store sign: how people find you.
In the course: The SEO machine

KPI / Key Performance Indicator

In plain words: Key performance indicators: your main metrics.
Definition: Critical metric tracked for business or product health.
Picture this: A dashboard: a few gauges for what matters most.
In the course: Product analytics

Kubernetes / K8s

In plain words: A container orchestrator that manages containers across a bunch of servers.
Definition: Container orchestrator for managing many containers across servers.
Picture this: The conductor of an orchestra of containers.
In the course: Deployment

KV store

In plain words: A "key-value" database: fast lookups.
Definition: Key-value database: simple, fast lookups by key.
Picture this: A coat check with numbered tickets: give the number, get your coat fast.
In the course: Deployment

Lambda

In plain words: AWS serverless functions: you run code without thinking about servers.
Definition: AWS serverless functions: run code without managing servers.
Picture this: A taxi on demand vs. owning a car.
In the course: Deployment

Landing page

In plain words: One page with one job: getting people to take an action.
Definition: Single focused page converting visitors to action.
Picture this: A shop window with one product front and center.
In the course: Lead Magnets

LangChain

In plain words: A framework for building LLM applications.
Definition: Python/JS framework for building LLM applications.
Picture this: Flat-pack furniture for AI projects.
In the course: Default Shift

LangGraph

In plain words: A LangChain library for stateful multi-agent workflows.
Definition: LangChain library for stateful multi-agent workflows.
Picture this: A state graph for complex AI scenarios.
In the course: Multi-agent orchestration

LangSmith

In plain words: An observability platform for LLM applications.
Definition: Observability platform for LLM applications (LangChain ecosystem).
Picture this: An airplane's black box for an AI service.
In the course: Production Observability

Latency

In plain words: The delay between a request and a response.
Definition: Time between request and response.
Picture this: An echo in the mountains: the pause between your shout and the reply.
In the course: Context Rot, Production Observability

Latent space

In plain words: The model's internal space where meanings live.
Definition: Internal high-dimensional space where a model represents meaning.
Picture this: A map of meanings with millions of coordinates.
In the course: How an LLM works

Lead

In plain words: A potential customer who has shown interest.
Definition: Potential customer who showed interest.
Picture this: A visitor left their phone number at your shop window: that's a lead.
In the course: Lead Generation, Lead Magnets

Lead generation

In plain words: Attracting potential customers.
Definition: Process of attracting potential customers.
Picture this: You set out hooks and wait for something to bite.
In the course: Lead Generation

Lead magnet

In plain words: Free, valuable material in exchange for contact info.
Definition: Free valuable resource exchanged for contact info.
Picture this: A perfume sample: take it, leave your email.
In the course: Lead Magnets

Library

In plain words: A ready-made package of code written by others.
Definition: Reusable code package others wrote.
Picture this: A bookshelf of ready-made recipes.
In the course: API keys and .env

LiteLLM

In plain words: A unified API for 100+ LLM providers.
Definition: Unified API gateway to 100+ LLM providers.
Picture this: A travel adapter that fits every outlet in the world.
In the course: Local models

Llama

In plain words: Meta's family of LLMs with open weights. In 2026, Meta's assistant Meta AI moved to the closed Muse Spark model; previously released Llama models remain available.
Definition: Meta's LLM family with open weights; in 2026 Meta moved its assistant to the closed Muse Spark model, older Llama models remain available.
Picture this: An engine you run locally: install it on your own machine for free.
In the course: Local models

LLM / Large Language Model

In plain words: A large language model: Claude, GPT, Gemini, Llama.
Definition: Massive neural net trained on text that generates language.
Picture this: A giant "book of everything" that answers questions.
In the course: How an LLM works

LM Studio

In plain words: A desktop app for running local models, with a friendly interface.
Definition: Desktop app for running local LLMs with a friendly UI.
Picture this: iTunes for AI models.
In the course: Local models

Load balancer

In plain words: Spreads requests across servers.
Definition: Distributes incoming requests across servers.
Picture this: The person directing the line: "you go to register 1, you go to register 3."
In the course: Deployment

Log

In plain words: A record of events for debugging and auditing.
Definition: Record of events for debugging and audit.
Picture this: An airplane's flight log.
In the course: Managing agents, Production Observability

/loop

In plain words: A command that runs a task repeatedly on a schedule.
Definition: Claude Code skill running a task on a recurring interval.
Picture this: An alarm clock for an AI task every N minutes.
In the course: Loop and Scheduled Tasks

Loss function

In plain words: A function measuring how bad the model's answer is; training minimizes it.
Definition: Math function measuring how wrong the model output is; training minimizes it.
Picture this: An error scale: the lower, the better.
In the course: How an LLM works

LTV / Customer Lifetime Value

In plain words: How much money one customer brings in over the whole relationship.
Definition: Total revenue from the average customer over the relationship.
Picture this: Not one purchase, but the whole history of the relationship in dollars.
In the course: First clients, Unit Economics

Machine learning

In plain words: The branch of AI where a system learns from data instead of rules.
Definition: AI subfield where systems learn patterns from data, not rules.
Picture this: Teaching a child by showing, not by explaining theory.
In the course: How an LLM works

