m{ai}geXR: AI + 3D for Desktop, iOS, iPadOS & Android

m{ai}geXR is an AI-assisted 3D and XR development project, spanning the Electron desktop app, iOS, iPadOS, and Android. I created it to bring conversation, editable code, and visual experimentation into one focused development workflow.

My role: Creator and software engineer — product direction, interface design, AI integration, and application development, using AI-assisted development throughout.

Status: Open-source beta and an active work in progress.

A promising 3D idea can disappear into setup: choose a graphics framework, configure an environment, prompt an AI model, move the output into an editor, and find out whether it runs. Every handoff adds friction to the creative process.

I wanted to connect those steps while keeping the code visible and editable. The goal was an approachable starting point for experimentation that also gives experienced developers room to inspect, adapt, and extend the result.

The workflow is deliberately direct: choose a framework and AI provider, describe a scene, generate code, inspect it, run it, and iterate. The Electron desktop app gives conversation, code, and scene previews room to work together. The iOS, iPadOS, and Android implementations bring the project to phones and tablets, extending the same prompt-to-creation concept across form factors.

The architecture pairs Electron desktop packaging with a React and TypeScript interface, Monaco editing, and a 3D preview workflow. Framework-specific prompts and templates give the model context about what it should build. Provider selection and generation settings let the creator shape how the assistant responds.

Keep the code in the loop. Generated output is a starting point that the creator can read, change, and test. Combining an editor with visual feedback supports both learning and deliberate iteration.

Make the assistant aware of its tools. Three.js, Babylon.js, and React Three Fiber have different APIs and conventions. Framework-specific context helps focus generation on the selected environment.

Make provider choice part of the product. Integrations with OpenAI, Anthropic, and Together.ai let users choose the model behind their workflow rather than rebuilding the experience around a single provider.

Design for specialization. The same concept can be adapted around a team's SDK, preferred libraries, coding conventions, or reusable templates. 3D/XR is the project's concrete starting point for a more focused AI coding experience.

The original demo below comes from the m{ai}geXR project site. It shows the iPad implementation with a generated scene and editable code, illustrating the project's prompt-to-scene workflow. This iPadOS footage demonstrates the creative workflow explored across desktop, iOS, iPadOS, and Android.

m{ai}geXR brings product thinking, AI integration, application engineering, and interactive graphics into one independently developed project. The open source makes the implementation available to inspect, and the demo makes the interaction visible.

The outcome is an evolving foundation for a specialized AI development tool: conversation connected to code and execution. I am continuing to refine the experience, and I plan to bring Nova64 into the broader project as future work.

View the desktop app source repository on GitHub — the repository is named WebMaigeXr and includes the Electron application code.

I help teams turn a development workflow into a usable product, connecting the interface, model integration, code editor, and domain-specific context. If you are building an AI application for desktop, iOS, iPadOS, or Android, a developer tool, or an interactive 3D experience, get in touch through my Contra profile.

I have documented the project's evolution from XRAiAssistant to m{ai}geXR, including the engineering decisions, framework integrations, and experiments along the way.

I'm building toward the next chapter of m{ai}geXR, with plans to bring Nova64 into the project. It is still a work in progress, and there is more to reveal. Follow my journey on Medium and here on Contra for what comes next.