Add-in

An add-in is a supplementary program that extends another program with functions. The main program itself remains unchanged; it simply gets something added on. You can think of it like a bicycle pannier: the bike works fine without it, but with the pannier it can suddenly carry groceries. In the spreadsheet program Excel, for example, there are add-ins that install a new button in the menu bar. Clicking it then does something Excel cannot do on its own. Other programs call the same principle plug-in, extension, or add-on — practically the same thing is meant.

Why software today rarely does everything on its own

Large programs are used by millions of people, but everyone needs something different. A tax advisor wants different functions than a physics teacher. If the manufacturer built in every wish, the program would become bloated and slow. Add-ins solve this problem elegantly: everyone installs only what they truly need.

For companies, this gives rise to an entire industry. Around Excel, the Chrome browser, or the design software Figma, markets with thousands of extensions have formed. Small providers earn money without ever having to write a complete program of their own. For the maker of the main program this is attractive too, since their software becomes more useful without requiring more work on their part.

Currently the term is back in the news mainly because of artificial intelligence. Many AI providers do not deliver their technology as a standalone program but as an add-in for tools people already use daily. This lowers the barrier: nobody has to switch their working program.

The catch with hooking into other programs

A program that allows add-ins provides fixed docking points for this purpose. These interfaces precisely define what a supplementary program is allowed to do: which data it can read, where it may place buttons, when it gets called. The add-in then runs inside the main program and reacts to its signals.

A typical AI add-in in a word processor works like this: it reads the selected paragraph, sends it over the internet to a language model — that is, an AI system for text — and inserts the response back into the document. The actual computation happens in a remote data center. On one’s own computer there is only a very small connecting piece.

This is exactly where the biggest risk lies. An add-in can usually see everything in the document and send data outward. With confidential contracts or health data, this is sensitive. That’s why operators like Microsoft or Google review extensions before publication, and many companies only allow their employees a vetted selection.

Add-ins in school life and the working world

Anyone who has installed an ad blocker or a translation tool in their browser is already using this principle. In the browser it’s just called an extension instead of an add-in. In Word there are add-ins that manage citations, in Excel ones that automatically fetch stock prices.

In business news, the term often comes up in connection with payment models. An AI assistant is no longer sold as a separate product but as a monthly paid add-in to an existing subscription. For providers this is lucrative, because existing customers can easily be persuaded to spend more money.

A common misconception is that an add-in is an update. An update improves the main program itself and comes from the manufacturer. An add-in often comes from a third-party company and can be removed again at any time. If an add-in disappears, the main program keeps working, just without the extra function.