ChatGPT

What ChatGPT is

ChatGPT is OpenAI’s generative AI assistant. You write a request in a chat window and it replies with text, and depending on the plan it can also read files, search the web, write code and create images. On this site it counts as a chatbot and as one example of generative AI, the wider category of systems that produce new content instead of only sorting or finding existing content.

Its place among the assistants comes from where it sits. ChatGPT is a destination of its own. A member of staff opens it deliberately, in the same way they open a browser tab, and it is not built into the applications a business already runs. That single fact drives most of what is different about it, from how it handles your documents to how a business decides whether to adopt it.

The product and the Model are Different things

People often use “ChatGPT” to mean the whole idea of AI writing tools, and that hides a useful distinction. A large language model is the underlying technology, trained on a very large amount of text to predict and generate language. ChatGPT is a product built on top of one. It adds a chat window, memory of the conversation and the features OpenAI chose to ship.

Copilot, Gemini and Claude have the same relationship to their own models. That is why the four behave differently even though each rests on a language model. The model is only part of the product. Each vendor trains its own, tunes it differently and wraps it in different features, and the wrapper decides much of what a business actually experiences.

One name, several ways to reach it

ChatGPT has one name and roughly one shape. It is a standalone interface at chatgpt.com, or the same models called through OpenAI’s API and built into something else. Team, Business and Enterprise are tiers of the same product line and not separate platforms, which is simpler to reason about than a vendor with two products under one name.

The tiers differ in price and in what they promise. The Business tier is billed per seat at a published rate, with an annual commitment priced below the monthly one and a higher-capacity seat for heavy users. Enterprise carries no published rate, and OpenAI negotiates it with each buyer. The Gemini comparison sets out the tiers side by side, and it is worth reading before a business assumes the Business rate simply scales up.

What people use it for

The course outlines are the most concrete list on this site of what ChatGPT is used for in a working week. The practical blocks run through summarising text and reading PDFs, writing and reformatting text, rewriting and translation, and drafting emails and replies. They cover content work such as names and slogans, a product blog, marketing campaigns and building presentations, and management work such as agendas, meeting minutes and working from a transcript.

Research is its own block, with deep search and web search, reading tables and asking questions of a document. Coding is covered too, from explaining code to writing it. The syllabus also includes image creation, with DALL-E and Sora named in the Basics outline. None of this needs technical skill. Most of it is a well-written request, which is why the same courses spend so long on prompt engineering.

Getting Reliable output from it

A request that works once is not the same as a request that works every time. The Basics course teaches prompting styles and prompt types and works through examples, so that participants write structured prompts to a repeatable pattern. The Advanced course goes further with one-shot and few-shot prompting, creating and reusing a persona, and controlling the shape of the output.

Two features carry that reliability between conversations. Custom instructions are standing rules applied to every request, and memory is what the tool keeps from one conversation to the next. Both sit in the personalisation and settings part of the syllabus. A prompt that has proved itself is worth keeping, and the Basics course sends participants away with a prompt library to reuse.

Projects and Custom GPTs

Beyond single conversations, the syllabus covers two ways of organising repeated work. Projects keep related conversations and material together. Custom GPTs are assistants that hold a role and a set of rules, so the person using one does not have to write the brief each time.

The courses build several. A prompting assistant rewrites a message into a stronger prompt. An HR assistant answers from the staff handbook. An expense categorisation assistant sorts spending, a business communication coach reviews your writing, and a simulator lets someone practise a difficult conversation. The Advanced course also asks when a custom GPT is worth building and where a saved prompt is enough. A custom GPT is a small AI agent in the loosest sense, and the more it acts on its own with tools, the closer it moves to the fuller version that page describes.

Answering from your own Documents

By default ChatGPT answers from its training and from whatever you put in the conversation. It has no native awareness of your organisation’s documents. A business that wants it to answer from a policy, a handbook or a contract has to give it that material, which is grounding.

The Advanced course builds exactly that. It asks questions of long documents, compares and analyses them, and builds document-grounded question and answer on the organisation’s own files. It teaches why the source behind an answer matters more than the answer, and it works through real tender documents with a structured evaluation while keeping a person in the loop on the judgement. The result is a tool that answers from your material and shows where, which is a different thing from a tool that sounds sure of itself.

What it does with your Data

This is the question most businesses ask first, and the site answers it from the providers’ own statements. OpenAI states that ChatGPT Business, Enterprise, Edu and the API platform are excluded from model training by default unless an organisation explicitly opts in, and that this covers both requests and responses. It offers data residency in the EEA and Switzerland for ChatGPT Enterprise, Edu and the API, and it holds named certifications including a SOC 2 Type 2 report and a family of ISO standards.

None of that replaces reading the agreement for the plan in question. The ChatGPT and Copilot comparison points out that the API’s documented policy and a business tier’s customer agreement are separate documents, and that neither should be assumed to carry over from the other. Each ChatGPT course opens with security and the use of private data, along with safety rules for working in third-party tools, before any tool is opened.

The wider rule is drawn by the kind of data, not by the tool. Our article on why banning AI at work backfires makes the case that staff will use these tools either way, so a business does better to say what may be entered than to prohibit the whole category.

Where it goes wrong

ChatGPT can produce an answer that reads well and is wrong. This is hallucination, and it is a property of how generative AI works, not a fault peculiar to one product. The ChatGPT course covers hallucinations and how to prevent them, and the Advanced course teaches how to check what comes back from research before you use it.

The practical habit is small. Ask for the source, read the source and treat the first reply as a draft. Grounding narrows the risk by tying answers to a document, but it does not remove the need for a person to check anything that matters. The syllabus is candid about limits too, with a section on where ChatGPT’s own limits are, so that a team learns what it cannot do as well as what it can.

ChatGPT compared with other Assistants

The foundational courses set ChatGPT beside Gemini, Claude, Copilot, Perplexity, Mistral, Grok and DeepSeek, so participants can tell which suits which job. The site has a glossary page for several of them, including Claude, Gemini, Perplexity and Mistral Le Chat.

The most common comparison is with Microsoft 365 Copilot. Copilot sits inside Word, Excel, Outlook and Teams and answers from content the user already has permission to see. ChatGPT is a separate destination with more room to build on, and no native access to your files. Neither is better in the abstract. The question is which ecosystem a business already runs on and what the workflow needs the tool to do.

How it is taught and where it appears in an Audit

One course is built on ChatGPT. ChatGPT Enterprise Essentials runs over two half days, from what generative AI is to building a custom GPT. The first session covers the foundations, the wider chatbot landscape, ChatGPT Enterprise itself and prompting. The second is practical throughout: content creation, meetings, research, coding and custom GPTs such as an HR assistant and expense categorisation.

It sits in the wider training catalogue.

An audit meets ChatGPT as a choice, not a default. Which tool fits is the question the Identify step is built to answer, weighed against what a candidate process actually handles. Once a tool is chosen, a course on it turns the decision into something the people using it day to day know how to do.