AI terms, in plain English

Every term below comes with two things: what it actually means, and a sentence you can say out loud in a meeting without hedging. Free to read, no account needed.

• Compute

  Listen

  The digital horsepower—specifically the specialised computer chips and cloud servers—needed to train and run AI models.

  Say this in a meeting

  “We need to check our compute allocation with the cloud provider before we commit to deploying this model across the business.”

• Text-to-Speech

  Listen

  Text-to-Speech is an AI technology that converts written text into spoken words, like a computer reading an email aloud or providing navigation instructions.

  Say this in a meeting

  “For accessibility, we should investigate using Text-to-Speech to read out our product descriptions.”

• Small language model (SLM)

  Listen

  A compact AI model that runs cheaply on your own laptop or server — good enough for specific tasks without paying for a giant model.

  Say this in a meeting

  “For this use case an SLM is probably enough — we don't need frontier-model economics for a classifier.”

  emerging

• Computer use

  Listen

  Giving an AI the ability to see your screen and click, type, and scroll the way a person would.

  Say this in a meeting

  “Computer use means we can automate the tools we already have, without waiting for every vendor to ship an API.”

  emerging

• Model Context Protocol (MCP)

  Listen

  A standard way to plug your company's tools and data into any AI assistant, like USB for AI.

  Say this in a meeting

  “If we adopt MCP for our internal tools, any AI our teams pick up can use them on day one.”

  emerging

• Bot

  Listen

  A bot is an automated software program that performs specific tasks, often repetitive ones, that would otherwise be done manually, like responding to queries or collecting data online.

  Say this in a meeting

  “We need to consider using bots for our customer service channels to handle basic queries more efficiently.”

• AI Risk

  Listen

  AI Risk refers to the potential negative consequences or harms that could arise from the development or deployment of artificial intelligence systems, ranging from misuse and ethical concerns to system failures.

  Say this in a meeting

  “We need to ensure our AI strategy thoroughly addresses potential AI risks, not just technical challenges, to safeguard our operations and reputation.”

• Voice AI

  Listen

  Voice AI refers to computer systems that can understand and generate human speech, like when you speak to a smart speaker or use a transcription service.

  Say this in a meeting

  “For our customer service, we should explore Voice AI to handle common queries, freeing up human agents for more complex issues.”

• Speech-to-text

  Listen

  Speech-to-text is a technology that converts spoken words into written text, similar to how a human transcriber would type out what is being said.

  Say this in a meeting

  “We could use speech-to-text technology to automate meeting minutes or process customer service calls more efficiently.”

• AI image generation

  Listen

  Software that produces original photos, illustrations, or graphics from simple written descriptions.

  Say this in a meeting

  “Before we spend budget on bespoke stock photography, let's see if AI image generation can produce the campaign visuals we need.”

• Test-time compute

  Listen

  Letting the AI "think longer" at the moment you ask it a question, instead of just relying on how much it learned in training.

  Say this in a meeting

  “The gains here are coming from test-time compute — the model reasons for longer before answering, which is why quality is jumping without a bigger model.”

  emerging

• Reasoning model

  Listen

  A newer type of AI that pauses to work through a problem step by step before answering, like showing its scratch work.

  Say this in a meeting

  “For anything analytical we should default to a reasoning model — it's slower but the answers hold up under scrutiny.”

  emerging

• Agentic workflow

  Listen

  An AI that doesn't just answer — it actually takes steps for you, like clicking through a website or filing a ticket.

  Say this in a meeting

  “The next wave isn't chatbots, it's agentic workflows — the AI does the task end-to-end, not just drafts it.”

  emerging

• Artificial Intelligence (AI)

  Listen

  Software that can do things we used to think only humans could do — like understanding language, spotting patterns, or making decisions.

  Say this in a meeting

  “When we say AI, we mean systems that learn from data instead of being explicitly programmed for every rule.”

  marketingoperationssaleshrfinancelegalgeneral

• Machine Learning

  Listen

  Instead of writing rules by hand, you show the computer lots of examples and it figures out the patterns itself.

  Say this in a meeting

  “This isn't hard-coded — it's a machine learning model, so it improves as we feed it more of our own data.”

  general

• Large Language Model (LLM)

  Listen

  A very large AI trained on huge amounts of text that can read, write, and answer questions in plain language.

  Say this in a meeting

  “The tool is powered by an LLM, so it can summarize documents and draft responses in natural language.”

  general

• Prompt

  Listen

  The instructions or question you type into an AI so it knows what to do.

  Say this in a meeting

  “Let's tighten the prompt — clearer instructions usually give us more consistent output.”

  general

• Hallucination

  Listen

  When an AI confidently makes something up that sounds right but isn't actually true.

  Say this in a meeting

  “We need a human review step here — LLMs can hallucinate facts, especially about our internal policies.”

  legalgeneral

• Generative AI

  Listen

  AI that creates new things — text, images, code, audio — instead of just classifying or predicting.

  Say this in a meeting

  “This is a generative AI use case, so we should think about brand voice and IP review, not just accuracy.”

  marketinggeneral

• Chatbot

  Listen

  A conversational interface, often powered by an LLM, that people can talk to instead of clicking through menus.

  Say this in a meeting

  “Instead of another form, let's put a chatbot in front of the workflow so employees can just ask.”

  operationshrgeneral

• Model

  Listen

  The actual "brain" of an AI system — a file full of learned patterns that turns inputs into outputs.

  Say this in a meeting

  “The model itself is what we're licensing here; everything else around it is our own product.”

  general

• Natural Language

  Listen

  Everyday human language, like the sentences we're speaking now, as opposed to code or structured data.

  Say this in a meeting

  “The advantage is our team can query it in natural language — no SQL or filters required.”

  operationsgeneral

Got a meeting coming up? Paste your agenda into Meeting Prep and get the terms you'll actually hear.