“Software Is Over” Has Been Said Before. Microsoft and Adobe Are Still Here.

This week, Ars Technica covered developer Brandon Thomas's announcement that he had used Anthropic's Claude Opus 5.5 to build open source alternatives to seven Adobe products. Photoshop, Illustrator, Premiere, Lightroom, After Effects, InDesign, and Acrobat Pro. All written in Rust, released under MIT/Apache licenses, and described as "clean-room" replacements. His conclusion: "Software is over."

He also said he had "one-shotted Photoshop. The whole thing." From there, the predictions went on to replacing Google Search, building our own smartphones, and fixing the rest of the tech world.

Well, hold on. We've had free, open source alternatives to Microsoft and Adobe for most of my career. Those companies haven't gone bust. Before declaring software over, I'd want to understand why those alternatives haven't already done it.

We've had free alternatives for a long time

OpenOffice.org went open source in 2000. LibreOffice, its fork, has been around since 2010. GIMP has been a free alternative to Photoshop since the late 1990s. Inkscape and Scribus date to 2003, covering vector graphics and page layout. Then there's Krita for painting, darktable for photo editing, Kdenlive for video. We aren't short of choices.

And these are capable products. People do real work with them. Governments have run migrations to LibreOffice. If a small business can do everything it needs with free office software, why pay Microsoft? Use what works for the business.

But Microsoft still sells Office, and Adobe still sells Creative Cloud. Free has been available for a long time. There's more to this decision than the price of the license.

Now get the company to switch

Say you decide to replace the software at a company. The replacement is free, the features look good, and you have the source code. So far, so good.

Then a contract arrives as a .docx with tracked changes. A designer sends a layered .psd. The print shop wants a packaged InDesign file. Can the replacement handle those files correctly? "Mostly compatible" sounds okay until somebody has to spend an afternoon fixing a layout, then do it again the next time a file comes in. That person's time costs money too.

Then there are the people using it. Job postings ask for Photoshop, Excel, and Premiere by name. Schools teach those products. You hire someone who already knows how to do the work, and now you need to train them on a different tool. Maybe that's worth doing. But you have to account for it, including when you hire the next person.

And what happens when it breaks? Procurement is going to ask about support, security reviews, and who is responsible. An MIT license gives you the code and permission to use it. It doesn't assign someone to fix your problem. You can arrange support, or have your own people maintain it, but somebody has to take that job.

The integration is another big part of this. In a Microsoft environment, Office sits alongside Exchange, Teams, SharePoint, identity, and compliance tooling. Creative Cloud has fonts, stock assets, cloud libraries, and review workflows. A company may depend on several of those things without thinking much about them. Replace the app and you still have to figure out what happens to everything connected to it.

None of this means you shouldn't switch. It means you have to work out whether switching makes sense for the way your company actually works. The fact that an AI wrote the replacement doesn't answer that for you.

Seven alphas is still an accomplishment

Thomas has a decade of Rust experience, and he got alphas of seven apps out with AI. That's impressive. These are categories of software Adobe has spent years building with large teams. Getting a first version built for less time and money does change who can attempt something like this.

But how much of Photoshop is there? Every filter? Every keyboard shortcut? What happens with the files people have been creating for years? Those details matter when you're asking someone to replace a tool they depend on for work.

Thomas initially promised "100% [feature] parity within a month." A few days later, that became "99% parity will take a while, but I'm sure it'll be measured in months and not years." The project's own description is "super early alpha." Okay, then let's evaluate it as an early alpha. There's a lot of work between that and having replaced the whole of Photoshop.

I would also want to know who maintains these apps five years from now. Thomas says they'll be funded by selling access to Artcraft's AI image model, and that this won't be "some orphaned small-team project." That's the proposed funding model. If your business is going to depend on the software, you'll still want to know what support you can count on.

Oh, and Adobe can use AI too. It already has the file format knowledge, the customer relationships, and the people selling into agencies and enterprises. If development gets cheaper, that opportunity is available to the existing software companies as well.

This is related to what I wrote about internal apps built from a prompt. Once you have the software, you still own the work of supporting it and getting people to use it. That work needs to be part of the decision.

Where the incumbents should worry

The part of the Ars article that interests me most is why Thomas started. He says he got "bit by the dark pattern 'cancellation fee' one too many times."

Think about that. A customer got annoyed enough with the terms of leaving that he went and built alternatives to seven of your products. Whatever happens to these apps, that should get Adobe's attention.

Adobe's cancellation practices also prompted a 2024 lawsuit from the U.S. Department of Justice on behalf of the FTC. This isn't just one developer complaining about a subscription.

This is where I think AI does put more pressure on the existing vendors. Free alternatives have been around for decades, but a customer still needs a reason to go through the trouble of switching. Keep making that customer angry, and you're giving them one. If AI helps an alternative become good enough sooner, there's less reason for that customer to keep waiting.

I wouldn't declare software over because seven alphas appeared. But if I were selling software, I'd want to know why my customers stay. Making it difficult to leave doesn't mean they want to stay.