One developer used AI to recreate some of Adobe's most popular applications and released them as open source. Days later, Microsoft Office alternatives followed. The software industry should be paying attention.
For decades, the software business model was straightforward. Build something people need, make it difficult to replace, and charge them to keep using it. Then, in October 2026, a developer decided he was tired of paying Adobe.
Software engineer Brandon Thomas used Anthropic's Claude to help build free, open-source alternatives to seven Adobe applications, including Photoshop, Illustrator, Premiere Pro, Lightroom, Acrobat, After Effects, and InDesign. The applications, developed in Rust and released under the ArtCraft project, are designed to recreate familiar functionality without requiring Adobe subscriptions.
A few days later, Thomas expanded the effort to Microsoft Office, releasing early alternatives to Word, Excel, and PowerPoint.
Let's be clear. These aren't production-ready replacements. Independent testing found missing functionality, usability problems, and significant limitations. Recreating an application's interface is not the same as replicating decades of engineering, reliability, and professional workflows. But that's not the most interesting part of this story.
The interesting part is that software once requiring substantial engineering resources can now be recreated, at least in early form, by a single experienced developer working with AI.
And that raises a much bigger question: What happens to the software business when building software becomes dramatically easier?
The Economics of Software Are Changing
Traditional software companies have benefited from a powerful economic advantage: building sophisticated applications is difficult, expensive, and time-consuming. That creates barriers to entry. It also helps justify recurring subscriptions, proprietary ecosystems, and long-term customer commitments. AI-assisted development is beginning to challenge those economics.
In a randomized controlled study involving 96 Google software engineers, researchers found that AI assistance reduced time spent on a complex development task by an estimated 21%, although the confidence interval was wide. However, another randomized study by METR found that experienced open-source developers working in familiar, mature codebases actually took 19% longer when using early-2025 AI tools.
The reality is that AI doesn't automatically make every software project faster or cheaper. The results depend on the task, developer, tools, and complexity of the existing system.
But the direction of change is becoming harder to ignore. AI is expanding what individuals and smaller teams can attempt. ArtCraft demonstrates that even complex, established software categories are no longer beyond the reach of relatively small development efforts.
Not every application will be replaced, but the cost and difficulty of competing is changing.
When Software Becomes Easier to Build, What are We Actually Paying For?
Consider the typical enterprise technology stack. Companies pay for software to manage customer relationships, analyze data, produce reports, create content, automate processes, and coordinate teams. Each application comes with its own interface, licensing model, integrations, permissions, and limitations. Over time, organizations accumulate more software than they can efficiently manage. But much of what businesses actually need isn't another application. They need a task completed. A report generated. A campaign analyzed. A customer issue resolved. A process automated.
Historically, accomplishing those tasks meant buying software designed around a predefined set of features. AI introduces the possibility of building or assembling the capabilities required for a specific task, then adapting them as business needs change. Instead of purchasing another application for every new problem, organizations may increasingly create workflows, agents, and interfaces around the outcomes they want.
The value shifts from owning access to a particular application to controlling what the technology can actually do.
That is a fundamental challenge to traditional software economics.
The Subscription Model Is Under Pressure
ArtCraft is particularly interesting because its creator wasn't trying to invent an entirely new category of software. He was responding to an existing one.
According to PetaPixel, Thomas said his frustration with Adobe's cancellation fees helped motivate the project. He wanted to create alternatives that users could access without the same subscription restrictions.
The project isn't completely free of commercial services. ArtCraft offers optional paid AI credits, while its core applications are open source. But the distinction matters. A subscription is easier to justify when the underlying product is difficult to replicate, deeply differentiated, or essential to a customer's operations. It becomes harder to defend when customers have credible, lower-cost alternatives.
That doesn't mean Adobe, Microsoft, or other established software companies are about to disappear. Their products offer extensive functionality, mature ecosystems, support, security, and reliability that early-stage alternatives haven't demonstrated.
But AI-assisted development could create new competitive pressure, particularly for applications with narrow functionality, high subscription costs, or limited differentiation.
And as the barriers to building software decline, customers may start asking a question that once seemed unreasonable: Why are we paying so much for something we could build ourselves?
The Next Generation of Software May Not Look Like Software
The bigger disruption may not be cheaper versions of existing applications. It may be an entirely different way of interacting with technology. Most enterprise software today is organized around applications. Users navigate menus, move between systems, export data, and manually coordinate work. AI agents and workflows introduce a different model.
Instead of opening several applications to complete a task, a user can increasingly describe the desired outcome and have software coordinate the necessary steps. Consider a marketing team preparing its monthly performance report.
Today, that might involve exporting campaign data, combining spreadsheets, reviewing performance, creating visualizations, and assembling a presentation. In an AI-driven workflow, those steps could be coordinated through a single interface, with automated data retrieval, analysis, report generation, and human approval.
The underlying systems may still exist. In fact, many will remain essential. But users may interact with them less directly. The application becomes less important than the capability it provides. This doesn't eliminate software. It changes where software creates value. The companies that succeed may not be those with the most features or the largest collections of applications. They may be the ones that make it easiest to connect capabilities, automate work, and adapt to changing requirements.
Enterprises Shouldn't Replace One Kind of Lock-In With Another
There is a catch. If AI makes software easier to build, it also creates the potential for more fragmentation. More agents. More workflows. More models. More tools. More infrastructure to manage. And replacing traditional software dependencies with dependencies on a single AI provider isn't necessarily progress.
Enterprises still need security, governance, reliability, interoperability, and control over their data. They need to know what their AI systems are doing, what information they can access, what they cost to operate, and how to change them without rebuilding everything. That's why the next phase of enterprise AI won't simply be about generating applications faster. It will be about making the entire AI operation work together.
At Rival, we believe that the future must be agnostic. Businesses should be able to build, run, connect, and govern AI across models, tools, applications, and infrastructure without being locked into a single provider. Because as technology becomes easier to create and replace, flexibility becomes more valuable, not less.
Software Isn't Going Away. Its Competitive Advantage Is Changing.
The ArtCraft applications may never seriously threaten Adobe's business. They may remain experimental, or they may develop into credible alternatives over time. That's still an open question. But the broader shift deserves attention. AI is changing who can build software, how quickly new products can emerge, and what customers may eventually expect to pay.
For established software companies, that means the barriers protecting their businesses could become less durable. For enterprises, it creates an opportunity to rethink how they acquire, build, and operate technology. And for the next generation of AI platforms, it points toward a different future. One where the most valuable technology isn't necessarily the application you buy.
It's the freedom to build, connect, and run whatever your business needs.
Software isn't dead.
But the era of paying a premium simply because software is difficult to build may be coming to an end.
Sources
PetaPixel, October 7, 2026: Someone Rebuilt Free, Open-Source Versions of Photoshop, Premiere, and Lightroom Using AI
The Register, October 8, 2026: Rust Upstart Reveals Reimplementation of Microsoft Office With Claude on Coding Duty
TechRadar, October 10, 2026: One Dev Is Recreating 7 Adobe Apps in Rust for Free — So I Tested It Out
ArtCraft: Original open-source project repositories on GitHub
IEEE / ICSE-SEIP, 2025: How Much Does AI Impact Development Speed? An Enterprise-Based Randomized Controlled Trial
METR, 2025: Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity