Can n8n coordinate an AI agent workflow for SketchUp 3D modelling and Twinmotion construction animation?

[danda_war]

Describe the problem/error/question

Hi everyone,
I am looking for advice on whether n8n could be used to coordinate an AI-agent workflow for 3D modelling and construction animation.
The goal is to create a workflow where an AI agent can help generate a basic SketchUp model of a small SIP-panel building, such as a garden room. Ideally, each part of the model would be created as a separate component, for example:
Foundation
Floor structure
SIP wall panels
Roof panels
Windows
Doors
External cladding
Internal finishes
The reason I need the parts separated is so the model can then be taken into Twinmotion and animated stage by stage, showing the construction process from foundation to finished building.
For example, the animation would show:
Foundation installed
Floor structure added
SIP wall panels installed
Roof panels fitted
Windows and doors added
Cladding and final finishes completed
My main question is whether n8n can help control this type of workflow, for example:
Written prompt → AI agent → SketchUp model generation → separate components → export model → Twinmotion animation workflow
I understand n8n may not directly create 3D models itself, but could it coordinate the process using APIs, scripts, SketchUp Ruby automation, external AI tools, or file exports?
What is the error message, if any?
No error message. This is a workflow design question.
Please share your workflow
I do not have a working workflow yet. I am trying to understand the best architecture before building it.
The intended workflow would be something like:
Prompt input in n8n
↓
AI agent interprets building dimensions and SIP panel layout
↓
SketchUp automation creates the model
↓
Each building part is saved as a separate component/layer/tag
↓
Model is exported
↓
Twinmotion is used to create a stage-by-stage construction animation
Questions
Has anyone built anything similar using n8n?
Can n8n trigger or control SketchUp scripts, especially Ruby scripts?
Is it realistic for an AI agent to create a clean SketchUp model from a written prompt, or would it need a lot of manual correction?
Would it be better to generate the geometry through scripted rules instead of asking AI to directly model everything?
Can Twinmotion animation steps be automated, or would that part still need to be mostly manual?
Any advice on the best tools, workflow structure, plugins, APIs, or AI-agent approach would be appreciated.
Thank you.

What is the error message (if any)?

Please share your workflow

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

Share the output returned by the last node

Information on your n8n setup

• n8n version:
• Database (default: SQLite):
• n8n EXECUTIONS_PROCESS setting (default: own, main):
• Running n8n via (Docker, npm, n8n cloud, desktop app):
• Operating system:

[Fabi-SPL]

yeah n8n can run this, but its the conductor not the thing making the model. worth being clear on that up front cause it changes the whole design.

the mistake most people make here is asking the AI to model the geometry directly. llms are bad at clean parametric cad, youll spend more time fixing it than if youd just scripted it. the way id split it: let the AI do the part its actually good at, turning your written brief into a structured spec (wall lengths, panel count and sizes, opening positions, all as json). then a deterministic script builds the actual geometry from that spec. AI for intent, code for geometry. that separation is what gives you clean repeatable models AND your separated parts for free.

for sketchup thats the ruby api. n8n cant reach inside sketchup and run ruby itself, but it doesnt need to. n8n produces the json spec and hands it off, and a ruby script inside sketchup reads that spec and builds each part (foundation, floor, wall panels, roof, openings…) as its own component with a tag named per stage. because the script makes them as discrete named components, your stage-by-stage breakdown is baked in from the start, not something you split by hand later.

the twinmotion end is where id set expectations. import via datasmith and your component hierarchy + tags carry over, so all the parts land already organised. but the actual construction sequencing is twinmotions Phasing feature and thats mostly a manual timeline setup, theres no real api for n8n to drive the animation itself. so realistically n8n + ruby automate everything up to and including the export, and the phasing in twinmotion stays hands on (just way faster since the parts are already split and named).

so straight to your questions: yes n8n coordinates it, yes it can trigger ruby (indirectly, via a file or endpoint the script watches), no dont have the AI model directly, yes generate the geometry by scripted rules, and the twinmotion animation is the one bit that stays mostly manual.

happy to go deeper on the n8n to ruby handoff if you get to building it, thats the part with the most gotchas.