AI SaaS Glossary: Terms from A to Z

A

Affiliate program: An affiliate program pays partners a commission for customers they refer through a tracked link, turning bloggers, creators and agencies into a sales channel.; Affiliate program
AI agency: An AI agency builds and runs AI solutions for clients, such as chatbots, content workflows or automations, and often resells a white-label platform under its own brand.
AI agent: An AI agent is an AI system that works toward a goal in several steps, deciding which tools to use, acting, checking the result and continuing, rather than giving a single reply.; Read the guide
AI assistant: An AI assistant is a chat model set up for a specific job, with its own instructions, personality, tools and often a knowledge base, such as a support agent or a legal research helper.; AI assistants
AI chatbot: An AI chatbot is a conversational interface powered by a language model, often embedded on a website to answer visitors’ questions from the business’s own content.; Website chatbots
AI credits: AI credits are a unit an AI SaaS sells to its own users and converts to tokens, images or seconds of audio at rates it sets, as opposed to the tokens AI providers bill by.; Read the guide
AI hallucination: An AI hallucination is a confident answer from a model that is false or made up, such as an invented fact, quote, link or source.; Read the guide
AI memory: AI memory lets a chat assistant remember facts about a user across conversations, such as their role, preferences or projects, and use them in later answers.; AI chat
AI SaaS: An AI SaaS is a software-as-a-service product whose core value comes from AI models, sold by subscription or usage, such as an AI writer, image generator or chatbot builder.; AI SaaS script
AI upscaling: AI upscaling increases an image’s resolution while adding plausible detail, so small or blurry images can be used at larger sizes.
AI web search: AI web search lets a chat model search the internet during a conversation and answer from current pages, with sources, instead of only from its training data.; AI chat
AI wrapper: An AI wrapper is a product that sells access to another company’s AI model through its own interface, workflow or niche, rather than training its own model.; Read the guide
AI writer: An AI writer is a tool that drafts text such as blog posts, ads, emails and product descriptions from a short brief, usually through ready-made templates.; AI writer
API: An API (application programming interface) is a defined way for programs to talk to each other, such as an app sending prompts to an AI provider and getting answers back.
API key: An API key is a secret token that identifies and authorises an app when it calls a service such as an AI provider; whoever holds the key is billed for its usage.
ARPU: ARPU (average revenue per user) is total revenue in a period divided by the number of active users or customers.
ARR (Annual Recurring Revenue): ARR (annual recurring revenue) is the yearly value of all active subscriptions, usually MRR multiplied by twelve.

B

Build vs buy: Build vs buy is the decision between developing software in-house and licensing an existing product, weighing cost, time to launch, control and ongoing maintenance.; Build vs buy calculator
BYOK (Bring Your Own Key): BYOK (bring your own key) lets customers of an AI product use their own API key from a provider such as OpenAI, so their usage is billed to them by the provider instead of through the product.; Read the guide