Marketplace

In plain words: A catalog of plugins/skills/agents you can install.
Definition: Platform listing plugins/skills/agents for install.
Picture this: The App Store for AI tools.
In the course: What Skills are, The skills ecosystem

MCP / Model Context Protocol

In plain words: An open protocol for connecting AI to tools and data.
Definition: Anthropic standard letting LLMs connect to tools and data via servers.
Picture this: The USB standard for AI: one port for everything.
In the course: MCP, MCP Builder

Message

In plain words: One step in a conversation: a turn from the user or the AI.
Definition: Single chunk of a conversation: a user or assistant turn.
Picture this: One message in a chat.
In the course: Tokens and context

Metric

In plain words: A number you measure over time.
Definition: Quantitative measure tracked over time.
Picture this: The reading on a pressure gauge.
In the course: Product analytics, Production Observability

MFA / Multi-Factor Authentication

In plain words: Authentication with 2+ independent factors.
Definition: Authentication with 2+ independent factors.
Picture this: A lock + a code + a fingerprint.
In the course: Permissions and security, Security and secrets

Microservice

In plain words: A small independent service inside a bigger system.
Definition: Small independent service doing one thing in a larger system.
Picture this: Shops in a mall: each does its own thing, but they're side by side.
In the course: Deployment

Midjourney

In plain words: One of the leading AI image generators; it runs in the browser on a subscription.
Definition: Image generation AI by subscription; works in the web app.
Picture this: An artist who paints from text: you write, you get a picture.
In the course: Image generation, AI video generation

Mistral

In plain words: A French AI lab: the Mistral Vibe assistant (formerly Le Chat), some models with open weights.
Definition: French AI lab; assistant Mistral Vibe (formerly Le Chat), some models with open weights.
Picture this: Europe's player in the LLM race.
In the course: Comparing models, Local models

Modal

In plain words: A serverless platform for AI/ML with GPUs.
Definition: Serverless platform for AI/ML workloads with GPUs.
Picture this: AWS Lambda, but with GPUs and built for AI.
In the course: Deployment

Monitoring

In plain words: Continuously watching the health of a system.
Definition: Continuous observation of system health and metrics.
Picture this: An ICU: beeping monitors and constant observation.
In the course: Production Observability

MRR / Monthly Recurring Revenue

In plain words: Monthly recurring revenue from subscriptions.
Definition: Predictable monthly subscription revenue.
Picture this: The business's paycheck: it comes in every month.
In the course: First clients, Unit Economics

MQL / Marketing Qualified Lead

In plain words: A lead "warmed up" by marketing and ready to hand off to sales.
Definition: Lead engaged enough by marketing to be passed to sales.
Picture this: The shopper already tried it on, but hasn't paid.
In the course: Sales AI

MTTR / Mean Time To Recovery

In plain words: The average time to recover after an outage.
Definition: Average time to fix an incident.
Picture this: Pit stop speed in Formula 1.
In the course: Production Observability

Multi-agent

In plain words: Several agents working on a task together.
Definition: Multiple AI agents collaborating on a task.
Picture this: A construction crew instead of one handyman.
In the course: Agent teams, Multi-agent orchestration

Multimodal

In plain words: AI that works with different formats: text, images, sound, video.
Definition: AI handling multiple input types: text, image, audio, video.
Picture this: An employee who can read, listen and see.
In the course: Multimodality

MVC / Model-View-Controller

In plain words: An architecture pattern: data / interface / logic kept separate.
Definition: Pattern separating data, UI and logic in apps.
Picture this: A factory: the shop floor / the showroom / the manager, each knowing their role.
In the course: Websites and web apps

Neural network

In plain words: Layers of mathematical nodes imitating the brain's neurons.
Definition: Layers of mathematical nodes mimicking brain neurons.
Picture this: A web of nodes where the signal travels from input to output.
In the course: How an LLM works

Niche

In plain words: A narrow market with specific needs.
Definition: Narrow target market with specific needs.
Picture this: A truffle shop vs. a supermarket.
In the course: Choosing a niche

NLP / Natural Language Processing

In plain words: The branch of AI dealing with human language.
Definition: Field of AI dealing with human language.
Picture this: All the technology around understanding text.
In the course: How an LLM works

North star metric

In plain words: The one main metric that reflects the product's value.
Definition: Single metric capturing the product's core value.
Picture this: The North Star: everyone steers by it.
In the course: Product analytics

NPM / Node Package Manager

In plain words: The package manager for JavaScript/Node.js.
Definition: Default package manager for JavaScript/Node.js.
Picture this: A brick supply store for building JS projects.
In the course: Installation and setup

Niche / Vertical SaaS

In plain words: SaaS for a specific industry: dental offices, real estate.
Definition: SaaS for a specific industry vertical (dental, real estate).
Picture this: A tailored suit vs. off the rack.
In the course: Choosing a niche

Observability

In plain words: The ability to understand what's going on inside a system.
Definition: Ability to understand system state from logs, metrics and traces.
Picture this: A see-through engine: you can see every part.
In the course: Production Observability

