A developer used AI to build Adobe and Office alternatives in a couple of months

Several years of work by a large team shrink to a couple of months of work by one developer with AI. That is the story Habr examines: the author takes on a suite of complex graphics and office applications meant to compete with Adobe and Microsoft Office. The original headline claims the giants have been "killed." That is a strong exaggeration: the announcement gives no project names, no user numbers, and no comparative tests.

What the article tells us

The available part of the article contains few facts, but they are telling:

• team: one developer and a small group of helpers;
• tooling: AI for generating code and, probably, a significant part of the interface;
• timeline: a couple of months versus several years under the traditional approach;
• goal: graphics and office applications comparable to Adobe and Microsoft products;
• distribution model: open source;
• questions the author promises to cover: legal risks, the impact on the future of development, and how close the apps come to commercial counterparts.

What is missing is missing: there is no data on support for PSD, DOCX, or XLSX formats, on performance with large files, or on the number of bugs. These are exactly the parameters that separate a demo from a replacement for a working tool. People value Photoshop and Office not for a set of buttons but for compatibility, handling of heavy documents, and decades of fixed bugs.

Where the legal line runs

Resemblance to the originals is the main risk here. Copyright protects specific code and presentation, but not ideas or functionality. In Google v. Oracle, the US Supreme Court ruled in 2021 that copying the Java API declarations was fair use, and in SAS Institute v. World Programming, the EU Court of Justice held that a program's functionality is not protected by copyright. Replicating Photoshop's capabilities is legal. Copying icons, names, signature interface elements, and recognizable design is already dangerous: trademarks and protection of appearance come into play.

AI-assisted development adds a second layer of problems. A model can reproduce fragments of code under copyleft licenses such as GPL, and a project risks becoming legally incompatible with its own license. In addition, under the approach of the US Copyright Office, code without substantial human contribution receives no protection. Such a project is harder to defend against forks and harder to license.

What changes for developers

The most interesting part of this story is not the specific applications but the price of entry. If a working prototype of an office suite can be built in months, the barrier is no longer writing code but quality checks, maintenance, and user trust. The first 80% of a product becomes cheap. The remaining 20%, including format compatibility, accessibility, security, and documentation, still takes time and people.

For Adobe and Microsoft, the real threat is not a single enthusiast but the prospect of many such projects. They have defenses: cloud services, plugin ecosystems, enterprise contracts, and the habits of millions of users.

How to read stories like this

If you are building a product, check three things. First, open real files from your own work and see whether the layout and formatting break. Second, study the project's license and the origin of its code, especially if you plan commercial use. Third, assess community activity beyond the creator: the number of contributors, how fast bugs get closed, and how regularly releases ship. Until that data is available, what we have is a convincing demonstration of the speed of AI-assisted development, not a replacement for Adobe and Office.