Will AI make some software obsolete?

For the past few years, we have heard a lot about AI as a technology capable of replacing developers, search engines, graphic designers, and even the software we use every day. But perhaps there is a more interesting question to ask: will AI actually replace some software itself?

The answer is probably yes. Not all software, obviously: it is a very partial "yes." But if we look at the stock market, investors already seem to be betting that AI could make some of today’s software less essential.

AI will not replace your operating system or database overnight. However, it could profoundly change the way we create and use many tools.

When software becomes just an interface

Let’s take a very simple example: a tool for resizing an image. Today, you open the software, import your image, choose a width, optionally select an output format, and then download the result.

With an AI capable of directly manipulating files, you could simply write: "Resize this image to 1200 pixels wide and convert it to WebP."

In this case, the software has not really disappeared. Its interface has simply become much less important. The user describes what they want to achieve, and the AI takes care of the rest. This is probably one of the most interesting changes: a conversation can replace part of a graphical user interface.

Asking AI to create its own software

The most spectacular change may not be there. Instead of looking for software, we can now ask AI to create it.

Imagine that you need a very specific little tool: compare two CSV files, rename hundreds of photos, automatically convert XML files to JSON, or generate QR codes from a spreadsheet.

A few years ago, you had to find the right software… or develop it yourself. Today, many people can simply write: "Create a tool that compares two CSV files and only displays the differences."

Within a few seconds, AI can generate a small working application. For one-off needs, this is quite impressive. And you get a tool that is completely adapted to your needs, without features you don't need.

Small tools are the most exposed

There are countless software tools that perform a single, very specific task: convert a file, reformat text, extract information from a document, generate a regular expression, summarize text, or convert one data format into another.

For many of these uses, asking AI directly for the result can be faster than finding the right tool and learning how to use its interface.

This is actually something I already do. For a one-off transformation, I often ask AI to perform it directly, or to generate a small script adapted to my needs.

And some people go even further: they ask AI to create a small graphical interface with a few buttons, a form, and the necessary logic. Within a few minutes, they have a tool that does exactly what they need. In this case, the specialized software can sometimes become completely unnecessary.

Why it's not that simple

You might then think that software companies are doomed. In reality, it is much more complicated. Creating a prototype is one thing. Maintaining real software is another.

A software company has to deal with updates, bugs, performance, security, backups, compatibility between different platforms, user permissions, data imports, edge cases… and the list goes on. As someone who works for a software company myself, I can confirm that there is no shortage of work, despite the help of AI!

It is precisely all this invisible work that is difficult to reproduce with a simple request to an AI. A tool generated in a few seconds can solve a one-off problem. Software used by thousands of people for several years is a completely different story.

Some people hope to create custom software for their own needs. It is appealing, but they will then have to maintain it, fix bugs, and make it evolve. In the meantime, they simply won't be doing the work they are actually supposed to be doing.

On the other hand, a software company spreads these costs across all its customers: development, maintenance, updates, and support. It will therefore often be difficult to do better… and especially cheaper, by developing a tool alone for your own needs.

The reliability problem

There is an important difference between getting an answer and getting a reliable result. If I ask AI to convert a few lines of data, an error will probably be easy to spot. But with several million lines, checking every result becomes practically impossible, without even considering the cost and the amount of data involved.

Specialized software has a major advantage here: its behavior is deterministic. Given the same input, it produces the same result, which is essential in many professional contexts.

The higher the level of reliability required, the more traditional software retains its value. And above all, not everyone is going to create their own software with AI to replace a tool that only costs a few euros. Between development time, bugs, and maintenance, the cheapest solution is often still… the existing software.

Complex software is not going away

Imagine an accounting, CAD, or business management software. You cannot simply say: "Prepare the company's accounts for me."

Behind this simple sentence are tax rules, permissions, histories, validations, audits, and dozens of integrations with other systems.

In these areas, AI looks more like a new assistant than a replacement. It makes the software easier to use, but it does not replace all the infrastructure developed by its creator.

Conclusion

AI will very probably not replace software as a whole. However, it could reduce the need for many small tools by allowing users to create applications adapted to their own needs, sometimes within a few seconds.

But this does not mean that the software industry is disappearing. Designing a reliable, secure, maintainable product at a reasonable price, and capable of evolving for years, remains a challenge that AI does not solve on its own.

The real change may be elsewhere: tomorrow, we may download less software… because we will be able to build some of it ourselves.