Cloud or Local AI: First Step to Connecting AI with Dynamo [Beginner’s Edition]

 

I have always been an active user of cloud-based AI like Google’s Gemini to enhance the efficiency of my Generative Design (GD) workflows and uncover optimal insights from countless design alternatives. Since the release of Gemini 2.5 Pro, its exceptional analytical capabilities have been a massive help in identifying patterns in complex data and suggesting design directions. However, it’s crucial to note that the quality of the AI’s response varies significantly depending on how well you craft your prompt.

But as I moved from experimentation to practical application — using the Gemini Pro model via an API key to iteratively analyze tens or even hundreds of design alternatives — I ran into a very real-world problem: the burn rate of API tokens. The more powerful the model, the more tokens were consumed for each analysis. For a GD workflow that requires free-form idea exploration and unlimited testing, this quickly became a significant cost burden.

This cost issue, combined with concerns over data security from sending sensitive corporate design data to external servers, led me to search for a new alternative. The result was a successfully configured local AI environment, where I built a Qwen3 model directly on my personal PC using a program called MSTY.AI. (You don’t have to use MSTY.AI, but in my early experience, I found it to be a very convenient way to connect various AI models.)

In this newsletter, I want to share the initial setup guides for both of these AI environments that I’ve personally navigated: Cloud AI (Gemini) and Local AI (MSTY.AI + Qwen3). Later, I will expand on this with use cases for geometry and data, as well as in upcoming AU sessions

(3406 | Generative design meets Digital twin: Transforming Semiconductor Facility Planning)

https://conferences.autodesk.com/flow/autodesk/au2025/sessioncatalog/page/inperson/session/1744876445688001oX1X

Press enter or click to view image in full size

(The session will be treated as a concept with security, but the use of AI in the Generative design process will be small.)

• Cloud AI (Gemini) offered the ‘power and convenience’ of experiencing high-level performance instantly without any complex setup. However, it came with the practical barriers of network latency, cost, and corporate security policies.
• Local AI (MSTY.AI) involved a more complex initial setup process. But once it was built, it provided tremendous value: ‘unlimited use’ with no API costs, ‘instantaneous response speed’ without network delays, and most importantly, ‘perfect data security.’

A significant drawback of the local approach is its poor accessibility to the Revit API. However, there are ways to reduce the frequency of errors by focusing on Python-based tasks. Since our method involves connecting to Dynamo rather than directly to Revit, I will explain this process step-by-step.

Based on the vivid know-how I gained from experiencing both of these approaches, I will share the beginning process of both approaches. And I will share the use cases step by step. I hope we can discuss various use cases together.

FIRST

Dynamo with the Google Gemini AI

(Beginner’s Edition)

1. Overview

This guide provides a step-by-step manual on how to communicate with Google’s cloud AI, Gemini, using Python scripts within Autodesk Dynamo. This workflow empowers you to ask questions in natural language and receive AI-generated responses directly inside the Dynamo environment. This is the first step toward a wide range of applications, such as searching for complex information, brainstorming ideas, or even generating simple code snippets.

2. Prerequisite: Get Your API Key from Google AI Studio

To communicate with Gemini, you need a unique “secret key” that authenticates your requests — an API key. You can get this key for free from Google AI Studio.

1. Access Google AI Studio: Open a web browser and navigate to the address below. Log in with your Google account. [https://aistudio.google.com/]
2. Create an API Key: After logging in, click the “Get API key” button, located either in the upper-left corner or the middle of the page. In the new window that appears, click the “Create API key in new project” button. After a moment, a long string of characters will be generated. This is your API key. Copy this key and save it in a secure location (e.g., a text file or password manager).

⚠️ Important Security Warning: Your API key is like a personal password. Never share it with others or post it in public spaces like blogs or code repositories (e.g., GitHub).

Next, for more information, please visit Newsletter