I’ve spent most of my working life as a technical writer, and most of that has been supporting enterprise software of one type or another. For the last couple of years I’ve been on the sidelines because venture capital has flooded into AI to the seeming exclusion of all else, and I have no direct AI experience. Now, thanks to AI, a large part of the software industry may be coming to an end.
First up: Eric S. Raymond:
The end of software secrecy is almost upon us. The death of closed source, and the final triumph of open source.
I should have realized this a few months ago when I successfully decompiled the game of Firefighter from an ancient DOS shareware binary. My robot friend was able to tell it had been written in Borland Pascal from looking at the data layout; it gave me back very readable Pascal code with sensibly chosen function and variable names. I transpiled it to Rust and now ship it as part of my heritage games collection.
The thing is, I thought of this as a fun stunt but didn’t have any confidence that the technique would scale up to large, real programs. I have since learned that people are now doing this sort of thing with entire AAA games, which are among the most complex software artifacts ever to be shipped as a binary. If they can be decompiled, anything can be.
This has implications. Massive implications. We need to think about what the world is like when, in general, there is no longer a secrecy moat around almost any software at all.
I say “almost” because there is going to be one way to maintain secrecy. Software as a service with the executable hiding behind a network connection to the cloud. But as of now, we must assume that anything shipped as an executable, or even a firmware image, is as transparent as glass. It won’t keep its secrets for any longer than it takes for somebody to be motivated to throw an LLM at it.
Snip.
There are still some obstacles. In US law, decompilation of a binary is a derivative work of the binary that falls under whatever copyright it had. However, it is also settled law that if you decompile binary code to a precise specification of what it does, then generate fresh code from that without looking at the decompiled stuff, you’re in the clear. This was the case law that allowed PC clones to exist, after Phoenix Technologies reverse engineered the original IBM PC BIOS.
The two-step process – binary to specification to unencumbered code – will no longer takes a large number of programmers and years of development time. With an LLM, it’s now a thing you can do in a day. Ubiquity will make it impractical to prosecute all the people who might skip the specification step.
Source code can also be covered by patents. You can be able to see their source code for a patented technique and not be able to use it without violating the law. Linux has evolved practices for dealing with this problem – one is not shipping patented video codecs, but requiring you to download them as plugins from jurisdictions where U.S. patents can’t be enforced. This presents patent holders with the impractical challenge of individually suing millions of end users, assuming it can even figure out who they are. To date, AFAIK, this has not been attempted.
I believe Raymond overstates the case a bit, as I bet AIs will find ways to obstuficate executables, but I bet that it will be true in enough cases to drain all the profit out of it for larger software companies that ship executables.
Forced migration from shipped binaries to tied cloud services will be tried – Adobe pioneered this, and Microsoft is pushing it as hard as they can given that their OS is a binary that has to run locally. Both companies are seeing massive user revolts over this. There is good reason to doubt that it’s a strategy that can hold customers in the long term.
However, most software these days doesn’t seem aimed at consumers, it’s aimed at other businesses, the bigger the better. I haven’t seen much if any revolt there. A whole lot of companies seem to have decided they don’t need to own their own software if they can simply have it in “the cloud” and pay a subscription that means they have lower startup costs and can offload the support headcount to a vendor. And it seems that increasingly, that software will be written by AI, tested by AI, probed for security holes by AI, documented by AI, and supported by AI. I wonder if companies will give their AI tech support Indian accents to make them more believable?
But Raymond and other Open Source advocates shouldn’t be so sure that this turn of events is good for their camp as well. Decompilation can just as easily be used to take Open Source executables and create private source clones of them in different programming languages for a variety of different purposes. Also, I suspect that using an AI to add features to Open Source products will be a lot cheaper than hiring an actual Open Source programmer. In fact, I would wager money it’s already being done.
But that’s not the only sign of AI’s threat to the existing software ecosystem. A man has cloned Adobe’s suite and is in the process of doing the same for Microsoft.
