Have you ever stared at another monthly software bill and wondered if there might be a better way? I certainly have. Just the other day I caught myself calculating how much those recurring charges add up over a year, and the number felt ridiculous. Then along comes a developer from Atlanta who decides enough is enough. Using an advanced AI coding assistant, he builds free, open-source versions of the most widely used productivity tools on the planet. First Adobe’s creative suite. Then, almost immediately, Microsoft Office itself. Word, Excel, PowerPoint – all released on GitHub for anyone to grab. Suddenly the idea of paying forever for basic document work starts looking a lot less inevitable.
One Engineer, Powerful AI, and a Bold Challenge to Subscription Software
Brandon Thomas, working under the company name ArtCraft, has pulled off something that feels almost unreal in its speed. Days after dropping early versions of seven Adobe applications, he turned his attention to the Office suite. The results appeared this week as WordCraft, GridCraft and DeckCraft. He makes no secret of leaning heavily on Anthropic’s Claude model to accelerate the work. And he is giving everything away.
I’ve followed plenty of open-source projects over the years. Most crawl along for months or years before anything usable appears. This one moved at a completely different pace. That speed alone changes the conversation about what a single skilled person can achieve when paired with modern language models.
From Adobe Recreations to Office Clones in Record Time
The Adobe chapter came first. Thomas used Claude and the Rust programming language to put together stand-ins for Photoshop, Illustrator, Premiere and several other well-known creative tools. Early reactions described the Photoshop version as glitchy yet genuinely functional. Thomas himself called the whole set early alpha, nowhere near finished. Still, the mere existence of working prototypes raised eyebrows across developer communities.
Then the Office tools arrived. WordCraft handles DOCX files for both reading and writing. ArtCraft claims it already covers roughly 87 percent of ribbon commands and sits at about 62 percent real feature parity. GridCraft, the Excel-style spreadsheet, reportedly implements 93 percent of the worksheet functions people actually use. DeckCraft, the presentation tool, reaches around 79 percent feature coverage and sits about 80 percent of the way toward a first proper alpha release.
Those percentages sound impressive until you dig into the caveats. Charts, SmartArt and certain embedded objects still disappear when documents move through WordCraft. Real-world business document validation has barely begun. Counting ribbon commands only proves a button can be clicked, not that the feature behaves exactly like the original. The project documentation is refreshingly honest about these gaps.
Why This Project Hits a Nerve Right Now
Timing matters. Microsoft has steadily increased the price of its consumer plans after folding AI features into the mix. What used to cost one amount now costs more, with a limited number of AI credits thrown in. Existing customers can sometimes stay on older pricing, but new sign-ups feel the difference immediately. The company counts tens of millions of individual and family subscribers. That is a massive recurring revenue stream.
Adobe faced its own pressure earlier when regulators challenged certain subscription practices. A large settlement followed. Against that backdrop, a free alternative that actually opens and saves real Office files starts looking more attractive than pure idealism.
Thomas frames the effort as a targeted protest against the subscription model itself. In my view, that framing lands because many people share the quiet frustration of paying indefinitely for tools that once came with a one-time license. Whether this particular set of clones becomes the long-term solution is secondary. The signal it sends is what counts.
My guy, I can code. In college I was building laser games on skyscrapers and cloning biological systems into plants. AI models enable me to work as an entire company.
That response came after some critics dismissed the Adobe work as mere AI-generated noise. Thomas pushed back hard, pointing to genuine technical background and claiming productivity that outpaces certain large tech firms on a revenue-per-employee basis. The claim is his own, of course. No independent audit has confirmed the exact figure. Still, the attitude is hard to miss.
Clean-Room Approach and the Legal Gray Area
ArtCraft describes the applications as clean-room reimplementations. The team says they worked from public specifications and observed behavior rather than original source code. Legal observers note that independent reimplementation without access to proprietary code can, in many jurisdictions, avoid copyright infringement. Everything still depends on the specific laws involved and how courts interpret similarity of look and feel.
