Tech glossary

Showing two slightly different versions of a page or feature to different users at the same time to see which one performs better. Widely used by product teams to make evidence-based design decisions.

The specific conditions a feature must meet before it can be marked as complete. Written by the Product Owner, they answer the question: 'How will we know when this is done?'

An AI system that can take actions autonomously to achieve a goal — browsing the web, writing and running code, or calling external tools — rather than just answering a single question.

A way of working that breaks software development into short cycles called sprints, allowing teams to adapt and improve continuously rather than following a rigid plan.

A step-by-step set of instructions for solving a problem. Sorting a list, recommending a video, or detecting spam in an email are all algorithms in action.

A way for different pieces of software to communicate with each other. When an app shows you the weather, it's using an API to fetch that data from a separate service.

The world's largest cloud platform, offering hundreds of services — from hosting websites to running AI models. Most tech companies run part or all of their infrastructure on AWS.

Microsoft's cloud platform, popular with large enterprises and public sector organisations. Closely integrated with Microsoft products like Office 365 and Active Directory.

A paid job that combines on-the-job training with study towards a qualification. Tech apprenticeships go up to degree level (Level 6) and are a strong alternative to university.

A prioritised list of all the work a team plans to do — features, bug fixes, and improvements. The Product Owner decides what gets tackled first.

An intensive short-duration course (typically 3–6 months) designed to teach technical skills quickly. Many focus on web development, data science, or UX design.

An error or flaw in software that causes it to behave unexpectedly or incorrectly. Finding and fixing bugs is called debugging.

A set of practices that automatically test and deploy code changes, allowing teams to release software frequently and reliably without manual effort.

Using computer resources — storage, processing power, servers — hosted on the internet rather than on physical machines. AWS, Azure, and Google Cloud are the big providers.

The complete collection of source code that makes up a software project, usually stored in a version control repository like GitHub.

A map of every step a user takes when interacting with a product — from first discovering it to completing their goal. Helps teams spot pain points and opportunities to improve the experience.

A lightweight, portable package that bundles an application with everything it needs to run — so it works the same on any machine. Docker is the most popular container tool.

An organised collection of data that software can store, search, and update. Most apps — from social media to banking — depend on one or more databases.

The process of finding, understanding, and fixing errors (bugs) in code. A core part of every developer's day.

Releasing a version of software so that real users can access it, typically by pushing code to a server or cloud environment.

A structured five-day process (developed by Google Ventures) for solving a design or product problem — from understanding the challenge on day one to testing a prototype with real users on day five.

A set of practices combining software development and IT operations, focusing on automation, collaboration, and fast, reliable software releases.

A large body of work — too big for a single sprint — that can be broken down into smaller user stories. For example, 'User authentication' might be an epic containing stories for login, signup, and password reset.

Taking a pre-trained AI model and training it further on a smaller, specialist dataset so it performs better on a specific task — like a customer service bot trained on a company's own FAQs.

A large AI model trained on broad data that can be adapted for many different tasks. GPT-4 and Claude are foundation models — they underpin hundreds of different AI applications.

A pre-built set of tools, libraries, and conventions that helps developers build software faster. React is a popular web framework; Django and Rails are backend examples.

Frontend is what users see and interact with — buttons, pages, animations. Backend is the server-side logic that handles data, authentication, and business rules behind the scenes.

A developer who works across both the frontend (what users see) and the backend (servers and databases), covering the full technology stack of an application.

AI that can create new content — text, images, code, audio, or video — rather than just classifying or analysing existing data. ChatGPT, DALL·E, and Suno are all examples.

A version control system that tracks every change made to code. It lets teams collaborate, see the full history of a project, and safely roll back mistakes.

Google's cloud platform — home to services like BigQuery (data analytics), Vertex AI (machine learning), and Firebase (app development). Part of the 'big three' alongside AWS and Azure.

When an AI model confidently produces information that is factually wrong or completely made up. A key reason why AI outputs always need to be checked by a human.

The three core technologies of the web. HTML structures content, CSS controls how it looks (colours, layout), and JavaScript makes it interactive and dynamic.

Managing servers and cloud resources through code files (e.g. Terraform) rather than manual configuration — making infrastructure reproducible and version-controlled.

An open-source system for automatically managing, scaling, and deploying containerised applications across clusters of machines. Often abbreviated to 'K8s' and widely used in production systems.

A visual method for managing work using a board with columns (To Do, In Progress, Done), helping teams see workload at a glance and spot bottlenecks.