Days after unveiling free, open-source stand-ins for seven Adobe applications, Atlanta engineer Brandon Thomas has turned to Microsoft Office. This week he posted WordCraft, GridCraft and DeckCraft – replicas of Word, Excel and PowerPoint – to GitHub, Tom’s Hardware reported. Thomas says he used Anthropic’s Claude to help build the software and is giving the results away.
The Adobe set came first. Gizmodo’s Mike Pearl reported on Oct. 7 that Thomas had used Claude Opus 5.5 and the Rust programming language to recreate Photoshop, Illustrator, Premiere and four other Adobe applications. Thomas wrote on Hacker News that they were in early alpha and “not anywhere close to ready,” while Pearl called the Photoshop stand-in “glitchy, but real.” Adobe, Pearl noted, agreed to a $150 million settlement after federal prosecutors sued it in 2024, alleging it was “trapping” people in subscription plans.
The Office clones followed within days. Paul Thurrott reported on Oct. 9 that WordCraft reads and writes DOCX files, and that ArtCraft, Thomas’s company, calls it 87 percent complete for ribbon commands and about 62 percent complete for real feature parity. For GridCraft, a pre-alpha Excel stand-in, ArtCraft claims 93 percent of Excel’s worksheet functions. DeckCraft, the PowerPoint stand-in, is said to cover about 79 percent of the feature set and to be 80 percent of the way to a first alpha.
Thurrott also reported that ArtCraft has removed the download links from its website, leaving GitHub as the way to get the code.
Gadget Pilipinas took a closer look at the project’s documentation and found some important caveats. WordCraft’s roadmap says charts, SmartArt and embedded OLE objects are lost when a DOCX file is loaded and saved, and validation against real business documents had not yet begun. The site also noted that counting ribbon commands measures whether a feature can be invoked, not whether it actually behaves the way Word does. It listed the licenses as Apache-2.0 or MIT. Release numbers differ between the two outlets: Thurrott reported version 0.3.0, while Gadget Pilipinas cited 0.5.0.
The larger target is the subscription model itself. When Microsoft added Copilot to its consumer plans in January 2025, PCWorld reported, Microsoft 365 Personal rose from about $7 a month to $10 and Family from about $10 to $13, with 60 AI credits a month included. Existing subscribers could remain at the old price on a “Classic” plan.
And the customer base is enormous. Microsoft last reported 89 million consumer Microsoft 365 subscribers in July 2025 and then stopped publishing the metric, according to Office-Watch, which estimates the total is now about 95 million. It also reported that consumer cloud revenue rose 24 percent in the quarter ended June 30.
Even as incomplete products, those efforts should put a chill through the hearts of Adobe and Microsoft executives. Plus, those two companies are among the most abusive in all of tech and richly deserving of their comeuppance, as Louis Rossmann notes.
This is all great for people who hate Adobe’s and Microsoft’s user-hostile and anti-competitive practices, but the capabilities are fairly ominous for the software industry as a whole. It used to take a year or more for a dedicated programming team to churn out even a basic clone of a popular software product. Now one guy with sufficient AI tokens can do it in less than a week. (And don’t get me started on the whole “token economy” BS, which is an entirely different rant.) Given that, the desire to fund new software startups is going to be deeply disincentivized.
It’s ironic how the one thing AI has proven good at is replacing the jobs of the software engineers that helped create it.
Before the 1980s, hardware was king in tech. Chips and computers were expensive, and software was merely a way to sell more hardware. Then a variety of software companies (Lotus, Microsoft, Adobe, Boreland, Ashton-Tate, Intuit, etc.) managed to build business empires worth many multiples of their higher cost software brethren, a trend that only accelerated with the arrival of the World Wide Web. (Apple is a special case, in that their software still exists to drive hardware sales.) Now, thanks to the AI boom/bubble, it’s hardware and AI companies generating the huge valuations, with software profits under threat.
Just as in previous tech revolutions, I’m sure someone will figure out how to build business empires on the top of AI tools, but just like the Internet circa 1988, no one has figured out just how yet…