AEO (or Answer Engine Optimisation)
Optimising content so AI assistants, chatbots, and answer engines can cite your site accurately. Overlaps with SEO but adds structured data, entity clarity, and machine-readable files such as llms.txt.
Plain-English translations of the terms, languages, and platforms we use in app and AI development. Search or browse A–Z.
Optimising content so AI assistants, chatbots, and answer engines can cite your site accurately. Overlaps with SEO but adds structured data, entity clarity, and machine-readable files such as llms.txt.
A modern project management approach to quickly and efficiently implement complicated development products. Project members are given specific tasks, such as "Scrum Master", and project deliverables are broken into lots of "Sprints", each with their own clear deliverables.
Computer systems capable of tasks that historically required human judgment: reasoning, decision-making, and problem-solving. Today this includes generative AI (text, images, audio, video) and agentic systems that plan, call tools, and complete multi-step workflows.
Software that uses a language model to pursue a goal: it can plan steps, call APIs or tools, and iterate until the task is done. A single agent handles one focused job; multi-agent setups split work across specialised agents.
Using AI tools (for example Cursor, Claude, or ChatGPT) in planning, coding, review, and testing so teams ship faster without cutting quality. At ADS this is a delivery method, not a marketing label: we measure outcomes, not token counts.
Google's open-source operating system is used by pretty much all smartphones other than Apple including Samsung, Huawei, HTC, Nokia, Sony, LG etc.
A term often used by developers to impress and confuse clients. It stands for "Application Programming Interface" and is essentially the set of rules one system requires to work with another system. For example: "if you want the answer to this question, you need to ask me in this way and I will answer it in this way". Includes many different types like Push and Get calls.
The file format of compiled Android apps, so APK files are effectively Android apps. To install APK files outside of Google Play users have to grant additional permissions in their phone settings.
Using software (such as a smartphone app) to add layers, for example, 3D or interactivity, on top of the "real world" as seen through the phone's camera. For example Snapchat's animated photo effects or Pokemon Go. Increasingly popular. Similar to and often confused with Virtual Reality.
App Store Optimisation. Like SEO but for the App Stores. The science of making sure your app ranks as highly as possible on the app stores by using things like app name, keywords, description, screenshots, and preview videos.
A cloud-based solution developers use to store, manage and update code and code versions. Similar to GitHub.
A generative AI tool from OpenAI, widely used as a general-purpose large language model trained on a very broad dataset. Useful for drafting, coding help, and exploration; production apps usually need grounding, guardrails, and your own data.
AI that interprets images or video: detecting objects, reading documents, or scoring damage from photos. Used in products such as i-DENT-ify (vehicle damage assessment) and document parsing pipelines.
How much text (prompt plus conversation plus documents) a language model can consider at once. Larger windows help with long specs or RAG chunks; they also cost more per request. Design workflows around what actually fits.
See hybrid.
Small server-side code that runs close to the user (at the CDN edge) rather than on a central server. Common on Vercel and Supabase for auth checks, webhooks, and lightweight API logic with low latency.
When an API interacts with another system, the touchpoints of this communication are considered endpoints.
A software engineer embedded inside a client's organisation to design and ship solutions against that client's real workflows, data, and tools, rather than only delivering from a remote project team. At App Developer Studio, forward-deployed engineers sit with enterprise teams to apply AI directly to operational bottlenecks.
Training an existing language model further on your own examples so it behaves more predictably for your domain. Useful when prompts alone are not enough; it adds cost and maintenance compared to RAG or careful prompting.
A project quoted as one agreed total for a defined scope (often an MVP). Contrasts with hourly billing. ADS publishes fixed-price MVP estimates for common app and AI product types; custom scope varies.
A relatively mature, well-documented and widely-used framework built by Google and written in the Dart language to work across Android and iOS (Apple) devices. Pretty much combines the best of hybrid (simplicity, cost, time) with the best of native (performance and UI).
FlutterFlow is a low-code visual app UI builder that also enables easy integration with backends and third-party platforms, and custom coding. Whilst it can't always do everything app development requires it does a lot of it and totally rocks.