OAuth

In plain words: The standard behind "sign in with Google/GitHub" without a password.
Definition: Standard letting third-party apps access user data without a password.
Picture this: A guest key: access to one room only, not the whole apartment.
In the course: Permissions and security, Security and secrets

Ollama

In plain words: A tool for running local models in one click.
Definition: Tool running local LLMs with a simple CLI.
Picture this: Docker for AI models.
In the course: Local models

Omnichannel

In plain words: One consistent experience across every channel: website, email, chat.
Definition: Consistent experience across all channels (web, mobile, email, etc.).
Picture this: The same salesperson recognizes you at any branch.
In the course: AI messengers, CRM on autopilot

Onboarding

In plain words: The process of getting a new customer to value quickly.
Definition: Process of getting a new user to value quickly.
Picture this: The first-day tour for a new employee.
In the course: Delivery and Retention

OpenAI

In plain words: The company behind ChatGPT, the GPT-6 family, Whisper and the gpt-image-2 image generator.
Definition: Company behind ChatGPT, the GPT-6 family, Whisper, gpt-image-2.
Picture this: Anthropic's main competitor.
In the course: Comparing models

OpenAPI

In plain words: A standard for describing REST APIs in YAML/JSON.
Definition: Standard for describing REST APIs in YAML/JSON.
Picture this: A restaurant menu: what's served, in what form, at what price.
In the course: APIs and integrations

Organic

In plain words: Unpaid traffic: search, social media, recommendations.
Definition: Unpaid traffic from search, social and referrals.
Picture this: A garden that grew on its own after planting vs. a heated greenhouse.
In the course: The SEO machine

ORM / Object-Relational Mapping

In plain words: A library that links database rows to objects in code.
Definition: Library mapping database rows to programming objects.
Picture this: An interpreter between two languages: the database and the code.
In the course: Websites and web apps

Overfitting

In plain words: The model memorized the examples instead of learning to generalize.
Definition: Model memorizes training data and fails on new data.
Picture this: A student who memorized the practice test falls apart on different questions.
In the course: How an LLM works

OWASP

In plain words: An open web security project, known for its Top 10 vulnerabilities list.
Definition: Open Web Application Security Project; Top 10 vulnerabilities list.
Picture this: The rulebook of web application security.
In the course: Permissions and security, Security and secrets

Paid acquisition

In plain words: Getting customers through paid advertising.
Definition: Customers acquired via paid ads.
Picture this: You pay for every visitor who walks into the store.
In the course: AI advertising

Paid trial

In plain words: A paid trial period with the option of a refund.
Definition: Trial that requires payment upfront, refundable.
Picture this: A deposit for a test drive.
In the course: Pricing strategy

Parameter

In plain words: An adjustable weight inside a network: the "size" of a model.
Definition: Tunable weight inside a neural net; a large model = many params.
Picture this: A screw in an engine. Modern models have billions of them or more, but companies usually don't disclose exact sizes.
In the course: How an LLM works

Password manager

In plain words: An encrypted safe for your passwords.
Definition: Tool securely storing passwords (1Password, Bitwarden).
Picture this: One master key for a thousand locks.
In the course: API keys and .env, Security and secrets

Paywall

In plain words: A "pay to continue" barrier.
Definition: Barrier requiring payment to access content/features.
Picture this: A subway turnstile: no ticket, no entry.
In the course: Pricing strategy

Payback period

In plain words: How long it takes for cash flow to return an investment.
Definition: Time to recover an investment from cash flow.
Picture this: How many months of rental income it takes to pay off the equipment.
In the course: Unit Economics

Penetration test / pentest

In plain words: Authorized hacking to find vulnerabilities.
Definition: Authorized hacking to find vulnerabilities.
Picture this: You hired a "burglar" to test the locks on your store.
In the course: Security and secrets

Perplexity

In plain words: An AI search engine with citations and real web search.
Definition: AI answer engine with citations and live web search.
Picture this: A search engine that read everything itself and summed it up for you.
In the course: Comparing models

Persona

In plain words: A made-up portrait of your ideal customer, with a name and a personality.
Definition: Fictional ideal customer with a name, traits and goals.
Picture this: "Megan, 32, a B2B marketer": a specific person in your head.
In the course: Choosing a niche

PII / Personally Identifiable Information

In plain words: Personal data that can identify a person.
Definition: Data that can identify a person (name, email, phone).
Picture this: A scrap of a passport: even a scrap can give everything away.
In the course: Permissions and security, Security and secrets

Pipeline

In plain words: A sequence of stages: for sales or for data.
Definition: Sequence of stages: sales (lead → close) or data (extract → transform → load).
Picture this: A conveyor: input → stage 1 → stage 2 → output.
In the course: Sales AI

Plan mode

In plain words: A Claude Code mode: first the plan, then the actions (enter with Shift+Tab or /plan).
Definition: Claude Code mode where it plans before executing (Shift+Tab or /plan).
Picture this: "Read the spec → draft a plan → get sign-off → do it."
In the course: Work modes