A free, open-source operating system that powers the majority of the world's servers, cloud infrastructure, and Android devices. Knowing your way around Linux is a core skill for DevOps and backend engineers.

A type of AI trained on enormous amounts of text that can read, write, summarise, and answer questions in natural language. ChatGPT, Claude, and Gemini are all LLMs.

A branch of AI where computers learn patterns from large amounts of data rather than following explicit instructions. Used in spam filters, product recommendations, and image recognition.

An architectural approach where an application is built as a collection of small, independent services that each do one thing and communicate over APIs — rather than one large, monolithic codebase.

The simplest version of a product that can be released to learn from real users. The idea is to build just enough to test your assumptions before investing in the full feature set.

A computing system loosely inspired by the human brain, made up of layers of connected nodes. The foundation of most modern AI, including image recognition and large language models.

Software whose source code is publicly available for anyone to use, modify, and share. Many essential tools — Linux, Python, React — are open source.

A fictional but research-based profile of a typical user — their goals, frustrations, habits, and context. Used by UX designers and Product Owners to keep real human needs at the centre of decisions.

The instruction, question, or text you give to an AI model to get a response. The wording of a prompt can dramatically affect the quality and accuracy of what the AI produces.

The emerging skill of crafting prompts that reliably get useful, accurate outputs from AI models. Techniques include giving examples, breaking tasks into steps, and specifying tone or format.

The point at which a product meets a real market need well enough that people actively want to use it and tell others. Often described as the most important milestone for any new product.

A high-level visual plan showing a product's vision, goals, and planned direction over time. It helps teams and stakeholders align on what's coming next.

An interactive, testable early version of a product or feature — more detailed than a wireframe but not yet the finished thing. Used to gather user feedback before committing to full development.

A proposal to merge code changes into a shared project. Other developers review the code before it's accepted, helping maintain quality and catch mistakes.

A popular, beginner-friendly programming language used widely in web development, data science, AI, and automation. Often the first language taught in schools.

Re-running existing tests after code changes to make sure previously working features haven't accidentally broken.

A storage location for a project's code, usually hosted on GitHub or GitLab. It contains the full history of every change ever made to the project.

A way of delivering software over the internet as a subscription, rather than installing it on your computer. Spotify, Notion, and Salesforce are all SaaS products — you pay monthly and access them through a browser or app.

A cloud model where you run code without provisioning or managing servers — you just deploy a function and the cloud provider handles the rest. You're charged only for the milliseconds your code actually runs.

A popular agile framework where teams work in sprints with defined roles (Product Owner, Scrum Master, Developers) and regular meetings like standups and sprint reviews.

A short, fixed period — usually 1–2 weeks — in which a team plans and completes a set of work. At the end of each sprint, the team reviews and demos what was built.

Anyone with an interest in or affected by a project — clients, end users, managers, and investors are all stakeholders. Managing stakeholder expectations is a key skill.

A short daily team meeting (usually 15 minutes) where each person shares what they did yesterday, what they're doing today, and anything blocking their progress.

The basic unit of text that an LLM processes — roughly a word or part of a word. A model's 'context window' is measured in tokens, which limits how much text it can consider at once.

The neural network architecture that powers most modern LLMs. The 'T' in GPT stands for Transformer. Introduced by Google in 2017, it revolutionised how AI processes language.

A UK technical qualification equivalent to 3 A-Levels, combining classroom learning with a 45-day industry placement. A strong alternative for students heading into tech.

An open-source Infrastructure as Code tool that lets you define and provision cloud resources (servers, databases, networks) using configuration files — so your entire infrastructure is version-controlled and reproducible.

A practice where developers write automated tests before writing the actual code, ensuring the code meets requirements from the start.

The visual elements of software that users interact with — buttons, menus, forms, layouts, and colours. Good UI makes software feel clear and easy to use.

An automated test that checks a specific, small piece of code works correctly in isolation — like verifying a single function returns the right result.

Watching real users interact with a product to identify problems and areas for improvement. A key research method for UX designers.

A short description of a feature written from the user's perspective, e.g. "As a user, I want to reset my password so I can regain access to my account."

The overall experience a person has when using a product — how easy, intuitive, and enjoyable it is. UX covers everything from research and design to usability testing.

A system that records changes to code over time so you can view the full history and revert to earlier versions if needed. Git is the most widely used version control system.

A simple, low-fidelity sketch or blueprint of a user interface, used to plan layout and content structure before detailed visual design begins. Often drawn in Figma.