So far no formal challenge from either Microsoft or Adobe has surfaced in the coverage. That could change. Large companies protect their ecosystems carefully. At the same time, the open-source licenses chosen – Apache 2.0 or MIT – make the code freely available for others to inspect, improve or fork.
I’ve watched enough software disputes to know that legal clarity often arrives years after the first public release. For now the projects sit in that interesting space where technical possibility races ahead of settled law.
What the Current Versions Actually Deliver
WordCraft can open and save standard DOCX files. Basic editing works. Many ribbon commands respond. Yet advanced layout elements, certain embedded objects and full fidelity on complex business documents remain incomplete. The roadmap openly lists these limitations.
GridCraft focuses on worksheet functions. Hitting 93 percent coverage of Excel’s function set is a solid achievement for a pre-alpha tool. Formula-heavy work already has a fighting chance. Full parity on charts, pivot tables, advanced data modeling and add-in ecosystems still lies ahead.
DeckCraft handles the core presentation flow. Slide creation, basic transitions and many formatting options appear present. Rich media embedding, complex animations and seamless collaboration features need more work before anyone would trust it for a high-stakes client meeting.
Download links on the project website have been removed. GitHub remains the primary distribution channel. That decision keeps the focus on the code itself rather than polished packaging.
- WordCraft currently sits near 62 percent real feature parity
- GridCraft claims strong coverage of common spreadsheet functions
- DeckCraft has reached a solid portion of presentation capabilities
- All three remain early-stage with documented missing pieces
- Code is openly licensed for community contribution
The Broader Shift AI Makes Possible
Perhaps the most interesting aspect is not any single feature percentage. It is the demonstration that one determined engineer, amplified by a capable coding model, can produce plausible competitors to mature commercial products in a matter of days. Traditional development cycles for tools of this complexity used to stretch across years and large teams. That equation has changed.
AI does not magically create perfect software. The resulting applications still show roughness. Bugs exist. Edge cases fail. Documentation must catch up. Yet the barrier to producing a working first version has dropped dramatically. That drop matters for anyone who has felt locked into a particular vendor’s pricing or feature roadmap.
In my experience following developer communities, this kind of project tends to attract two reactions. Some celebrate the disruption and start testing the tools immediately. Others dismiss the early builds as unfinished toys. Both reactions miss the longer arc. The first usable versions are rarely the final story. Community contribution, iterative improvement and real-world pressure often transform rough prototypes into something more substantial.
Community Reaction and What Comes Next
The project already has an active discussion space. Users have begun requesting additional tools, with CAD software among the most frequent suggestions. That demand signals real appetite for alternatives beyond the usual commercial offerings.
Thomas continues to emphasize the protest element. Subscription fatigue is real for many professionals and small teams. When the same company that once sold perpetual licenses now pushes continuous payments, some users inevitably look for exits. Free open-source options, even imperfect ones, give those users leverage.
Will these particular clones become daily drivers for large numbers of people? Probably not in their current state. The gap to full feature parity remains meaningful. Reliability for mission-critical work still needs proof. Collaboration features that many teams rely on have not reached maturity.
What has already changed is the perceived cost of experimentation. Building a credible alternative no longer requires the resources of a major corporation. That shift alone puts quiet pressure on incumbents to justify their pricing and lock-in strategies more carefully.
Looking at the Numbers Behind the Frustration
Consumer subscription counts for the major productivity suite run into the tens of millions. Cloud revenue continues to grow at healthy double-digit rates. Those figures explain why companies invest so heavily in keeping users inside their ecosystems. They also explain why a free alternative, even a partial one, attracts attention.
Price increases that accompany AI feature rollouts intensify the conversation. Some customers accept the higher fee for the new capabilities. Others feel they are being asked to pay more for tools they already used successfully. The existence of working open-source options gives the second group somewhere to look.