Plugin

In plain words: An add-on that extends what a tool can do.
Definition: Add-on extending tool functionality (Claude Code plugins).
Picture this: An attachment for a drill: same base, more capabilities.
In the course: Plugin Security, The skills ecosystem

Post-mortem

In plain words: An incident review: what happened, why, and how to keep it from happening again.
Definition: Incident analysis: what happened, why, how to prevent it.
Picture this: A crash investigation so it doesn't happen again.
In the course: Production Observability

PPC / Pay-Per-Click

In plain words: Advertising where you pay per click.
Definition: Ads where you pay each time someone clicks.
Picture this: You pay for every person who walks into the store.
In the course: AI advertising

Pricing tier

In plain words: A level of a paid plan.
Definition: Level of a paid plan (Starter/Pro/Enterprise).
Picture this: Economy / Business / First Class.
In the course: Pricing strategy

Production

In plain words: The live environment with real customers.
Definition: Live environment used by real customers.
Picture this: A stage with an audience: no more rehearsals.
In the course: Deployment

Project

In plain words: A Claude Code workspace: one folder, one CLAUDE.md.
Definition: Workspace in Claude Code: one folder, one CLAUDE.md.
Picture this: A folder for one case, not a mishmash of everything.
In the course: CLAUDE.md, The WAT framework

Prompt

In plain words: A text instruction sent to an LLM.
Definition: Instruction text sent to an LLM.
Picture this: Your order to the waiter: what you want.
In the course: Prompting fundamentals

Prompt caching

In plain words: Caching repeated chunks of prompts so you don't pay for them again.
Definition: Storing reused prompt parts to avoid re-tokenization charges.
Picture this: You don't buy milk twice if you already have some.
In the course: Prompt Caching and the Batch API

Prompt injection

In plain words: An attack: user input contains hidden instructions for the LLM.
Definition: Attack where user input contains hidden instructions overriding the system prompt.
Picture this: Someone hid a note inside a form, and the secretary read it aloud to the boss.
In the course: Prompt Injection Defense

Qualification

In plain words: A sales stage: is this lead worth the effort?
Definition: Sales stage assessing whether a lead is worth pursuing.
Picture this: A questionnaire at the door so you don't waste time on people who aren't a fit.
In the course: Sales AI

RAG / Retrieval-Augmented Generation

In plain words: The LLM gets access to an external knowledge base when it answers.
Definition: Combining an LLM with retrieval from an external knowledge base.
Picture this: A student with the textbook open during the exam: writes in their own words but checks the facts.
In the course: RAG

RBAC / Role-Based Access Control

In plain words: Permissions are assigned to roles, not to people.
Definition: Permissions assigned by role, not individually.
Picture this: "The cashier sees the register, the accountant sees the reports": by job title, not by name.
In the course: Permissions and security, Security and secrets

Region

In plain words: The geographic region of a cloud data center.
Definition: Geographic location of a cloud data center (us-east, eu-west).
Picture this: The city your server lives in.
In the course: Deployment

Regression

In plain words: A task of predicting a number: a price, a temperature.
Definition: ML task: predict a continuous number (price, temperature).
Picture this: "How much is this apartment worth?": not a category, a number.
In the course: How an LLM works

Reinforcement learning / RL

In plain words: Learning by trial and error, with rewards.
Definition: Learning by trial and error with rewards/punishments.
Picture this: Training a dog with treats.
In the course: How an LLM works

Replicate

In plain words: A platform that hosts open-source AI models behind an API.
Definition: Platform hosting open-source AI models via API.
Picture this: Heroku for AI models.
In the course: Image generation

Request

In plain words: A message from the client to the server.
Definition: Message sent from client to server.
Picture this: Your request at the DMV window.
In the course: APIs and integrations

Response

In plain words: The server's answer to a request.
Definition: Server's answer to a request.
Picture this: What they hand you back at the window.
In the course: APIs and integrations

REST / Representational State Transfer

In plain words: The standard HTTP API pattern, with action verbs.
Definition: Standard HTTP API design pattern (GET, POST, PUT, DELETE).
Picture this: Standardized doors in a store: everyone knows where the entrance, the register and the exit are.
In the course: APIs and integrations

Retainer

In plain words: A regular payment for a service, like a membership.
Definition: Recurring fee for ongoing service.
Picture this: Having a lawyer on retainer: a fee every month.
In the course: First clients, Pricing strategy

Retention

In plain words: The share of customers who stay: the opposite of churn.
Definition: % of customers staying over time.
Picture this: How many guests come back a second time.
In the course: Delivery and Retention

Revenue

In plain words: Money coming in, before expenses are taken out.
Definition: Money from sales before costs.
Picture this: Everything in the cash register at the end of the day.
In the course: Unit Economics

RLHF / Reinforcement Learning from Human Feedback

In plain words: Training models using ratings from people.
Definition: Training method using humans rating outputs to align the model.
Picture this: Food tasters: the cook improves based on their scores.
In the course: How an LLM works

ROI / Return on Investment

In plain words: Return on investment.
Definition: Profit relative to investment cost.
Picture this: You put in $100 and got back $150: a 50% ROI.
In the course: Unit Economics