C

CAC (Customer Acquisition Cost): CAC (customer acquisition cost) is what you spend on marketing and sales to win one new paying customer.
CAC payback period: The CAC payback period is how many months of gross profit from a customer it takes to earn back what you spent to acquire them.
Canvas (AI document editing): A canvas is a side-by-side editor in an AI chat where the model writes a document or code into a separate panel that both the user and the AI can then edit.; Canvas
CDN (Content Delivery Network): A CDN (content delivery network) serves files such as images, scripts and videos from servers close to each user, so pages load faster and the main server does less work.
Chain of thought: Chain of thought is when a model works through intermediate reasoning steps before giving its answer, either because it is prompted to or because it was trained to.
Chargeback: A chargeback is when a customer disputes a card payment with their bank and the bank reverses it, taking the money back from the seller and usually adding a fee.; Read the guide
ChatGPT clone: A ChatGPT clone is an app that reproduces the ChatGPT experience, chat with AI models, file uploads and conversation history, under your own brand and usually with several model providers.; ChatGPT clone script
Checkout: Checkout is the step where a customer picks a plan or product, enters billing details and pays; every extra field or step costs conversions.
Chunking: Chunking is splitting documents into smaller passages before embedding them, so retrieval can return just the relevant part rather than a whole file.
Churn rate: Churn rate is the share of customers, or of recurring revenue, lost in a period, usually a month, through cancellations and failed renewals.; Read the guide
Content moderation: Content moderation is checking user prompts and AI outputs for harmful or policy-breaking material, such as violence, hate or sexual content, and blocking or flagging it.; Security and privacy
Context window: The context window is the maximum number of tokens a model can take into account at once, covering the instructions, the conversation, any attached documents and its own answer.; Read the guide
Conversion rate: Conversion rate is the share of people who take a target action, for example visitors who sign up or free users who start paying.
Coupon (discount code): A coupon is a code customers enter at checkout for a discount, as a fixed amount or a percentage, once or for several billing periods.; Plans and credits
Credit top-up: A credit top-up is a one-off purchase of extra usage credits on top of a plan, so heavy users can keep going without upgrading.; Plans and credits
Cron job: A cron job is a command a server runs automatically on a schedule, for example every minute, used for renewals, clean-up and other recurring tasks.
Custom domain: A custom domain is your own web address, such as app.yourbrand.com, used for a product instead of the vendor’s address, a core part of white-labelling.
Custom GPT: A custom GPT is a version of a chat assistant configured with its own instructions, knowledge files and tools for a particular task; the name comes from OpenAI’s GPTs feature.; AI assistants

D

Data residency: Data residency is the requirement or choice to store data in a particular country or region, often for legal or customer reasons.
Diffusion model: A diffusion model is a type of generative AI that creates images or video by starting from random noise and removing it step by step until the result matches the prompt.
Dunning: Dunning is the process of recovering failed subscription payments, by retrying the card and emailing the customer to update their details before access is cut.

E

Embeddings: Embeddings are lists of numbers that represent the meaning of a piece of text, so that texts with similar meaning end up close together and can be found by similarity search.; Read the guide

F

Few-shot prompting: Few-shot prompting is including a handful of worked examples in a prompt so the model copies their format and style, without any training.
Fine-tuning: Fine-tuning is further training an existing model on your own examples so it learns a style, format or narrow task; for adding knowledge, RAG is usually cheaper and easier to update.
Foundation model: A foundation model is a large model trained on broad data that can be adapted to many tasks, such as GPT, Claude, Gemini or Llama, as opposed to a model built for one job.
Free trial: A free trial gives new users full or partial access to a paid plan for a limited time, after which they pay or lose access.; Read the guide
Freemium: Freemium is a pricing model with a permanently free plan that has limits, alongside paid plans that unlock more usage or features.
Function calling (tool use): Function calling, or tool use, lets an AI model ask the app to run a tool, such as a web search, a calculator or an API call, and then use the result in its answer.; Read the guide

G

GDPR: GDPR is the EU’s data protection law, which sets rules for collecting and processing personal data of people in the EU, including consent, access and deletion rights.; Security and privacy
Generative AI: Generative AI is AI that creates new content, text, images, audio, video or code, rather than only classifying or predicting from existing data.
GPU: A GPU (graphics processing unit) is a processor built for massively parallel maths, which makes it the standard hardware for training and running AI models.
Gross margin: Gross margin is revenue minus the direct cost of delivering it, such as AI provider fees, hosting and payment fees, as a share of revenue.; AI SaaS profit calculator
Grounding: Grounding is giving a model trusted source material, such as documents or search results, and asking it to answer from that material to reduce made-up answers.
GST: GST (goods and services tax) is the consumption tax used instead of VAT in countries such as India, Australia, Canada, New Zealand and Singapore, and often applies to foreign digital services.
Guardrails: Guardrails are the rules and checks around an AI model that keep it on topic, safe and within policy, such as instructions, moderation filters and output validation.

