Glossary
AI glossary in plain English
The words that come up when you start running real work through an AI, explained the way we would explain them to a friend. Each one says how we use it, with links to the longer notes.
- AI agent
- An AI agent is an AI model that can take actions, such as running tools, reading files or sending a draft, rather than only answering in a chat.
- AI reviewer
- An AI reviewer is a second AI, separate from the one that did the work, that checks it against written rules before it goes out.
- API
- An API (application programming interface) is the way one piece of software asks another for data or tells it to do something, without a person clicking through a website.
- Churn
- Churn is the share of subscribers who cancel in a given period. Monthly churn of 5% means about one in twenty subscribers leaves each month.
- Closed-loop automation
- Closed-loop automation is a scheduled job that does a task, checks whether it worked, and fixes or reports the problem without waiting for someone to notice.
- Context window
- The context window is how much text an AI model can keep in view at once: your instructions, the files you share and the conversation so far.
- Fine-tuning
- Fine-tuning is training an existing AI model further on your own examples so it picks up a style or task, rather than only following instructions.
- Hallucination
- A hallucination is when an AI model states something false as if it were true, such as a made-up number, quote, link or fact.
- Human in the loop
- Human in the loop means a person reviews or approves an AI's work at set points before it takes effect.
- LLM (large language model)
- An LLM is the kind of AI model behind tools like ChatGPT and Claude: it was trained on huge amounts of text and predicts useful text in response to what you give it.
- MCP (Model Context Protocol)
- MCP is an open standard that lets an AI assistant connect to outside tools and data, such as your email, calendar or files, through small add-ons called servers.
- Micro SaaS
- A micro SaaS is a small subscription software product, often run by one person, that solves one narrow problem for a specific group of customers.
- MRR (monthly recurring revenue)
- MRR is the total a subscription business can expect to bill each month from its active subscribers, with yearly plans divided by twelve.
- Operator file
- An operator file is a plain text file of standing instructions an AI agent reads at the start of every session: who it works for, what it may do and how work gets done.
- Permission lane
- A permission lane is a written rule for what an AI agent may do on its own, for whom and within what limits, such as "reply to existing customers about their own account, from our domain, logged."
- Prompt
- A prompt is what you give an AI model to work from: your instructions plus any examples, files or data you include.
- RAG (retrieval-augmented generation)
- RAG means looking up relevant documents first and handing them to the AI model, so it answers from your information instead of from memory.
- Scheduled job (cron)
- A scheduled job is a task a computer runs on its own at set times, such as every morning at 8 or every 10 minutes. "Cron" is the classic tool for it.
- Token
- A token is a small chunk of text, often part of a word, that AI models read and write in. Usage limits and prices are counted in tokens.
- Webhook
- A webhook is a message one service sends to another the moment something happens, such as "a payment just succeeded", so the other side can react right away.
Written by our AI operator and checked by our AI reviewer. Missing a term? Ask it in the community.