Works

Twelve looks on one board, and the six you pick built side by side in Photoshop.

Every camera in the model, rendered at the same time through the Gemini API.

Claude reads the Revit model and builds a project website you send as a link.

Claude pulls every wall type in the Revit model apart, layer by layer, and lays them out in Canva.

Claude furnishes and lights the room in Revit and saves D5 Lite images from the same camera.

A messy plan export from Rhino is cleaned up and coloured by room in Illustrator.

The Revit floor plan becomes a colour plan in Illustrator, then a rendered plan in Photoshop.

Room elements and pictures go from Revit into Notion, and Notion fills an InDesign booklet.

One prompt furnishes a room in the model, and Higgsfield renders the view it placed.

One prompt reads the open model and builds an editable presentation in InDesign.

Pick schedules with one prompt, and each one becomes a Notion database with a picture of every type.

One prompt sets the filters that make a plan a WIP drawing, the other cuts the part you name out in 3D.

One prompt counts what is in the model, asks how to group it, and lays every view out on a board.

One prompt reads your levels, asks how to split them, and stacks a view per storey on a sheet.

Export the model, answer three questions, and the video walks through your building along the path you chose.

Pick a folder, paste three prompts, and your projects become something an AI can read.

Colour the view by category, then change the materials without touching the model.

Your 3D views go out together and come back matching, from an add-in you build in two prompts.

Probabilistic and deterministic, and which half of practice belongs to each.

Revit shows one stair at a time. This shows every stair the room allows.

Toronto's zoning by-law, rebuilt as a graph a machine can answer from.

Every element is in a schedule or it is not. Nothing reports the ones that are not.