Where Are the Vibecoded Photoshops?

I’d like to share my thoughts as someone who uses Python and Claude Code on a daily basis (I’ve been running a research codebase and trading bot for several months).

I generally agree with the comment that “architecture is the bottleneck,” but based on my own experience, I’d like to elaborate further.

I don’t think the issue lies in code generation capabilities. The code generated by LLMs is competent on its own; the real bottleneck is cross-cutting consistency, which I believe is the primary challenge for applications on the scale of Photoshop.

For example, when I had Claude perform the task of “adding a new order type” to my trading bot: -Implementation in the relevant file: 90% success on the first try -Compatibility fixes on the backtesting engine side: 60% success with no oversights -Cross-cutting concerns like logging, metrics, and notifications: 40% of these were missed

The missed parts pass both compilation and testing. I’ve experienced the most troublesome kind of failure: the code is broken in terms of specifications but cannot be detected mechanically.

Photoshop has an estimated tens of thousands of cross-cutting invariants. Every these tools must operate without conflict across all layer types, selection ranges, and color modes. However, reconciling all of this with a single LLM inference seems impossible with the current architecture.

In other words, the absence of a “vibecoded Photoshop” isn’t due to a superficial lack of capability in the LLM; rather, the current context window and attention mechanisms are structurally unsuited for maintaining global invariants. This may not be the kind of problem that can be solved simply by “scaling up.”

Conversely, the direction of “personalized bespoke small apps” pointed out by stevex has fewer cross-cutting invariants (since the functionality is localized) and aligns with areas where AI excels. My personal conclusion is that Photoshop and AI development are not competing; they are simply solving different problems.

Since these observations are based on Python-based projects, this cross-cutting failure pattern might be less pronounced in statically-checked languages like Java or Rust. I’d like to hear others’ observations on this.

I'd say that the main challenge for vibe-cloning big and known programs comes down to cost. Plenty of extremely competent devs and companies out there are porting FOSS programs from one language to another, while offloading all the heavy lifting to the AI models - but even then, it can cost a non-trivial amount of money, and we're still talking about programs which are 1/10th in size of the ones you've mentioned.

I'm not sure exactly how many lines of code say Excel is, but if we estimate 10 MLOC of closed-source, then it would take a serious pile of cash for some SOTA LLM to reverse-engineer it. Especially given the fact that we don't have the code at hand to port.

Of course, we could use some open version of it, as best approximation, but there's still a ton of reverse engineering involved, which will burn tokens like crazy.

So the question becomes: What incentives are there for lone devs, or smaller teams, to vibe-clone some of the products, if it is going to cost them a fortune to do so? These things need funding.

We don't have vibecoded Photoshop, but... I have a vibecoded hexeditor. Vibecoded debugger. Vibecoded small document writer. Vibecoded file browser. Vibecoded virtualization runner on macOS. All for my personal usage and not released anywhere.

Why would I release it? Everyone can vibecode their own.

The recent ChatGPT Images 2.0 has awe-inspiring image editing and composition capabilities. I can totally see what people mean when they say "Photoshop killer".

I was writing a comment about the durability of this app, but you beat me to it. Something tells me that the burden of maintaining this thing through various OS updates, security policy changes (from Apple and Google), new devices, etc. is going to be frustrating for him. It's great that he vibecoded something useful to him in this moment, but I do think these stories are "counting their chickens before they hatch" so to speak.