Kinzoku Blueprint Forge: AI-Powered Prototyping for Engineers

Source: Dev.to

What I Built

I set out to build Kinzoku Blueprint Forge, a conceptual prototyping tool designed for engineers and industrial designers. Rather than creating another generic image generator, I aimed for a tool that bridges the gap between a rough idea and a technical specification.

The app takes a user’s text description of a mechanical concept (e.g., “A hydraulic spider leg for a Mars rover”) and uses a multi‑model approach to generate a prototype:

Gemini 3 Pro preview and Gemini 1.5 Flash act as the engineer, analyzing the request to generate a `Tech_Specs_V1.0` card containing hypothetical material compositions (e.g., Titanium Ti‑6Al‑4V), weight estimates, and dimensional envelopes.

Imagen 3 acts as the drafter, generating a high‑fidelity, isometric technical blueprint on a grid background.

Reverse Engineering Module

I also prototyped a Reverse Engineering Module (utilizing Gemini 1.5 Pro’s vision capabilities) that allows users to upload existing sketches for analysis, turning the tool from a simple generator into a comprehensive workflow assistant.

Key Prompt Strategy

To achieve the distinct “CAD software” look, I used a strict system instruction for the Imagen model:

"Always rewrite the user prompt to append: 'in the style of a high-contrast technical blueprint, white lines on dark blue grid paper, isometric view, hyper-detailed, 8k resolution.'"

Demo

Live application: Kinzoku Blueprint Forge

• Blueprint Forge Interface – Generating a hydraulic system with accompanying material analysis.
• Reverse Engineering Module – Analyzing uploaded schematics.

My Experience

My background is in software development and 3D modeling (Solid Edge), so my goal was to create something that felt “polished and different”—a tool I would actually want to use.

Key Takeaways

• Prompting for Precision – The biggest challenge was moving the AI away from “artistic” interpretations toward “technical” accuracy. Treating Gemini as a logic engine to structure the data before visualizing it creates a much more cohesive result.
• Speed to Polish – I struggle with perfectionism, which often delays projects. Google AI Studio allowed me to handle the heavy lifting of the backend logic instantly, letting me focus on UI/UX and the overall “feel” of the application.
• The Power of Context – Wrapping the AI models in a professional, industrial‑themed UI (Tailwind CSS) increased the perceived value of the output. Presentation proved to be just as important as the underlying model.

I built this to push my understanding of how AI can aid the mechanical design process, and I’m proud of the result.