I

Image-to-image: Image-to-image is AI that takes an existing picture plus instructions and returns a changed version, for restyling, editing, removing backgrounds or keeping a character consistent.; AI image generator
Inference: Inference is running a trained model to get an output, as opposed to training it; every chat message or generated image is an inference request.
Inpainting: Inpainting is editing only a selected area of an image with AI, for example replacing an object or fixing a hand, while leaving the rest unchanged.
Invoice: An invoice is the formal billing document for a sale, showing seller and buyer details, items, amounts, tax and a sequential number; businesses need one to reclaim tax.; Invoicing

J

Job queue (background jobs): A job queue holds tasks that should run in the background, outside the user’s web request, and worker processes pick them up and run them one by one or in parallel.; Read the guide

K

Knowledge base (AI): In AI apps, a knowledge base is the set of documents, pages and files an assistant or chatbot can search and answer from.; Knowledge bases

L

Latency (time to first token): Latency is how long a model takes to respond; for chat, the key figure is time to first token, how soon the answer starts appearing.
Lifetime deal: A lifetime deal sells permanent access to a product for a single payment; for AI products it usually comes with usage limits, because each request has an ongoing cost.
LLM (Large Language Model): An LLM (large language model) is an AI model trained on very large amounts of text to predict the next token, which lets it write, summarise, translate, reason and hold a conversation.; Read the guide
Local LLM: A local LLM is a language model that runs on hardware you control, such as your own server or computer, instead of being called through a cloud provider’s API.; Read the guide
Localization (i18n): Localization adapts software to other languages and regions, translations, date and number formats and right-to-left layouts; i18n is the engineering that makes it possible.; Multilingual
LTV (Customer Lifetime Value): LTV (customer lifetime value) is the total gross profit a business expects from one customer over the whole time they stay a customer.; Read the guide

M

MCP (Model Context Protocol): MCP (Model Context Protocol) is an open standard that lets an AI model connect to outside tools and data sources through a common interface, so one connector works with any app that supports the protocol.; Read the guide
Merchant of record: A merchant of record (MoR) is the entity legally responsible for a sale: it processes the payment, collects and remits sales tax or VAT, and handles refunds and chargebacks.; Read the guide
Mixture of experts (MoE): Mixture of experts is a model design where only some of the model’s sub-networks, the experts, run for each token, which makes very large models faster and cheaper to run.
Moat: A moat is a lasting advantage that stops competitors from easily taking your customers, such as a loyal niche audience, proprietary data, integrations or distribution.
Model router: A model router is a service that gives access to many AI models through one API and account, and can route each request by price, speed or availability; OpenRouter is the best-known example.; OpenRouter integration
MRR (Monthly Recurring Revenue): MRR (monthly recurring revenue) is the predictable subscription revenue a business earns each month, with yearly plans counted as one twelfth per month.; Read the guide
Multi-tenant architecture: Multi-tenant architecture is a design where one installation of an application serves many customers, the tenants, each with their own isolated data, users and settings.; Read the guide
Multimodal AI: Multimodal AI is a model or system that works with more than one kind of data, such as text, images, audio and video, in its input, its output or both.; Read the guide
MVP (Minimum Viable Product): An MVP (minimum viable product) is the simplest version of a product that real customers will use and pay for, launched to learn quickly what to build next.

N

Niche: A niche is a narrow, well-defined market, such as AI tools for real-estate agents or for dentists, where a focused product can beat general tools.
NLP (Natural Language Processing): NLP (natural language processing) is the field of AI concerned with understanding and generating human language, which today is dominated by large language models.

O

OAuth: OAuth is the standard that lets a user grant an app limited access to an account on another service without sharing their password, as in "Sign in with Google" or connecting an MCP server.
Object storage: Object storage keeps files such as uploads and generated images in a scalable cloud bucket, like Amazon S3 or Cloudflare R2, instead of on the app server’s disk.
OCR (Optical Character Recognition): OCR (optical character recognition) extracts text from images and scanned documents, so that photos, scans and image-only PDFs can be searched, edited or read by an AI model.; Read the guide
One-time payment: A one-time payment is a single charge with no renewal, used for credit packs, lifetime deals or one-off services.
Open-weight model: An open-weight model is an AI model whose trained weights are published, so anyone can download it and run it on their own hardware or through any hosting provider.; Read the guide

