- Published on
Developer Behind the AI-Built Adobe Clones Says They Are “Not Anywhere Close to Ready
- Authors
- Name
- DeepHorus
Brandon Thomas put his AI-built open-source replacements for seven Adobe applications in front of Hacker News this week and immediately deflated them. "I wasn't ready to post this to HN," he wrote, adding that the project is "not anywhere close to ready" and that he would repost once it clears beta. The admission matters because announcement posts across five subreddits days earlier had promised a beta and roughly 100% feature parity within a month. The repository's own roadmap sides with the Hacker News version, not the marketing.
Thomas built the applications with Claude Opus 5.5 in Rust, compiling to WebAssembly for the browser, as open-source mirrors of Photoshop, Illustrator, Premiere, Lightroom, After Effects, InDesign and Acrobat Pro. Everything ships under MIT and Apache licenses. The timeline is startlingly short: the first commit landed September 30 and the latest on October 13, a fourteen-day sprint that produced an entire suite of professional creative tools rather than a single prototype.
PhotoCraft's ROADMAP.md, updated October 11, puts feature breadth at roughly 76%, derived from counts: 628 of 628 menus, 56 of 68 tools, 30 of 35 panels and 17 of 31 file formats. Then comes the number that matters. Readiness for real work is about 45% overall, 43% for a mainstream practitioner and 61% for someone using only core features. Beta is an estimated 1,100 to 1,800 Opus 5.5 agent wall-clock hours away, with full parity at 2,300 to 4,100; the target is named exactly as Photoshop 2026, version 27.11.0, measured from the installed application on October 10.
The roadmap is harder on the project than its critics are. Users of versions 0.5 and 0.6 have filed 579 open issues, 22 of 25 performance budgets are missed, and, in the document's own words, "PSDs we write are not yet reliably accepted by Photoshop." For software whose entire purpose is displacing Photoshop, broken file interchange is not a rough edge; it is a structural failure that no menu count can paper over.
The scale of the effort is real regardless. In fourteen days the repository absorbed 2,152 commits, 1,004 merged pull requests and 539,017 lines of Rust across 1,306 files. Git records 317 distinct contributor identities, so the "solo developer" framing collapsed almost immediately, and Claude appears as a named commit author 277 times.
People who installed the applications split along roughly the line the roadmap describes. Reddit commenter Painted-Arcana was blunt about the Photoshop clone: "It's bad. You can't even eye dropper by pressing alt. You can't change the layouts much beyond the static presets," adding that two minutes with the app made clear it is not a Photoshop replacement. Photographers hit the RAW pipeline hardest — Sea-Temporary-6995 reported noise and banding when lifting shadows on a Sony a7c file, and OspreyChaser found no RAW support for several Nikon bodies. The simplest app fared best: PDFCraft reportedly ran without bugs.
Little of the coverage noted how much process sits in the repository. AGENTS.md imposes a clean-room rule: the team studied Photoshop and other proprietary editors "for behaviour and look only," never copying code, shaders, profiles or assets, and implemented from public specifications such as the Adobe PSD spec, ICC profiles, ISO 32000 blend modes and papers. A panic_hunt test fuzzes every command with adversarial parameters, format work is validated against a corpus of Photoshop-authored PSDs, and the apps expose a JSON control channel and MCP server so the agent can drive the running application and screenshot its own interface work. Commenter kmeisthax rejected the framing outright: "There is no such thing as clean-room vibe coding," noting the doctrine normally assumes two separated teams writing specs and implementing blind.
The most measured reaction came from someone running the same experiment: commenter az_liberal_geek, building an open-source SketchUp replacement called Hew with Claude, described the very first version as by far the hardest part of the process. For now the project sits where its author placed it — an alpha with genuine momentum and an unproven core. Thomas has promised to repost at beta, and the roadmap's own 1,100-hour estimate suggests that reckoning is months, not weeks, away.