Rollback

In plain words: Going back to the previous version after a bad rollout.
Definition: Reverting to the previous version after a bad deploy.
Picture this: "That was a mistake, let's put it back the way it was."
In the course: Deployment, Production Observability

Runtime

In plain words: The environment code runs in.
Definition: Environment executing code (Node.js, the Python interpreter).
Picture this: The stage where the play is performed.
In the course: API keys and .env

Runway

In plain words: An AI video generation platform.
Definition: AI video generation platform.
Picture this: A Pixar studio in one web app.
In the course: AI video generation

SaaS / Software as a Service

In plain words: Software you access over the internet on a subscription.
Definition: Software accessed via subscription over the internet.
Picture this: Renting an apartment vs. buying a house.
In the course: First clients

Sampling

In plain words: How the next token gets picked during generation.
Definition: How the model chooses the next token; temperature and top-p control randomness.
Picture this: A lottery with rules: sometimes looser, sometimes stricter.
In the course: How an LLM works

SAM / Serviceable Addressable Market

In plain words: The part of the TAM you can actually serve.
Definition: Portion of the TAM you can actually serve with your model.
Picture this: Of everyone who wants in, how many you can actually take.
In the course: Choosing a niche

Scaling

In plain words: Adjusting capacity to match the load.
Definition: Adjusting capacity to handle changing load.
Picture this: Opening more registers when the line gets long.
In the course: Deployment

Schema markup

In plain words: Structured data for search engines.
Definition: Structured data for search engines (JSON-LD).
Picture this: A detailed ingredients label: easier for the search engine to read.
In the course: The SEO machine

Scheduled task

In plain words: A task that runs on a schedule.
Definition: Job running on a schedule (cron-like).
Picture this: An alarm clock for code.
In the course: Loop and Scheduled Tasks

SDK / Software Development Kit

In plain words: Tools and libraries for building on a platform.
Definition: Tools and libraries for building with a platform.
Picture this: A specialist's toolkit for one particular job.
In the course: API keys and .env, Subagents

SDR / Sales Development Representative

In plain words: A salesperson who does cold prospecting and qualification.
Definition: Salesperson focused on outbound prospecting and qualification.
Picture this: A scout who finds targets before the battle.
In the course: Sales AI

Secret rotation

In plain words: Regularly replacing secrets to reduce the risk of a leak.
Definition: Periodically replacing secrets/keys to limit damage.
Picture this: Changing your password every six months.
In the course: Security and secrets

Secrets

In plain words: Keys, passwords, tokens: never in the code, only in the environment.
Definition: API keys, passwords, tokens: must not be in code.
Picture this: The key to the safe isn't left sitting on the door.
In the course: API keys and .env, Security and secrets

Segmentation

In plain words: Dividing an audience into groups by traits.
Definition: Splitting an audience into groups by traits.
Picture this: A store groups its shoppers: students / families / retirees.
In the course: Choosing a niche

SEM / Search Engine Marketing

In plain words: Paid ads in search results.
Definition: Paid search ads (Google Ads).
Picture this: Buying a spot at the top of Google.
In the course: AI advertising

SEO / Search Engine Optimization

In plain words: Optimizing for search engines to get free traffic.
Definition: Improving organic search rankings.
Picture this: Making your stall so good that the market managers give it the best spot for free.
In the course: The SEO machine

Sentiment analysis

In plain words: Figuring out the emotional tone of a text.
Definition: Detecting emotion/opinion in text (positive/negative).
Picture this: A mood thermometer for reviews.
In the course: How an LLM works

Sequence

In plain words: A series of emails sent on a schedule.
Definition: Series of emails sent on a schedule.
Picture this: A TV series: episode by episode, it warms you up.
In the course: AI communications

Server

In plain words: A computer that serves clients over a network.
Definition: Computer providing services to clients over a network.
Picture this: The kitchen cooks, the waiters carry the plates out: the kitchen is the server.
In the course: Deployment

Serverless

In plain words: A cloud where you write functions and the servers are the provider's problem.
Definition: Cloud model where you write functions and the provider handles the servers.
Picture this: A taxi on demand vs. owning a car.
In the course: Deployment

settings.json

In plain words: Claude Code's config: permissions, hooks, variables.
Definition: Claude Code config: permissions, hooks, env vars.
Picture this: An app's settings: what's allowed, what's turned on.
In the course: Hooks, Permissions and security

Skill

In plain words: A reusable Claude capability: a folder with a SKILL.md and resources.
Definition: Reusable Claude capability defined in a folder (SKILL.md + assets).
Picture this: A process template Claude applies when it's needed.
In the course: What Skills are, Building a skill, Skills architecture

Slash command

In plain words: A Claude Code command that starts with /.
Definition: Claude Code command starting with / (/clear, /init).
Picture this: A hotkey.
In the course: Built-in commands

SLA / Service Level Agreement

In plain words: A commitment on availability/quality, for example 99.9%.
Definition: Promise of uptime/performance (99.9%).
Picture this: A TV warranty: it works for 2 years or we replace it.
In the course: Deployment, Production Observability

