Breaking

Solo dev builds 3D cell-learning app with GPT and Gemini

July 8, 2026 at 17:55 EDT

Media Generation
Software Dev & Coding
Education

Software engineer Dilum Sanjaya built an interactive 3D learning app called "Cell Architecture Studio" that lets users rotate, disassemble and compare organelles, completing it over a weekend by combining GPT Images 2 and Gemini 3.1 Pro.

Weekend Build · AI-Assisted Development

A 3D biology learning app, built in a weekend — with no design tool

Software engineer Dilum Sanjaya assembled "Cell Architecture Studio" — a rotatable, disassemblable 3D explorer of cells and organelles — by chaining generative AI models, each given a single distinct job.

AI tools chained into one pipeline

77.1%

Gemini 3.1 Pro on the ARC-AGI-2 reasoning benchmark

Token context length for complex coding tasks

The "Image-to-Code" workflow

Step 1 · UI Design

GPT Images 2

Generates UI mockup images — an instant "design spec"

→

Step 2 · Code

Gemini 3.1 Pro

Reads the images, outputs working React + Three.js app code

→

Step 3 · 3D Assets

Tripo / Hunyuan3D

Generate 3D models of cells and organelles

The point: images become the interface between models — proof that "you can build this much without Figma."

Warm reception

Educators, learners and homeschooling parents want it in class
"Learn biology as if playing a game"
Open implementation already spawned an AI-tutor fork

The limits

Models favor "inspiration over perfect accuracy" — needs fact-checking
Some warn of misinformation risk
No full public release yet; UI took multiple tries, fine detail still manual

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.