WAX: A Word Processor as a Platform

I've been working with Mathias Romeo on the development of WAX. It’s a project I haven't talked much about as I've been focusing a lot on Kotahi recently, but it's coming along pretty fast now.

The market for word processors is highly competitive, dominated by ubiquitous architecture systems like Google Suite and Microsoft 365. Competing with such offerings is challenging, but these dominant players have notable downsides:

• They are slow to innovate, offering a one-size-fits-all approach that often fails to meet specialized user needs.
• They are locked into their own AI models—Google Suite with Gemini, Microsoft with Copilot (OpenAI-based, but MS looks to want to diversify this to cut reliance on OpenAI)—limiting their ability to adopt the best models for specific tasks.
• These platforms raise data privacy concerns; if you don’t want Google or Microsoft to have access to all your documents, these tools are not ideal.

On the open-source side, word processing remains largely stuck in the 1990s, with desktop-based applications still dominating. Sophisticated, web-based, cloud-native solutions have been lacking. This is where WAX comes in.

The WAX Advantage

WAX is designed for users who need a flexible, AI-integrated word processor that allows for interaction with different AI models based on task specificity. For example, users could employ Stable Diffusion for image generation, OpenAI’s O3 for code generation, and Claude 3.5 for text processing. Unlike proprietary alternatives, WAX is fully open-source, self-hostable, and free from major tech platform data surveillance.

More importantly, WAX is advancing faster in the word processing space than its major competitors.

Built-In Design Mode

One of WAX’s standout features is its Design Mode, which enables seamless PDF export from predefined templates. Users can switch between Edit Mode and Design Mode to create polished outputs, whether for slides, proposals, or formatted documents. This mode allows users to interact with the AI to adjust document layout, resize pages, or even convert a document into slide format—simply by requesting it from the AI.

It incorporates snippets, small design styles that can be applied to document elements, enhancing customization and formatting flexibility.

It also recently added image generation with drag-and-drop functionality, allowing users to generate and insert images directly into their documents. Additionally, WAX automatically generates alt text and captions for these images, improving accessibility with minimal user effort.

The Future of WAX

We see a lot of exciting possibilities ahead. One key strength of WAX is its contextual sidebar, which adapts dynamically based on whether you're in Edit or Design Mode. This extensibility allows us to introduce new features, such as deep research tools. For example, users could enter a research assignment, let AI-powered tools gather insights in the background, and continue working in other modes, effectively parallelizing workflows.

We are also exploring slash commands for autocompleting information, which could assist researchers with automated citation management. Additionally, we envision the ability to store CSV files, allowing users to drag them onto the interface to generate interactive charts within the document.

Another promising direction is pre-production asset handling. Currently, most AI word processing tools operate directly on documents, but we are thinking about how metadata could be added as files are stored. For example, dragging multiple PDFs—such as open-access articles—into a folder could prompt WAX to interpret them, extract DOIs or citations, and automatically add them to a citation manager. This would allow users to effortlessly access references by typing /cite, streamlining their workflow.

We also see folder and nested file structures as an opportunity to map hierarchies, similar to book organization. This could enable exporting a folder as a book, EPUB, or print-ready PDF. Additionally, we are considering ways to generate accessible outputs, leveraging AI and UX advancements to further optimize workflows.

While WAX is already a powerful tool, development is far from complete. We are iterating quickly and will soon introduce a web-based version with an intuitive UX model. Features like RAG-based knowledge bases have been built but are yet to be fully integrated as we explore ways to enhance them before deployment. Beyond individual users, we are considering how WAX can serve researchers and businesses, providing a flexible, AI-powered alternative to mainstream word processors.

If you are interested in collaborating, testing, or providing feedback on WAX, drop me a line. We’re always eager to work with forward-thinking users who see the potential in an open-source, AI-enabled word processor built for modern workflows.