SOC 2

In plain words: A security standard enterprise customers require.
Definition: Security compliance standard demanded by enterprises.
Picture this: A UL certification mark for cloud services.
In the course: AI Regulation & Compliance

Social proof

In plain words: Evidence that other people trust you: reviews, logos.
Definition: Evidence others trust you (testimonials, logos, reviews).
Picture this: A line outside a restaurant: the food must be good.
In the course: Funnels for AI products

SOM / Serviceable Obtainable Market

In plain words: The share of the SAM you can realistically reach.
Definition: Realistic share of the SAM you can capture.
Picture this: Of the people you could serve, how many will actually find their way to you.
In the course: Choosing a niche

SQL / Structured Query Language

In plain words: The standard language of relational databases.
Definition: Standard language for relational databases.
Picture this: A universal language for asking the warehouse for things.
In the course: Websites and web apps

SQL (Sales Qualified Lead)

In plain words: A lead that has passed sales qualification and is ready for active work.
Definition: Lead vetted by sales as ready for active selling.
Picture this: A shopper who's already holding out a card: definitely buying.
In the course: Sales AI

SSH / Secure Shell

In plain words: A secure remote terminal.
Definition: Encrypted remote terminal access.
Picture this: A secure tunnel to another machine.
In the course: Deployment

SSL/TLS

In plain words: A cryptographic protocol that protects data in transit.
Definition: Cryptographic protocol securing internet communications.
Picture this: A sealed envelope for letters on the internet.
In the course: Permissions and security, Security and secrets

Stack

In plain words: The set of technologies in a project.
Definition: Combination of technologies used to build an app (LAMP, MERN).
Picture this: A row of tools: hammer, screwdriver, power drill.
In the course: Websites and web apps

Status page

In plain words: A public page showing the state of a service.
Definition: Public page showing service uptime and incidents.
Picture this: The departures board at an airport: what's flying, what's delayed.
In the course: Production Observability

Stream

In plain words: A continuous flow of data that's processed as it arrives.
Definition: Continuous data flow processed as it arrives.
Picture this: Claude streaming its answer: text word by word, not as one whole file.
In the course: APIs and integrations

Subagent

In plain words: A sub-agent called from the main session to handle subtasks.
Definition: Smaller Claude agent invoked from the main session for subtasks.
Picture this: You hand a separate task to an intern while you focus on the main thing.
In the course: Subagents

Suno

In plain words: An AI music generation platform.
Definition: AI music generation platform.
Picture this: A composer working from a description: you write "sad guitar," you get a track.
In the course: Music with AI

Supervised learning

In plain words: Learning from labeled data: "input → answer" pairs.
Definition: ML with labeled training data (input → known correct output).
Picture this: A teacher showing the correct answers.
In the course: How an LLM works

TAM / Total Addressable Market

In plain words: The total size of the market for a product.
Definition: Total market demand for a product/service.
Picture this: Every potential buyer in the world.
In the course: Choosing a niche

Target audience

In plain words: The specific group a product is aimed at.
Definition: Specific group of people content/a product is aimed at.
Picture this: Not "everyone," but "these people right here."
In the course: Choosing a niche

Telemetry

In plain words: Automatic collection of a system's metrics and events.
Definition: Automated collection of system metrics and events.
Picture this: An airplane's black box records everything.
In the course: Production Observability

Temperature

In plain words: A setting for how random generation is: 0 = strict, 1 = creative. In the API for the newest Claude models, the manual setting has been removed (the current models are on the What's current page).
Definition: LLM parameter controlling output randomness (0 = strict, 1 = creative); removed for the newest Claude models in the API.
Picture this: A dial from "strict accountant" to "free-spirited poet."
In the course: Prompting fundamentals

Terminal

In plain words: A text interface for giving commands to the operating system.
Definition: Text interface for typing commands to the OS.
Picture this: Talking to the system directly, no middleman.
In the course: Installation and setup

Throughput

In plain words: How many requests get processed per unit of time.
Definition: Requests processed per unit of time.
Picture this: How many cars cross the bridge in an hour.
In the course: Production Observability

Together AI

In plain words: A platform for hosting and fine-tuning open-source models.
Definition: Platform hosting and fine-tuning open-source models.
Picture this: A marketplace of cloud AI models.
In the course: Local models

Token

In plain words: The basic unit an LLM processes: a little over half a word of English on current Claude models (about 0.75 on older ones).
Definition: Basic unit an LLM processes: a little over half a word of English on current Claude models (about 0.75 on older ones).
Picture this: No letters, no words: just tokens, and you pay for each one.
In the course: Tokens and context

Tokenization

In plain words: Splitting text into tokens before processing.
Definition: Splitting text into tokens before model processing.
Picture this: Slicing bread before you eat it.
In the course: How an LLM works

Tool

In plain words: A function an LLM can call to take an action.
Definition: Function an LLM can call to perform real-world actions.
Picture this: A hammer for the agent: it can drive a nail, not just talk.
In the course: MCP

