Glossary

AI in plain English

You do not need to speak fluent AI to work with us. Still, if you want the words behind the tools, here they are, defined the way we would explain them to a friend who runs a business.

Prefer the long read? Browse the guides.

Foundations

Artificial intelligence (AI): Software that handles tasks we used to think needed a person, like reading, writing or deciding, well enough to be useful.
Machine learning: A way of building software that learns patterns from examples instead of being told every rule by hand.
Model: The trained system that actually does the work, turning an input like a question into an output like an answer.
Large language model (LLM): A model trained on huge amounts of text that can read and write in natural language, the engine behind most business AI today.
Neural network: The layered math structure most modern AI is built on, loosely inspired by how brain cells connect.
Training: The one time, upfront process of teaching a model by showing it many examples until it gets good at a task.
Inference: The everyday act of using a trained model to get an answer, which is what happens each time you send it a request.
Prompt: The instruction or question you give a model, phrased in plain language, that shapes what it does.
Hallucination: When a model states something confidently that is simply wrong, which is why good systems check their work against real data.

Technology

Agent: A program that does a task on its own, deciding the steps, using tools and following up, rather than answering one question at a time.
RAG (retrieval augmented generation): A technique that lets a model look up your real documents before answering, so it speaks from your facts instead of guessing.
Embedding: A way of turning text into numbers so software can find things by meaning, not just by matching exact words.
Vector database: Storage built for embeddings, letting a system instantly find the most relevant piece of your knowledge for any question.
Fine tuning: Adjusting an existing model on your own examples so it fits your voice, format or task more closely.
API: A doorway one piece of software offers so another can talk to it, which is how we connect your tools together.
MCP (model context protocol): A shared standard that lets AI agents plug into tools and data safely, the way a USB port lets any device connect.
Token: The small chunks of text a model reads and writes in, and the unit most AI usage is measured and priced by.
Context window: How much text a model can hold in mind at once, which sets how much of your document or history it can consider.
Multimodal: A model that handles more than text, such as images, audio or documents, in the same conversation.

In practice

Automation: Letting software carry out a routine task from start to finish without someone doing it by hand each time.
Workflow: A repeatable series of steps to get something done, the kind of thing that is worth automating once it is clear.
Dashboard: A single screen that pulls numbers from your different systems into one live, readable view.
Integration: A connection between two of your tools so information flows automatically instead of being copied across.
Chatbot: A system that talks with customers or staff in plain language, answering questions and taking simple actions.
Copilot: AI that works alongside a person, drafting and suggesting while the human stays in charge of the final call.
OCR: Turning a scan or photo of a document into text a computer can read, search and act on.
Classification: Sorting incoming things, like emails or tickets, into the right category automatically.
Summarization: Condensing a long document, thread or call into the few lines that actually matter.
Lead scoring: Ranking incoming enquiries by how likely they are to become customers, so your team calls the best ones first.
Knowledge base: A single, searchable home for how your business works, that both people and AI can draw answers from.
Second Brain: Our name for a knowledge base an AI can search on your behalf, so the answer is always one question away.

Trust and law

Privacy and data protection: The rules and practices that keep personal information safe and used only for what it was given for.
PII (personally identifiable information): Any data that can identify a specific person, which deserves extra care in how it is stored and shared.
Data residency: Where your data physically lives, which can matter for legal and compliance reasons.
Audit log: A tamper resistant record of who did what and when, so every automated action can be traced.
Human in the loop: A design where a person reviews or approves the important steps, so AI assists rather than acts unchecked.
Bias: When a system treats some groups unfairly because of skew in its data, which is worth testing for and correcting.