> ## Documentation Index
> Fetch the complete documentation index at: https://docs.runchat.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Invent worlds, then deduce designs

> Imagine the world an artefact came from, photograph it, then deduce the architecture that would belong there

Start from a context instead of a building. An agent invents the place a random artefact came from, an image model photographs it, and a second step works out what its people would build.

![A museum artefact, the invented landscape it came from, and the building deduced for that world](https://mintcdn.com/runchat/ExzxwftWKPCOoCF3/images/examples/inventing-worlds.webp?fit=max\&auto=format\&n=ExzxwftWKPCOoCF3\&q=85\&s=2d75d5c9fdb42808ed1eb12d577163e6)

> [card] [Open the example workflow](https://runchat.com/editor?id=97ad6c47-8fef-4d37-b192-c79b24fa8e57)
>
> Make a copy of this workflow in your account to try it with your own images.

## How it works

1. **Pick a random Smithsonian object** fetches an artefact.
2. **Invent the world this object came from** is an Agent (vision) that writes a landscape photo prompt: terrain, climate, light and traces of culture, but no buildings.
3. **Photograph the invented world (Nano Banana Pro)** renders that landscape.
4. **Deduce a building for this world, as a card model (Nano Banana Pro)** takes the landscape and the artefact and designs a building that responds to both.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.