Tool use

In plain words: An LLM's ability to call tools while it reasons.
Definition: LLM capability to invoke tools/functions during reasoning.
Picture this: It doesn't just talk, it picks up the tools itself.
In the course: MCP

Tracing

In plain words: Following a single request through every service.
Definition: Following a single request through all services for debugging.
Picture this: Package tracking on the shipper's website: where is it right now?
In the course: Production Observability

Training

In plain words: The process of teaching a network on data, adjusting its weights.
Definition: Process where a neural net learns from data, adjusting weights.
Picture this: A college student: months of study before the diploma.
In the course: How an LLM works

Transfer learning

In plain words: Taking a ready-made model for a new task: saves time.
Definition: Using a pretrained model for a new task: saves time and data.
Picture this: An experienced employee gets up to speed on a new project faster than a rookie.
In the course: How an LLM works, Fine-tuning

Transformer

In plain words: A neural network architecture (2017): the foundation of every modern LLM.
Definition: Neural net architecture (2017), the basis for all modern LLMs.
Picture this: "Attention": what matters right now, with everything else in the background.
In the course: How an LLM works

Trial

In plain words: A trial period for testing a product.
Definition: Free or low-cost period to test a product.
Picture this: A test drive.
In the course: Pricing strategy

ultrathink

In plain words: A keyword in Claude Code that asks for deeper thinking. These days depth is controlled with the /effort command.
Definition: Claude Code keyword that raised the thinking budget (depth is now steered via /effort).
Picture this: "Focus, this is important": a flag for the AI.
In the course: Extended Thinking

Underfitting

In plain words: The model is too simple: bad everywhere.
Definition: Model too simple to capture patterns; bad on both training and test data.
Picture this: Measuring a curve with a straight ruler: it doesn't work.
In the course: How an LLM works

Unsupervised learning

In plain words: Learning without labels: the model finds patterns on its own.
Definition: ML on unlabeled data: the model finds patterns itself.
Picture this: An archaeologist sorting finds without a textbook.
In the course: How an LLM works

Uptime

In plain words: The percentage of time a service is working.
Definition: % of time a service is operational.
Picture this: Out of 365 days, how many the store was open.
In the course: Production Observability

Upsell

In plain words: Selling a more expensive version to an existing customer.
Definition: Selling a more expensive version to an existing customer.
Picture this: "Would you like to make that a double?"
In the course: Pricing strategy

Upmarket

In plain words: Moving from small customers to big ones.
Definition: Moving from small customers to larger enterprises.
Picture this: Café → restaurant → banquets for 200 people.
In the course: First clients, Monetization case studies

Value-based pricing

In plain words: Pricing by value to the customer, not by your costs.
Definition: Price based on value to the customer, not cost to produce.
Picture this: Not "cost + 20%," but "how much will this save them?"
In the course: Pricing strategy

Vector

In plain words: An array of numbers that represents meaning.
Definition: Numerical array representing meaning (embeddings).
Picture this: Coordinates in a space of meanings.
In the course: RAG

Vercel

In plain words: A frontend hosting platform with serverless functions, from the makers of Next.js.
Definition: Frontend hosting platform with serverless functions; creators of Next.js.
Picture this: A home for React/Next.js apps.
In the course: Deployment

Version control

In plain words: A system that tracks changes to files over time.
Definition: System tracking changes to files over time (Git).
Picture this: Track Changes in Word, on steroids.
In the course: GitHub Worktrees

vLLM

In plain words: A high-performance LLM inference server for self-hosting.
Definition: High-performance LLM inference server for self-hosting.
Picture this: A sports car for running local models.
In the course: Local models

/voice

In plain words: A Claude Code command for voice input/output.
Definition: Claude Code command for voice input/output.
Picture this: A microphone for the AI.
In the course: Voice tools

Vulnerability

In plain words: A weak spot in a system that can be exploited.
Definition: Weakness in a system that attackers can exploit.
Picture this: An unlocked window in a house.
In the course: Permissions and security, Security and secrets

Webhook

In plain words: A callback: a service knocks on your door when an event happens.
Definition: HTTP callback: a service calls your URL when an event happens.
Picture this: A doorbell: a guest is here, go open the door.
In the course: APIs and integrations

Whisper

In plain words: OpenAI's speech recognition model, with open weights.
Definition: OpenAI's speech-to-text model: open weights, widely used.
Picture this: A stenographer with near-perfect accuracy.
In the course: Voice tools

Work mode

In plain words: Claude Code's different working modes (permission modes); switch between them with Shift+Tab.
Definition: Different operational modes of Claude Code (permission modes: Manual, acceptEdits, plan, auto, dontAsk, bypassPermissions).
Picture this: Bicycle gears: uphill, flat, downhill.
In the course: Work modes

Worker

In plain words: A serverless function on Cloudflare's edge network.
Definition: Cloudflare's serverless function platform.
Picture this: A function that runs close to the user.
In the course: Deployment

Workspace

In plain words: A Claude Code container that can hold several projects.
Definition: Top-level Claude Code container that can hold multiple projects.
Picture this: An office floor with project rooms on it.
In the course: The WAT framework

