One developer wrote Photoshop in a weekend? Hold on. The truth about ArtCraft is more interesting

Over the past few days, a story along these lines has been doing the rounds on social media:

one experienced developer got fed up with Adobe, fired up Claude Opus 5.5 and recreated the entire Creative Cloud in a few days — Photoshop, Illustrator, Premiere, Lightroom, After Effects and the rest.

It sounds good. There is AI, David versus Goliath, expensive subscriptions and “one person did the work of hundreds of developers”.

The reality is a little less cinematic.

And at the same time — in my view — much more interesting.

Let us start with what is true

The project exists. It is called ArtCraft, and its so-called Crafting Apps are published as open source.

PhotoCraft is an image editor inspired by Photoshop workflows, VectorCraft targets vector graphics, FilmCraft video editing, LightCraft RAW processing, EffectCraft motion/VFX, DesignCraft desktop publishing and PdfCraft working with PDFs.

They really are written from scratch in Rust, and the official project website emphasises that they are native applications: no Electron and no webview. The code is public, under MIT/Apache 2.0 licences.

The official application list:
https://getartcraft.com/apps

That alone is interesting.

The pace is even more interesting.

PhotoCraft's public repository history starts on 30 September 2026, the first version, 0.1.0, appeared on 2 October, and by 18 October the project had reached version 0.10.0.

Repository:
https://github.com/storytold/photocraft

Releases:
https://github.com/storytold/photocraft/releases

So: yes.

The pace is absurd.

But this is already where the first difference between reality and the viral headline appears.

“One developer” quickly stopped being one developer

Brandon Thomas did launch the project and used AI agents, primarily Claude Opus 5.5, to generate enormous amounts of code.

Once published, however, the project rapidly attracted a community.

More contributors, pull requests, fixes, tests and new features arrived. The latest PhotoCraft releases credit Brandon Thomas and dozens of other people involved in its development.

So the statement:

“one person wrote the whole of Adobe Creative Cloud”

makes a very good headline.

But a poor description of what actually happened.

Closer to the truth would be:

an experienced developer used AI agents to launch a series of very ambitious open-source reimplementations, which a rapidly growing community then began to develop.

Less clickable.

But more accurate.

Is it really “free Photoshop” already?

This is where it is worth looking at the project itself.

PhotoCraft's developers state plainly:

this is early alpha and not yet a replacement for Photoshop in everyday professional work.

They point to missing tools, weaker typography, gaps in professional workflows, no plugin compatibility and missing generative features, among other things.

PhotoCraft repository:
https://github.com/storytold/photocraft

That is a fairly significant detail that posts declaring “Adobe just died” sometimes mysteriously fail to highlight.

ArtCraft's official application list is even more telling.

It now includes more applications — WordCraft, GridCraft, DeckCraft, CADCraft, SoundCraft and FigCraft too.

But most have the status “In development”. Some are labelled “Early alpha”.

So we have installers and working programs.

We do not have a dozen mature alternatives to Adobe, Office, AutoCAD and Figma.

Those are two different things.

This is where the most interesting part begins

Because it is easy to swing to the opposite extreme:

“Oh, alpha. Nothing interesting, then.”

No.

This is very interesting.

PhotoCraft really does open and save PSD files, and it has layers, masks, blend modes, filters and a substantial part of a typical image-editing workflow.

All of this appeared at a pace that would have been hard to imagine just a few years ago.

And this, in my view, is where ArtCraft's real story lies.

Not:

“AI wrote Photoshop in a weekend.”

But:

“AI radically reduced the cost and time needed to build the first 50–70% of a very large system.”

That could matter much more.

The first 70% and the last 30% are not the same thing

Anyone who has spent a while building software knows this pattern.

Building a version that:

looks like a product,
starts up,
has the main screens,
delivers the basic workflow,
handles the most common cases,

is one thing.

Building a program that a professional will trust with a client's file representing several days of work is on an entirely different level.

Creative Bloq's hands-on tests offer a good example.

PhotoCraft opened a PSD and basic editing worked, but automatic selection, masking, healing and some filters still fell short of Photoshop.

LightCraft had problems with camera profiles and masking, among other things.

FilmCraft could handle a basic edit, but limitations still emerged in more professional workflows.

Test:
https://www.creativebloq.com/news/i-tried-those-free-ai-coded-adobe-creative-cloud-clones-and-they-have-some-big-problems

It is precisely those “boring” things, built over many years or decades, that account for much of professional software's value.

Not the start screen.

Not the toolbar.

Not even the first 50 features.

But thousands of edge cases, compatibility, stability, plugins, colour profiles, imports, exports, data recovery, performance on unusual hardware and confidence that the project will still open tomorrow morning.

Will AI kill expensive SaaS products?

My view: some of them could have a very serious problem.

But not because Claude will replace Adobe tomorrow.

The issue is that one of the most important barriers to entry is falling very quickly:

the cost of creating competing software.

Until recently, the answer to:

“Why does nobody just build a competitor?”

was very often:

“because you need to hire dozens or hundreds of developers and spend several years developing the product”.

ArtCraft shows that this argument is starting to weaken.

One very good engineer with AI agents can now reach, in an incredibly short time, a point that once required a small team.

They can then publish the project.

And the next day, a community from around the world is working on it.

But SaaS is more than code

That is why I would not bet on Adobe's, Microsoft's or the other giants' funerals just yet.

Their advantage is no longer solely that they own millions of lines of code.

It also includes:

ecosystems,
file formats,
integrations,
plugins,
documentation;
training,
user habits,
support,
brand,
distribution
and years of testing cases that a new program's creator does not even know about yet.

AI can dramatically compress development.

It does not automatically compress twenty years of user experience.

And that is exactly why ArtCraft is both impressive and a long way from being “the end of Adobe”.