P

Payment gateway: A payment gateway is the service that securely takes a customer’s card or wallet payment online and passes it to the banks for approval, such as Stripe, PayPal or Razorpay.; Read the guide
Payout: A payout is money sent out of a business, such as gateway settlements to your bank or commissions paid to affiliates.
PCI DSS: PCI DSS is the security standard for handling card data; using a gateway’s hosted checkout or payment fields keeps card numbers off your server and most of the burden with the gateway.
PHP: PHP is a widely used server-side programming language for web applications, supported by almost every web host.
Plugin (extension): A plugin is an add-on package that extends an application with new features or integrations without changing its core code.; Aikeedo marketplace
Product-market fit: Product-market fit is when a product clearly satisfies a market’s need, shown by users who stay, pay and recommend it without being pushed.
Prompt: A prompt is the input you give an AI model, the question, instruction or description, that tells it what to produce.; Prompt library
Prompt caching: Prompt caching lets a provider reuse the processed start of a prompt that repeats across requests, such as long instructions or a document, and bill it at a lower rate.
Prompt engineering: Prompt engineering is the practice of writing and testing prompts, with clear instructions, context, examples and output formats, to get reliable results from a model.
Prompt injection: Prompt injection is an attack where text hidden in user input, a web page or a document tries to override a model’s instructions, for example to leak data or misuse tools.
Prompt template: A prompt template is a reusable prompt with blanks, such as product name or tone, that users fill in through a simple form instead of writing the prompt themselves.; Writer templates
Proration: Proration adjusts a subscription charge when a customer changes plan mid-cycle, crediting the unused part of the old plan against the new one.

Q

Quantization: Quantization stores a model’s weights at lower precision, for example 4-bit instead of 16-bit, so it needs less memory and runs faster with a small loss in quality.

R

RAG (Retrieval-Augmented Generation): RAG (retrieval-augmented generation) is a technique where relevant passages are first retrieved from a knowledge base and then given to an AI model with the question, so the answer is grounded in that content.; Read the guide
Rate limiting: Rate limiting caps how many requests a user or key can make in a period, protecting a service from abuse and runaway costs; AI providers also rate-limit your account.
Reasoning model: A reasoning model is an LLM trained to think through a problem step by step before answering, which improves results on maths, code and planning at the cost of more time and tokens.; Read the guide
Receipt: A receipt confirms a payment has been made; unlike an invoice, it is not always enough for a business to reclaim tax.
Recurring billing: Recurring billing charges a customer automatically on a schedule, monthly or yearly, using a saved payment method, which is the basis of any subscription business.
Referral program: A referral program rewards existing customers for bringing in new ones, with credits, discounts or cash, so happy users become a growth channel.; Referral program
Refund: A refund returns all or part of a payment to the customer at the seller’s choice, which is far cheaper than the customer forcing a chargeback through their bank.
Reseller: A reseller sells another company’s product to its own customers, often under its own brand and at its own prices, keeping the margin between cost and price.
REST API: A REST API is a common style of web API where resources are reached at URLs and changed with standard HTTP methods such as GET and POST, usually returning JSON.; For developers
Retention rate: Retention rate is the share of customers still active after a given time; it is the opposite side of churn.
Reverse charge: Reverse charge is a VAT rule for cross-border business sales where the seller charges no VAT and the business buyer accounts for it themselves, using their VAT number.
Role-based access control (RBAC): Role-based access control gives each user a role, such as owner, admin or member, and decides what they can see and do from that role.
RTL (right-to-left): RTL (right-to-left) support lays out an interface for languages written from right to left, such as Arabic, Hebrew and Persian.; Multilingual

