This vibecoded image editor cost $2,000 to build — was it worth it?

Generative AI is everywhere, whether it’s used as a cornerstone of a service, used to build apps, or employed to boost functionality within them. But with so many new products appearing all the time, which ones are worth checking out?

We’re here to help with a new regular series covering the best and freshest AI apps and services that you should know about.

Catch our previous edition of new AI apps and services here!

Photon Studio

Adobe Photoshop is arguably the industry standard for professional image editing, but perhaps its one issue is that access to it is now tied up with Adobe’s larger Creative Cloud subscription service. For many potential users, this is a massive investment for a product that they may not even use.

Photon Studio gives users another incredibly similar option. This vibecoded image editing suite is modeled on Photoshop, can open PSD files with layers intact, uses Photoshop’s default shortcuts to ensure a seamless workflow, and keeps and processes files locally.

The story behind Photon is perhaps as interesting as the app itself. It was vibecoded by a former Photoshop user seeking a viable alternative to Adobe’s dominance in the creative space. The project used GPT-6 Astra and initially required $2,000 worth of tokens — a number that has reportedly increased since it launched around two months ago. The end product, however, is a free photo editor that “supports 99% of Photoshop’s features,” per the developer.

As mentioned, Photon is free to use and is available on Windows, macOS, and Linux. There’s an optional one-time fee to unlock more related goodies.

There is no Android build at present, and there’s no indication that it’s in the development pipeline. Either way, it’s an impressive showcase of vibecoding’s possibilities when the human behind it has a clear vision.

Other new AI apps and services you should know about this week

There are plenty of other great new AI apps and services out there, including:

Archivist

My desktop and I really like the idea behind Archivist. In theory, the AI tool aims to automate the most monotonous and mind-numbing organizational necessity on a desktop: renaming files.

While many services already do this pretty well, Archivist uses AI to give files names that are more searchable and relevant to their contents.

It offers various naming presets that best suit the image’s end use. Users can then drop said images into these presets, and the tool spits out files with relevant names. You can then share all the files with a single link. It’s a nifty time-saving idea that’ll benefit those who process large volumes of media files daily.

Archivist is free to try on the web without an account, but to rename more than 50 files per month, you’ll need to set up login credentials.

Knotwork

There’s a lot to like about the hands-off nature of prompting a chatbot or agent. You call, it responds. Simple. But what if you want more control over this process, which models are used, and everything between your call and its response?

Knotwork is a nifty open-source Android AI agent. The difference from the likes of Gemini is that users decide which models it uses, how the agent uses them, and what output they want. It’s worth quoting the app’s F-Droid description to explain this pipeline in more detail.

Instead of a prompt box wired to somebody’s API, every conversation runs through a pipeline you can open and edit: a graph of typed nodes — input, on-device LLM, optional cloud LLM, tool call, routing, decomposition, evaluation, clarification, output. 

In a sense, this may be the perfect AI agent for those who don’t necessarily like AI itself. Knotwork is far more of a tinkerer’s tool, akin to Tasker, than any other assistant I’ve come across.

It’s available on Android through F-Droid and is free to use. You will need to provide your own API keys to access cloud or premium models.

Radaric

It’s been years, but I still occasionally mourn the loss of TweetDeck. It was the easiest way to keep tabs on the multiple Twitter streams I had running on my device, and made covering real-time content so much easier. Times have changed, and Twitter’s now X and TweetDeck is dead, but Radaric is perhaps the closest any product has come to recreating its essence.

Radaric isn’t necessarily an X client, though. It’s a broad content aggregator that pulls feeds from social platforms and RSS into its interface. Where it differs from the likes of TweetDeck, Inoreader, and Feedly is how it handles these information streams.

The product, as described on its website, “learns what you approve, reject, and explain.” By outlining their interests, users can essentially teach it, in plain language, to ignore certain content types and privilege others. This saves you from manually editing feeds or deleting stories. The web app also consolidates multiple sources into a single story, similar to Kagi News.

Access to Radaric requires an account and a $9/month fee.