What does that
actually mean?

The AI safety company that built and operates Claude. Anthropic was founded in 2021 by former members of OpenAI and focuses on making AI systems that are safe and easy to understand.

For your team: If your team uses Claude, Anthropic is the company behind it. Think of Anthropic as the manufacturer and Claude as the product.

A conversational AI tool made by OpenAI. You type a question or request, and it responds in plain language. ChatGPT is one of the most widely used AI assistants in the world.

For your team: Many staff members encounter AI for the first time through ChatGPT. It can help draft emails, summarize documents, answer questions, and much more.

An AI assistant made by Anthropic. Like ChatGPT, Claude responds to written prompts in plain language and can help with writing, analysis, and answering questions. Claude is often noted for being careful and thoughtful in its responses.

For your team: Some organizations prefer Claude for tasks that require careful handling of sensitive topics, such as donor communications or policy summaries.

Microsoft's AI assistant, built into products like Word, Excel, Outlook, and Teams. Copilot can help you draft emails, summarize documents, generate formulas, and more, directly inside the tools your team already uses.

For your team: If your organization uses Microsoft 365, you may already have access to Copilot. It is one of the most practical entry points for staff who are not yet comfortable with standalone AI tools.

Google's AI assistant, available at gemini.google.com and built into Google Workspace products like Docs and Gmail. Gemini can help with writing, research, and summarization.

For your team: If your organization runs on Google Workspace, Gemini is the natural AI assistant to explore first. It works directly inside the tools your team already uses every day.

The company that built ChatGPT and the GPT family of AI models. OpenAI is one of the most influential AI research organizations in the world and makes its models available through ChatGPT and its API.

For your team: When someone says they are using 'GPT' or 'ChatGPT,' they are using technology made by OpenAI. The company also licenses its models to other software products you may already use.

A formal document that tells staff how they are and are not allowed to use AI tools at your organization. It covers things like which tools are approved, what kinds of information can be shared with AI, and who is responsible when something goes wrong.

For your team: An Acceptable Use Policy is one of the first things every organization should put in place before staff start using AI on the job. It does not need to be long or complicated to be effective.

An AI system that can take a series of actions on its own to complete a goal, rather than just answering a single question. An AI agent might browse the web, send an email, update a spreadsheet, and write a report, all as part of one task, with little or no human input along the way.

For your team: Agentic AI is more powerful than a chatbot but also carries more risk, because it acts without asking permission at each step. For most nonprofits, the right approach is to understand agents before adopting them, not to avoid them entirely.

A process where an existing AI model is trained further on a specific set of examples so it becomes better at a particular task or learns a particular style. Fine-tuning is more involved than prompting and typically requires technical expertise.

For your team: Most nonprofits will not need to fine-tune a model. Getting better at prompting usually delivers the results you are looking for without the added cost and complexity.

A category of AI that creates new content, such as text, images, audio, or video, in response to a prompt. ChatGPT, Claude, and Gemini are all examples of generative AI tools.

For your team: When people talk about 'AI' in everyday conversation today, they usually mean generative AI. It is distinct from older AI applications like spam filters or recommendation engines.

When an AI generates a response that sounds confident and plausible but is factually wrong or entirely made up. The AI is not lying; it simply has no reliable way to know when it does not know something.

For your team: This is one of the most important things for any team to understand. Never rely on AI-generated facts, statistics, citations, or legal guidance without verifying them from a trusted source first.

The underlying technology that powers tools like ChatGPT and Claude. An LLM is a type of AI that has been trained on enormous amounts of text and learned to predict and generate language in a way that sounds natural and coherent.

For your team: You do not need to understand how an LLM works under the hood to use it effectively. It helps to know that these models generate responses based on patterns in their training data, which is why they can sound authoritative even when they are wrong.

A prompt is the instruction or question you give to an AI tool. Prompting is the act of writing those instructions. The quality of what you get back depends heavily on how clearly and specifically you ask.

For your team: Prompting is a skill your whole team can learn. Small changes, like adding context, specifying a format, or defining the audience, can dramatically improve the results you get.

The practice of crafting prompts in a deliberate, structured way to get consistently useful results from an AI tool. It goes beyond asking a question and includes techniques like giving the AI a role, providing examples, and breaking complex tasks into steps.

For your team: You do not need to hire a prompt engineer. Building a library of prompts that work well for your team's most common tasks, such as drafting grant updates or preparing board summaries, is something any staff member can do with a little practice.

The unit AI models use to process and generate text. A token is roughly three to four characters, or about three quarters of a word. AI tools have a limit on how many tokens they can handle in a single conversation, which affects how long a document or discussion can be.

For your team: You will mostly encounter this word when an AI tool tells you that a document is too long to process. If that happens, try summarizing the document first, or break it into smaller sections.

The large collection of text, images, or other content that an AI model learned from. The quality, diversity, and recency of training data shapes what the model knows and what biases it may carry.

For your team: Because training data has a cutoff date, AI tools may not know about recent events. And because training data reflects patterns in human-generated content, AI tools can sometimes reflect biases present in that content.

A casual term for using an AI tool to write code by describing what you want in plain language, without necessarily understanding the code it produces. The name captures the idea of steering by intuition rather than technical expertise.

For your team: Vibe coding can help non-technical staff automate small tasks or build simple tools, but it comes with risk if the resulting code handles sensitive data or automates critical processes. Have someone technical review anything mission-critical before it goes live.