S

SaaS (Software as a Service): SaaS (software as a service) is software delivered over the internet by subscription, run by the provider, so customers use it in the browser without installing anything.
SaaS boilerplate: A SaaS boilerplate is starter code with the common parts of a SaaS already built, such as sign-up, billing and an admin area, so developers can begin from there instead of from scratch.; AI SaaS boilerplate
Sales tax: Sales tax is a tax collected at the point of sale; in the US it is set by states and local areas, and many states now tax SaaS once a seller passes a sales threshold there.
Self-hosted software: Self-hosted software runs on a server you control, your own or one you rent, instead of on the vendor’s cloud, so you own the installation and the data.; Read the guide
Semantic search: Semantic search finds results by meaning rather than exact keywords, by comparing embeddings of the query and the content.
SMTP (transactional email): SMTP is the protocol apps use to send email; transactional emails such as sign-up confirmations, password resets and receipts are usually sent through an SMTP provider.
Social login: Social login lets users sign up and sign in with an existing account, such as Google, GitHub, Facebook or LinkedIn, instead of creating a new password.
Speech-to-text (transcription): Speech-to-text, or AI transcription, converts spoken audio into written text, often with timestamps, punctuation and translation.; Read the guide
SSL/TLS certificate: An SSL/TLS certificate enables HTTPS, encrypting traffic between the browser and the server; free certificates are available from Let’s Encrypt.
Streaming responses: Streaming sends a model’s answer to the user token by token as it is generated, so text appears straight away instead of after the whole reply is finished.
Structured output: Structured output makes a model reply in a fixed machine-readable format, usually JSON that matches a schema, so apps can use the result without parsing free text.
Subscription: A subscription is an agreement to pay a recurring fee for ongoing access to a product, renewing automatically until it is cancelled.; Plans and credits
System prompt: A system prompt is the hidden instruction an app sends to a model before the conversation, setting its role, tone, rules and what it should or should not do.

T

Tax ID (VAT number): A tax ID, such as an EU VAT number, identifies a business for tax; customers enter it at checkout so it appears on invoices and, where it applies, VAT is reverse-charged.
Temperature: Temperature is a setting that controls how random a model’s output is: low values give focused, repeatable answers, high values give more varied and creative ones.
Text-to-image (AI image generation): Text-to-image is AI that creates a picture from a written description, the prompt, and often also edits existing images from instructions.; Read the guide
Text-to-speech (TTS): Text-to-speech (TTS) is AI that turns written text into natural-sounding spoken audio in a chosen voice and language.; Read the guide
Text-to-video (AI video generation): Text-to-video is AI that generates a short video clip from a written prompt, and image-to-video does the same starting from a still image.; Read the guide
Tokens: Tokens are the small pieces of text an AI model reads and writes, roughly four characters or three quarters of an English word each, and the unit most AI providers bill by.; Read the guide
Transformer: The transformer is the neural network architecture behind modern language models, which uses attention to weigh how every token relates to every other token.

U

Unit economics: Unit economics is the revenue and cost of a single unit, one customer, subscription or request, which shows whether growth makes the business more or less profitable.
Upsell: An upsell moves an existing customer to a higher plan or adds paid extras, such as more credits, more seats or premium models.
Usage-based pricing: Usage-based pricing charges customers by how much they use a product, such as messages, credits or minutes, rather than only a fixed fee per seat or per month.; Read the guide
User onboarding: User onboarding is the first-run experience that guides a new user to the moment they get value, which strongly predicts whether they stay and pay.

V

VAT on digital services: VAT on digital services is the value-added tax many countries charge on software and online services sold to consumers there, usually at the customer’s local rate, wherever the seller is based.; Read the guide
Vector database: A vector database stores embeddings and finds the ones closest to a query vector quickly, which makes it the search engine behind RAG and semantic search.; Read the guide
Voice cloning: Voice cloning is AI that creates a reusable synthetic copy of a specific person’s voice from a sample recording, which can then read any text aloud.; Read the guide
Voice isolation: Voice isolation uses AI to separate speech from background noise and music in a recording, leaving a clean voice track.; AI voice tools
VPS (Virtual Private Server): A VPS (virtual private server) is a rented virtual machine with its own resources and full control, the usual choice for hosting a self-hosted web application.; Install Aikeedo on a VPS

W

Webhook: A webhook is an HTTP request one service sends to another when something happens, such as a payment gateway telling your app that a subscription renewed or failed.
White-label software: White-label software is a product built by one company that another business rebrands and sells as its own, with its name, logo and domain, and without the original maker’s branding.; Read the guide
Workspace: A workspace is a shared space in a SaaS app where a team works together, with its own members, roles, subscription, usage and files.; Workspaces and teams

AI SaaS glossary

A

B

C

D

E

F

G

I

J

K

L

M

N

O

P

Q

R

S

T

U

V

W

Launch your SaaS

Latest