In many marketing and development teams, Adobe Photoshop has been a fixture for years. It is therefore no surprise that projects promising an alternative attract close attention. One of these is Artcraft: a developer is working on an image editor that puts AI functions at its centre and is intended as a counter-design to the Adobe Creative Cloud. The tool is currently still in development.

The motivation behind it is not new. Criticism of high licence costs and of the cloud connection – that is, the coupling of software, account and file storage to the manufacturer's infrastructure – has accompanied Adobe for some time. The fact that a single developer is setting up his own product in this environment shows above all one thing: the technical hurdles for image editing software have fallen thanks to AI models.

What can be said with confidence – and what cannot

At this point in time, Artcraft is a project in progress, not a finished product. Anyone planning budgets or restructuring processes today should keep the two apart. Statements about the scope of functions, maturity, pricing model or availability can only be seriously assessed once a version exists that can be tested in day-to-day operation.

A different approach to such reports is therefore sensible: not as a purchase decision, but as an occasion to take stock of your own toolchain. Which tasks do teams actually perform in Photoshop? How many of them are pure routine – cut-outs, retouching, format derivatives, exports to web formats? And how many require in-depth design work?

Why alternatives are interesting from a business perspective

Licence costs are rarely the largest item in medium-sized companies, but they scale with team size. More relevant is usually the dependency: if file formats, brushes, templates, colour profiles and approval processes are tied to one provider, a switching effort arises that goes far beyond the monthly fee.

This is exactly where new tools come in. They do not set out to replicate every function of a program that has grown over decades, but cover the most common work steps – often faster, because AI models largely automate tasks such as masks or object removal.

Assessment criteria for evaluation

Anyone evaluating AI-supported image tools should look beyond the pure list of functions. The following questions have proven useful:

  • Data flow: Are images processed locally or transferred to a service? This is crucial for client material and unpublished product images.
  • Format compatibility: Can existing working files be opened, or does the library start from zero?
  • Export quality: Does the tool support modern web formats such as WebP or AVIF and clean colour profiles?
  • Automatability: Are there interfaces or command-line tools to integrate batch processing into existing pipelines?
  • Operating model: One-off purchase, subscription or usage-based billing – and how does this behave as the team grows?
  • Project maturity: Is there a viable model behind the software that makes maintenance plausible over years?

The last point deserves particular attention. A tool developed by a single person can be technically convincing and still be a risk for business-critical processes. For secondary strands – quick social media graphics, internal mockups, image optimisation – the risk is manageable. Different standards apply to the central production of campaign material.

Integration into web workflows

In web projects, image editing does not end at export. The content management system handles crops, responsive variants and compression – in TYPO3, for example, via image processing libraries and defined cropping variants. The more clearly this boundary is drawn, the less the overall process depends on a single desktop program.

A practical consequence: anyone setting up design processes today so that source files exist in open or at least widely used formats and automated steps take place outside the graphics program keeps their options open. Switching a tool later then becomes a question of familiarity, not a migration project.

AI functions need clear rules

Generative image functions add questions that purely technical comparisons cannot answer. Teams should define in which cases AI retouching is permissible, how altered images are documented and which material remains fundamentally unedited – for example in product images or in editorial contexts. Such guardrails are useful regardless of the specific tool and can already be formulated now.

Assessment

Today, Artcraft is above all a signal: the market for image editing is opening up because AI models take over tasks that previously required specialised software. For web and software projects, this means less cancelling licences at short notice than keeping processes portable – open formats, automation outside the graphics program, documented handover points to the CMS. Anyone who prepares for this can test new tools as soon as they are mature, without endangering existing production.

Sources