I keep coming back to the practical question. How much functionality does a person actually need for everyday work? Many users never touch the advanced corners of these applications. For those users, an 60-to-90 percent solution that costs nothing and runs locally can feel liberating.
Honest Limitations Worth Remembering
It would be misleading to claim the current releases replace commercial Office suites. They do not. Complex documents still lose information. Full behavioral parity remains distant. Performance, stability and polish all need further development. Security review of the open-source code will take time as more eyes examine it.
The clean-room claim itself has not been independently verified in public legal proceedings. Until that happens, residual risk remains for anyone who decides to use the software in sensitive environments.
Still, dismissing the entire effort as mere demonstration would miss the point. The projects prove that the technical barrier has fallen. Future iterations, whether by Thomas or by others who fork the code, can close more of the remaining gaps.
What This Means for Everyday Users and Teams
For individual freelancers and small businesses watching every expense, the appearance of free alternatives creates breathing room. Even if the tools stay imperfect for a while, they expand the set of realistic choices. That expansion can influence how commercial vendors price and package their own offerings.
Larger organizations will move more slowly. Compatibility testing, security policies and training investments make sudden switches costly. Yet even those organizations benefit when competitive pressure keeps feature development and pricing honest.
Developers themselves gain a new reference point. Seeing what one person plus AI can accomplish in days recalibrates expectations about project scope and timelines. Some will feel inspired. Others may feel pressure to adapt their own workflows.
I’ve found that the most lasting impact of projects like this often appears years later. The first wave generates headlines. The second and third waves, built by communities that improve the original code, sometimes produce the tools people actually adopt at scale.
The Bigger Picture of Software Ownership
At its core this story touches a longer debate about who controls the tools we use every day. Subscription models provide continuous revenue and continuous updates. They also create continuous dependence. Perpetual licenses once gave users a clearer sense of ownership. The pendulum has swung hard toward the subscription side. Projects that restore free, inspectable alternatives pull the pendulum back a few degrees.
Whether that pull becomes permanent depends on many factors. Code quality. Community energy. Legal outcomes. User willingness to accept incomplete features in exchange for independence. None of those factors is settled yet.
What feels settled is the demonstration itself. AI-assisted development has reached a stage where a motivated individual can produce working, file-compatible versions of major commercial applications in a remarkably short window. That fact alone reshapes the landscape.
Final Thoughts on Speed, Protest and Possibility
Brandon Thomas did not invent open-source software or AI coding assistants. He combined them with focused effort and a clear point of view about subscription fatigue. The resulting applications remain early. Important pieces are missing. Real-world reliability still needs to be proven at scale.
Yet the speed of delivery and the open licensing force a useful question. If one engineer can reach this stage so quickly, what becomes possible when more people apply the same approach? The answer is still unfolding. For now the code sits on GitHub, free for anyone to examine, run or improve.
I keep returning to that simple reaction Thomas posted when critics called the work AI noise. He can code. The models multiply his reach. The combination produces results that would have seemed improbable only a few years ago. Whether these specific tools mature into daily replacements or simply serve as proof of concept, they have already succeeded in one important respect. They make the idea of free, capable alternatives feel less theoretical and more immediate.
That shift in perception may turn out to be the most lasting contribution of the entire effort. Users who once accepted subscription pricing as the only realistic option now have a concrete example of something different. Developers who wondered whether AI coding tools could handle complex, multi-feature applications have a live case study. And the companies that built their businesses around continuous payments have one more reason to keep listening carefully to what their customers actually need.
The story is still early. More versions will appear. More contributors may join. Legal questions could surface. Performance will improve or stall. Through all of that uncertainty, one fact remains clear. The tools that shape how we write, calculate and present ideas no longer sit exclusively behind paywalls controlled by a handful of large firms. A different path has been shown. Whether enough people choose to walk it will determine what comes next.
