AI-Built Adobe Clones Fall Far Short of Promised Feature Parity

Brandon Thomas launched an ambitious open-source project to build replacements for seven Adobe applications using Claude Opus 5.5, but moved quickly to temper expectations. After the work appeared on Hacker News, he acknowledged being unprepared, stating the project is "not anywhere close to ready" for public use—a stark contrast to earlier announcements claiming near-complete parity within a month.

The Gap Between Promise and Reality#

The PhotoCraft repository demonstrates genuine technical scope: 2,152 commits in just two weeks, spanning 539,017 lines of Rust across 1,306 files. Yet the project's own roadmap reveals a sobering reality. While the clones cover roughly 76% of feature breadth across menus, tools, and panels, real-world readiness sits around 45%—43% for typical professionals, 61% for those using only core features.

Thomas estimates beta requires 1,100 to 1,800 more hours of work, with full parity taking 2,300 to 4,100 hours, measured as sequential wall-clock time for a single agent. The target is precise: Photoshop 2026 version 27.11.0. Thomas has since told Hacker News that "99% parity will take a while, but I'm sure it'll be measured in months and not years."

Critical Shortcomings in Production Use#

Early users identified fundamental gaps. The Photoshop clone lacks basic workflows—no keyboard shortcut for the eyedropper tool, rigid layout presets, and insufficient feature coverage. Photographers encountered serious issues with RAW processing pipelines, including noise and banding artifacts when adjusting shadows on Sony files, plus absent support for several Nikon bodies in Lightroom's replacement.

Perhaps most damaging: files exported from PhotoCraft don't reliably open in actual Photoshop. For software pitched as a replacement, file interchange failure is not a cosmetic issue. The roadmap itself counts 579 open issues from version 0.5 and 0.6 users, along with 22 of 25 missed performance budgets.

Engineering Approach and Controversy#

The team documented a "clean-room" methodology, studying Adobe applications for behavior and appearance while implementing from published specs (Adobe PSD format, ICC standards, ISO 32000 blend modes) and observation alone—never copying code or proprietary assets. The repository includes a panic_hunt fuzzing test for all commands and validates output against a corpus of Photoshop-authored files.