Claude and AI,defined plainly.

Building

An AI system that works towards a goal over several steps, deciding which tools to use and what to do next, rather than answering a single question. Good business agents run within clear limits and hand decisions back to a person.

Strategy

A structured review of where AI can help a business, team by team, ending in a prioritised list of opportunities, risks and first steps. It is the usual starting point for an AI strategy.

Strategy

A plan for where and how a business will use AI, tied to its goals: which opportunities to pursue first, which tools to use, how data stays safe and how the team will be trained.

Safety and data

A short document that tells staff which AI tools they may use for work, what data must never go into them and who checks the output.

Claude

The AI safety and research company that builds Claude. Anthropic offers Claude through its own apps, business plans and the Claude API.

Building · Application programming interface

A way for software to talk to other software. The Claude API lets developers build Claude into their own tools, websites and workflows, and is billed by usage.

Claude

Standalone pieces of content that Claude creates in a separate window alongside the chat, such as documents, code, diagrams or small interactive tools, which you can edit, reuse and share.

Building

Sending many API requests to be processed together, rather than one at a time, in exchange for lower cost and slower turnaround. Suited to non-urgent, high-volume work.

Claude

Anthropic's AI assistant, used by businesses for writing, analysis, research, coding and reasoning across long documents. It is available in apps for web, desktop and mobile, and through the Claude API.

Claude

Anthropic's agentic coding tool. It works in a codebase from the terminal, an IDE or the web, and can read code, run commands, make changes across files and prepare pull requests for engineers to review.

Claude

Anthropic's business plan for larger organisations and those with stricter security, identity, audit or contract requirements.

Claude

Anthropic's business plan for small and medium teams, with a shared workspace, admin controls, central billing and commercial data terms.

Building

A project memory file in a code repository that tells Claude Code how the project works: commands, architecture, conventions and anything that needs care. It is read at the start of each session.

Claude

A secure link between Claude and another tool, such as email, file storage, a calendar or a CRM, so Claude can find information and, where allowed, take actions there. Connectors work within each person's own permissions.

AI basics

The amount of text an AI model can consider at once, including your instructions, documents and the conversation so far. A larger context window lets Claude work with longer documents in one go.

Safety and data · Data protection impact assessment

A process under UK and EU data protection law for identifying and reducing the risks of processing personal data. It may be required before using AI on certain kinds of personal data.

Building

Further training a model on your own examples to change its behaviour. Most businesses get what they need faster and more cheaply with good instructions, Projects, Skills and retrieval instead.

AI basics

AI that creates new content, such as text, images or code, in response to instructions. Claude is a generative AI assistant.

AI basics

When an AI model states something that sounds plausible but is wrong or made up. The safeguards are grounding answers in your own sources and having a person check facts before anything is used.

Safety and data

A working rule that a person reviews and approves AI output before it is sent, published or acted on. It is the most important safeguard in a business rollout.

AI basics · LLM

The type of AI model behind assistants like Claude. It is trained on large amounts of text and predicts language, which lets it write, summarise, analyse and reason.

Building · Model Context Protocol

An open standard for connecting AI assistants to other systems, describing what the assistant can look up and which actions it may take. Many Claude connectors are built on MCP, and custom ones can be built for your own tools.

AI basics

A specific version of an AI system. Claude comes in a family of models: larger ones for harder reasoning and smaller, faster ones for simple, high-volume tasks.

Strategy

A list of every place AI could help a business, scored for impact and effort, used to decide what to do first. Quick wins are high-impact, low-effort opportunities.

Claude · Claude Projects

A shared workspace in Claude that holds a team's documents and instructions, so every conversation in it starts with the right knowledge and context.

AI basics

The instruction or question you give an AI model. Clear prompts with context, examples and the format you want produce much better results.

Building

An API feature that stores repeated context, such as long instructions or documents, so later requests that reuse it are faster and cheaper.

Safety and data

An attempt to hijack an AI system by hiding instructions in content it reads, such as a web page, email or document. It is guarded against with limited permissions and human approval for actions.

Strategy

An AI opportunity that delivers clear value with little effort or risk, such as drafting routine replies. Rollouts usually start with quick wins to build confidence.

Building · Retrieval-augmented generation

A technique where relevant information is looked up from your own documents or data and given to the model with the question, so answers are grounded in your sources.

Claude · Claude Skills

A package of instructions, and optionally templates, examples and scripts, that teaches Claude a repeatable process, such as writing a proposal your way. Claude uses a Skill automatically when a task calls for it.

Safety and data · Single sign-on

Signing in to many tools with one company account. On business AI plans it lets IT control access centrally and remove it instantly when someone leaves.

Building

Standing instructions given to a model before the conversation, setting its role, rules and tone. In custom tools, the system prompt is where most of the behaviour is defined.

AI basics

A small chunk of text, often part of a word, that AI models read and write in. API usage and context windows are measured in tokens.

Safety and data

The data used to train an AI model. On Anthropic's commercial plans, business conversations are not used to train models by default; always check the current terms for your plan.