AI just replicated the entire Adobe suite
AI just cloned the entire Adobe suite, and now it's targeting Office. Are closed-source software companies nervous?
Word, Excel, and PowerPoint—the three most widely used office software in the world—have been cloned by one person using AI, and all of them are open source.
The person behind this is Brandon Thomas, a former Square engineer who has been writing code for fifteen years. Last month, he first used AI to...
💡 What You Will Learn
AI just cloned the entire Adobe suite, and now it's targeting Office. Are closed-source software companies nervous? Word, Excel, and PowerPoint—the three most widely used office software in the world
📜 Table of Contents
AI Just Cloned Adobe's Entire Suite, Now It's Aiming at Office — Should Closed-Source Software Be Worried?
Word, Excel, PowerPoint — the three most widely used office applications in the world — have been cloned by a single person using AI, and all of it is open source.
The person behind this is Brandon Thomas, a former Square engineer who's been writing code for fifteen years. Last month, he used AI to clone seven Adobe applications in one go, racking up 40,000 GitHub stars. When netizens accused him of "just AI slop," he laid out his credentials and pivoted his sights to Microsoft: GridCraft for Excel, DeckCraft for PowerPoint, WordCraft for Word — all written in Rust, all generated by Claude.
Free alternatives come around every year. GIMP's been at it for nearly thirty years, and LibreOffice is a veteran too. So why is this worth talking about separately? Because the cost of cloning professional software has been shattered by AI — and this is the first time in thirty years that the software industry has seen anything like it.
Four Weeks, 550,000 Lines, Two People
Someone on Hacker News dug into the repo: two contributors, four weeks, one person contributing 550,000 lines of code. Three years ago this would've been unthinkable — an Adobe engineering director said publicly just two years ago that rewriting all of Photoshop was nearly impossible.
That statement wasn't wrong at the time. Rewriting mature software is hard because of the invisible work: color management pipelines, every single byte of file formats, edge cases accumulated over a decade. The UI is actually the easy part. This knowledge is buried in millions of lines of code and hundreds of engineers' heads — anyone who wants to clone it has to pay that same human labor bill all over again. AI rewrote the bill. Code went from a scarce commodity to something that can be mass-produced.
Open Source Lacks People — And AI Doesn't Mind the Dirty Work
Open source has been trying to replace closed source for twenty years. What's been the bottleneck? People. Engineers willing to spend a decade on dirty work like file formats and compatibility are rare. Between working for free on open source and earning a big salary at a big company, most choose the latter. AI has no career plan, doesn't want equity, and can be directed to implement Office's format specifications page by page. WordCraft can now read and write docx directly, with essentials like tables of contents, footnotes, and APA citations. Of course, it's still far from mature — Thurrott's assessment was blunt: "Not even close." But the direction is clear, and the community is already calling for CADCraft, targeting Autodesk.
Closed Source's Moat Was Never in Lines of Code
If you think closed source is doomed, here's a bucket of cold water. Code can be generated overnight; trust can't: enterprises dare to put financial models into Excel because they trust twenty years of stability — something a four-week-old alpha release can't offer. The ecosystem is the same: format compatibility is just the entry ticket. Plugins, collaboration habits, and someone to hold accountable when things go wrong — all of that lives outside the code. And one more thing: open source has historically had a higher probability of fizzling out than closed source. ArtCraft's hype died down, sponsorships stopped — who's going to fix the next bug? A top-voted Reddit comment put it well: open source alternatives were never in short supply — what's scarce is ones that can actually compete.
The Most Ironic Part: Using Closed-Source AI to Build Open-Source Software
Ars Technica reported that ArtCraft survives by selling tokens: its own closed-source vision model and IDE are billed by usage, and the revenue funds this open-source software. An open-source shell wrapped around a closed-source core — the actual means of production haven't been open-sourced at all. Code isn't valuable anymore; what's valuable now is data, compute, and trust.
My Take: Closed Source That Coasted on Easy Money Is Dead
The "good enough" tier will be rapidly devoured by "open-source shell + AI-built." Writing documents, making spreadsheets, editing vlogs — within five years, you probably won't need to pay subscriptions for these anymore. Anyone still coasting on revenue at this layer is in danger. The professional tier won't change: film-grade color grading, enterprise compliance, support systems with accountability — these moats live outside the code, and AI can't clone them. There's only one path left for closed source: earn back your subscription fee as value the user can actually see.
As for open source, what it may end up winning is shifting industry profits from "the people who write code" to "the people who have data and distribution." Whether that counts as a victory — and for whom — is hard to say.
Would you trust a four-week-old AI software with your docx files? Or would you wait and see — either for it to fizzle out or to grow up?
Written by our editorial team; tools listed here are tested or verified against public sources. Links point to official sites or GitHub repos for reference only — no paid placements.