One developer used Claude to rebuild Microsoft Office in Rust
A solo engineer used Claude Opus 5.5 to build free open-source clones of Word, Excel, and PowerPoint. The team-size moat is collapsing.
Tools · Source: Thurrott
What happened
Brandon Thomas is a solo developer with 15 years of software engineering experience. He just released open-source clones of Microsoft Word, Excel, and PowerPoint. He built WordCraft, GridCraft, and DeckCraft from scratch in pure Rust. He used Anthropic Claude Opus 5.5 to help write the code. The apps read and write native DOCX, XLSX, and PPTX files. They do not require a Microsoft 365 subscription or a license key. You do not even need an account.
The project is moving incredibly fast, but it still has functional gaps. GridCraft covers 93 percent of Excel worksheet functions and 89 percent of its ribbon commands. However, the developer notes it is only about 50 percent ready for real work. WordCraft sits at 62 percent overall feature parity. The surface user interface was easy for AI to replicate quickly. The deep formatting logic is taking much more time to catch up.
This launch follows Thomas releasing seven Adobe Creative Cloud clones and an AutoCAD alternative called CADCraft just days prior. He started this massive project after getting hit with Adobe cancellation fees one too many times. Developers are paying close attention. WordCraft quickly pulled in over 5,100 stars on GitHub. GridCraft and DeckCraft grabbed 2,600 and 1,800 stars respectively. He plans to fund these free tools by selling token access to his own visual AI model and IDE.
Key facts
- 5,100 — GitHub stars WordCraft received by October 11
- 93 percent — Excel worksheet functions covered by GridCraft
- 15 — Years of software engineering experience Brandon Thomas has
- Claude Opus 5.5 — The Anthropic AI model used to help write the code
Why it matters
The barrier to entry for enterprise software just hit the floor. Microsoft and Autodesk built their empires on the sheer cost and time required to write compatible alternatives from scratch. Their moats were never about unique features. They were about the massive headcount needed to build and maintain them. Now, a single experienced engineer with an AI coding assistant can credibly challenge decade-old software suites in a matter of weeks. The team-size advantage is disappearing rapidly.
This also redefines the debate around AI-generated code. Critics called his earlier work AI-generated slop and vibe coding. But thousands of developers starring the GitHub repositories suggest a different reality. This is not a non-technical founder building a simple web wrapper. This is a veteran engineer using AI as a massive leverage tool. He is reverse-engineering proprietary formats and writing complex desktop software across Windows, Mac, and Linux. The definition of what one person can build has permanently changed.
For builders
Target bloated subscription software
Users are tired of bundled pricing models and cancellation fees. Solo founders can now use AI to build clean-room alternatives to expensive legacy tools. The incumbents lose their monopoly on complex feature sets.
Sell access to fund free tools
Thomas plans to fund these free open-source apps by selling token-based access to his visual AI model and IDE. Builders can use open-source utility tools as a massive top-of-funnel acquisition strategy for paid AI products.
The final ten percent is hard
AI gets you the UI and basic functions fast, but deep logic takes time. Thomas walked back his claim of reaching full parity in a month, admitting the final stretch takes months. Do not underestimate the complexity of legacy edge cases.
My take
I love this. A single developer got mad at an Adobe cancellation fee and decided to burn down the entire subscription software ecosystem using Claude. The big tech companies thought their massive legacy codebases were an impenetrable moat. They were wrong. AI just gave solo builders the firepower to hunt elephants.
Original reporting: Thurrott. This is my rewrite and opinion.