The approach of offering a free, "lite" version of the app that the user needs to pay (see IAP) to normally access additional content, value-added services or to remove adverts.
A cloud-based solution developers use to store, manage and update code and code versions. Similar to BitBucket.
When a language model states something plausible but wrong: invented facts, fake citations, or unsafe instructions. Mitigations include RAG over trusted sources, human review, and clear UI when answers are uncertain.
A broad term to cover programming languages that are designed to work across iOS (Apple), Android and often web. It's a great idea: one code base means reduced time, cost, chance for errors to both build and maintain. But the downside is often reduced stability, performance and security and there are many hybrid languages and some are better than others. Popular hybrid languages include Flutter, Cordova, React Native, Xamarin and PhoneGap.
A paid product (once off or subscription) that is made from within a free app through the App Stores using the payment the user loaded with the store. The App Stores will take a 30% commission.
Apple's operating system for iPhone and iPad. Major updates are usually once a year.
LLMs are machine-learning models that can comprehend and generate human language text. They work by analyzing massive data sets of language to predict, based on the text written, the most likely meaning of prompts and generate replies using the same data set.
A plain-text file at /llms.txt (and often a fuller /llms-full.txt) that lists what your site offers, written for AI crawlers and answer engines. It complements sitemaps and structured data by giving models a curated map of your services and key pages.
Low code development is a broad term for the space between no-code development and native/building-from-scratch development. Depending on the platform, it can address some of the issues of no-code development, but generally requires the support of a development team to complete the more complex tasks. FlutterFlow, Retool and Bubble are good examples of low-code platforms that we like.
The number of distinct users who open your app or product at least once in a calendar month. Hosting bands and infrastructure costs often scale with MAU and how data-heavy each session is.
An open standard for connecting AI assistants to tools and data sources (repos, tickets, docs) in a consistent way. Think of it as USB for AI integrations: one protocol, many connectors. Used with IDEs such as Cursor and Claude Code.
Several AI agents working together, each with a role (research, planning, coding, review). ADS builds these for clients and uses a four-agent workflow in ChatStack for requirements and estimates.
An initial version of the app with just enough core features to appeal to early adopters and gain feedback on priorities for future development.
Code built in the phone's own operating language. For iOS (Apple) this is Swift and Android it is Java or Kotlin. However increasingly it's used to mean "proper" apps, downloaded from the app stores and doing the things that only phone apps can do, rather than web apps or responsive websites.
A React framework for web apps and marketing sites, with server rendering, API routes, and strong SEO defaults. ADS uses it with Vercel for production deployments and structured data on key pages.
An approach to development that, as it states, doesn't involve actual coding. Typically it uses SaaS platforms and drag-and-drop interfaces to simplify the processes and open it up to non-technical people. However it often has limited flexibility, you can't access the code, you don't own the IP and it comes with license fees. But it can be useful for MVP.
Non-tabular databases that store data differently than relational tables. NoSQL databases come in different types based on their data model. They provide flexible schemas and are easy to create MVPs. Firebase is a good example.
For us, this is the opposite of a prototype. We use it where we're trying very new or complex technology or integrations. Rather than spending time and money designing something that might not be possible, we bite the bullet up front and take on the most challenging part of the project up front.
A powerful, open source object-relational database system suitable for complex backends that require their functions, services or integrations at scale. Supabase is a good example.
Designing instructions, examples, and constraints so a language model gives reliable outputs. Matters for chatbots, agents, and internal copilots. Not a substitute for good product requirements, but it reduces rework.
A pretty but dumb version of your app. So it looks great with the right designs, colour scheme, font, menus etc. but it doesn't actually do anything, although they can be installed on a phone with an icon and buttons etc. can be tappable to navigate around it. A very important step we do before we start to code (the clever but ugly bit!)
Those pop-up notifications that smartphone users receive on some apps, for example when new WhatsApp's are received.
Progressive Web Apps are somewhere between websites and native apps. They should work offline and be able to do some things native apps can do, for example receive push notifications, use GPS and interact with the camera. But exactly what they can do differs between Apple and Android, and between for example Android 10 and 6. The nice thing about them is they can be downloaded outside of the App Stores so users can just download from any website. And they avoid that dreaded Apple approval process!
Fetching relevant documents or database rows at query time, then asking the language model to answer using only that context. Grounds answers in your data (policies, tickets, manuals) and cuts hallucinations compared to a model guessing from training data alone.
An open source, cross platform mobile framework developed by Facebook that can also be used for web development.
A fixed monthly fee for a set number of hours or capacity, common for app maintenance, hosting support, and dedicated AI squads. Predictable for clients; keeps the same team familiar with the codebase.
Database rules (especially in Postgres/Supabase) that restrict which rows each user can read or write. Essential when AI or apps expose user-specific data: the server enforces access, not the client.
Whether money and time spent on a project pay back in measurable outcomes: cycle time saved, tickets deflected, revenue, or adoption. Enterprise AI consulting should define ROI before picking a model, not after.
Increasingly popular business model for software development companies where you don't own or control the code but pay a monthly or annual fee typically based on the number of users or amount of usage. Used in particular for common apps that are based on third party platforms.
Software Development Kit that companies provide that is typically embedded in the code of an app to allow your app to access their services.
Search Engine Optimisation. The science of making sure your site ranks as highly as possible in organic search on search engines, in particular Google, using keywords, tags, load speed, mobile-friendliness, descriptions and minimizing errors. Ultimately the quality and interest of the content to users (shown by low bounce rates, time spent on site, pages visit) are the most important factors over time.
A cloud-based platform used by over 800,000 businesses to quickly and easily develop world-class eCommerce websites. The Shopify ecosystem, with its own third party apps makes running a store simple, including payment and courier integration.
A popular project communication platform used by developers with different project chats and useful integrations. Think "WhatsApp" for desktop and geeks.
The combination of technologies, languages, tools and libraries that the project will be built on.
Speech-to-text (STT) turns voice into text; text-to-speech (TTS) turns text into spoken audio. Together they power voice-first AI agents, such as internal assistants staff talk to instead of typing.
Machine-readable labels embedded in a page (often as JSON-LD script tags) that describe entities, FAQs, services, and products. Helps search engines and answer engines show rich results and cite the right facts.
An open-source backend built on Postgres: auth, APIs, storage, and row-level security, with a generous free tier and predictable scaling. A common ADS stack pairing with Next.js on Vercel.
Everything it costs to own a product over time: initial build, hosting, maintenance, third-party fees, and team time. ADS's app cost calculator models a five-year TCO, not just launch cost.
Apple's app that allows developers to distribute test versions of apps that are not yet available to the public.
Beloved by designers with oversized glasses and questionable facial hair, these terms actually stand for User Interface and User Experience. Used almost interchangeably but in theory UI is the actual design of the screens/ pages users will be interacting with and UX is their overall experience (e.g. journey or flow) through the app. Very, very important in creating successful apps.
A cloud platform optimised for Next.js and front-end deployments: previews per branch, edge network, and serverless functions. ADS hosts marketing sites and many client web apps on Vercel.
Informal term for building software by prompting AI with minimal planning or structure. Fast for experiments; risky for production without requirements, tests, and review. ADS prefers disciplined AI-accelerated development over vibe coding for client work.
Using software (like AR) but also hardware (such as headsets) to create a completely immersive experience that replaces rather than builds on top of the "real world".
Well, wouldn't we like to know! It is often used as a fancy term by developers to cover up the fact that they've only built you a website designed for phones. In theory web apps should have some offline functionality and better performance and be able work with the hardware of the phone (e.g. photo upload, GPS). See PWA.
An HTTP callback: when something happens in System A (payment received, form submitted), System A POSTs a payload to a URL you configure in System B. Common for integrations, notifications, and serverless workflows.
We design, develop, and help you launch apps, websites, AI solutions, and bots.