YAML

In plain words: A human-readable data format for configs.
Definition: Human-readable data format used for configs.
Picture this: A form that uses indentation instead of brackets.
In the course: Hooks

Zero-shot

In plain words: Asking a model to do a task with no examples in the prompt.
Definition: Asking a model to do a task without examples in the prompt.
Picture this: Straight to business: "do X," with no "like this."
In the course: Prompting fundamentals

Zero trust

In plain words: A security model: trust no one by default.
Definition: Security model trusting nothing inside or outside the network by default.
Picture this: Every door needs a badge, even inside the building.
In the course: Security and secrets

Cross-reference: the course's 100 key terms

AI fundamentals: AGI, ANI, Alignment, Safety, Attention, Backpropagation, Bias, CoT, Classification, Clustering, Completion, Context window, CNN, Deep learning, Decoder, Embedding, Emergent capabilities, Encoder, Eval, Few-shot, Fine-tuning, Function calling, Generative, GAN, Gradient descent, Ground truth, Hallucination, In-context learning, Inference, Instruction tuning, Latent space, LLM, Loss function, Machine learning, Multimodal, Neural network, NLP, Overfitting, Parameter, Prompt, RAG, Regression, RL, RLHF, Sampling, Sentiment, Supervised, Temperature, Token, Tokenization, Training, Transfer learning, Transformer, Underfitting, Unsupervised, Vector, Zero-shot.

Claude Code: Agent, Agent SDK, CLAUDE.md, Computer Use, Custom command, eval.json, Extended thinking, Hook, /clear, /compact, /context, /loop, MCP, Plan mode, Plugin, Project, Prompt caching, Scheduled task, settings.json, Skill, Slash command, Subagent, ultrathink, .claudeignore, Work mode, Batch API, Tool, Tool use, Agent teams, /init, Computer use tier, Headless, /voice, .claude folder, claude-mem, Marketplace, Workspace, Message, Multi-agent.

Programming: API, API key, async, Backend, Cache, CLI, CDN, Container, Cron, Database, Docker, Endpoint, Frontend, Function, Git, GitHub, HTTP, HTTPS, IDE, JSON, KV store, Latency, Library, Microservice, MVC, ORM, Request, Response, REST, Runtime, Serverless, SDK, SQL, SSH, Stack, Stream, Terminal, Throughput, Version control, Webhook, YAML.

Cloud/Deploy: AWS, Cloudflare, Vercel, Deploy, Docker compose, Edge, Environment variable, GitHub Actions, GPU, Hosting, Kubernetes, Lambda, Load balancer, Monitoring, NPM, Observability, OpenAPI, Region, Scaling, Secrets, Server, SLA, Status page, Uptime, Worker.

Business: ARR, B2B, B2C, BANT, CAC, Churn, Closing, Cold outreach, Conversion, CRM, Customer success, Demo, Discovery call, Freemium, Funnel, Gross margin, ICP, KPI, Lead, Lead magnet, LTV, MRR, MQL, Niche, North star metric, Onboarding, Paid trial, Paywall, Payback period, Persona, Pipeline, Pricing tier, Qualification, Retention, Retainer, Revenue, ROI, SAM, SaaS, Segmentation, SDR, Sequence, SOM, SQL (sales), TAM, Trial, Upsell, Upmarket, Value-based pricing, Vertical SaaS.

Marketing: AEO, Attribution, A/B test, Brand voice, CAC payback, CMS, Conversion rate, Copy, CPC, CTR, Deck, Dwell time, Engagement, GTM, Hook (marketing), Keyword, Landing page, Lead generation, Omnichannel, Organic, Paid acquisition, PPC, SEO, SEM, Social proof, Target audience.

Production/DevOps: Alert, Audit log, Canary deploy, CI/CD, Dependency, Deployment, Distributed system, Downtime, Environment, Error budget, Feature flag, GitOps, Incident, Infrastructure as code, Log, Metric, MTTR, Observability, Post-mortem, Production, Rollback, Secret rotation, Telemetry, Tracing, Uptime.

Security: 2FA, AAA, API key rotation, Authentication, Authorization, bcrypt, Breach, CSRF, Encryption at rest, Encryption in transit, GDPR, Hash, IAM, JWT, MFA, OAuth, OWASP, Password manager, Penetration test, PII, Prompt injection, RBAC, SOC 2, SSL/TLS, Vulnerability, Zero trust.

AI ecosystem: Anthropic, ChatGPT, Claude, Copilot, Cursor, DeepSeek, ElevenLabs, fal.ai, Flux, Gemini, Hugging Face, LangChain, LangGraph, LangSmith, LiteLLM, LM Studio, Llama, Midjourney, Mistral, Modal, Ollama, OpenAI, Perplexity, Replicate, Runway, Suno, Together AI, vLLM, Whisper.

Glossary maintenance: updated whenever new terms appear in the lessons. Format: five lines per term, sorted alphabetically, with a "picture this" image where it helps. Every term links to the lessons where it's covered in detail.

Last updated: 2026-10 Total terms: 300+ Version: 2.1