Launching today

Morable

Launching today

AI repairs room layouts. Geometry verifies what works.

1 follower

AI repairs room layouts. Geometry verifies what works.

1 follower

Visit website

Morable helps people make better decisions before buying furniture or changing a room. It combines AI-generated design proposals with deterministic spatial validation, so geometry and function—not the model—decide whether a layout works. The product was built primarily with GPT-5.6 Sol. For this challenge, we connected GPT-6 Astra for one bounded repair step: it proposed new x/y positions, and Morable independently revalidated the result from INVALID to VALID

Free

Launch tags:Design Tools•Artificial Intelligence•Interior design

Launch Team / Built With

Framer AI AgentsDesign and publish professional sites with AI

Promoted

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

What became practical with Astra was moving from detecting a spatial problem to attempting a constrained repair in the same loop. Morable already had deterministic geometry validation, so it could detect when a layout was physically wrong — for example, when furniture blocked the required clearance in front of a wardrobe. For this challenge, we let Astra make one bounded repair attempt, changing only the x/y positions of the affected furniture. Then the same independent Morable validator checked the layout again. In our test, the result went from INVALID → VALID without changing furniture dimensions or weakening the spatial rules. The part I find most important is the separation of roles: Astra proposes. Morable verifies. The model does not certify its own answer. That makes AI-assisted spatial correction much more practical for a product designed to help people make decisions before spending money on their homes.

Report

11h ago

Maker

📌

Hi Product Hunt! I’m Carlos, founder of Morable. I started Morable around a simple problem: AI can generate beautiful room ideas, but a beautiful layout can still be physically wrong. A desk can block a wardrobe, furniture can interfere with circulation, and an image can hide those problems. So Morable separates the roles: AI interprets and proposes, while deterministic spatial logic validates geometry and function. For this challenge, we tested a bounded repair loop: an invalid layout was detected by Morable, GPT-6 Astra was allowed one repair using only x/y positioning, and the same independent spatial validator checked the result again. The proposal went from INVALID to VALID without changing furniture dimensions or weakening the rules. Most of Morable has been built with GPT-5.6 Sol. Astra was used here for this specific repair experiment, which gave us a glimpse of what stronger reasoning models could unlock for spatial decision-making. I’d love feedback on the approach, especially the idea that AI should be able to propose — but not certify its own answer.

Report

14h ago