Odin - AI assisted Mendix development
Content Type: Module
Categories: User Interface,Connectors,Artificial Intelligence
Overview
Odin is an extension for your Studio Pro which allows you to connect to a public/private LLM model (like GPT 4.1 mini) to support your app development.
Documentation
Typical usage scenario
- Getting started with your app
- AI assisted development
- Discuss with your LLM model and build apps
Limitations
- Only supports OpenAI models
- Not recommended for Production apps use
Features
- Use your AI model to intelligently build your apps starting with the app model.
- Domain model artifact CRUD
- Entities
- Attributes
- Associations
- Generalization
- Enumerations CRUD
Dependencies
- Open AI models (can be hosted on Azure with Azure AI Foundry)
- Model used MUST have Tool calling/Function calling capability
- Model used MUST have vision capability IF you wish to use image recognition
Configuration
You need the following to connect to the model -
- Access to an Open AI model (like GPT 4.1 mini) - Possible to do using Azure AI Foundry in public/private clouds
- Endpoint of the model
- Name of the deployment
- Authentication via one of the following methods -
- az login in terminal ( Entra ID)
- API Key
Frequently Asked Questions
- Which models are supported / preferred ?
- You can use any model which supports Tool calling and vision(only required if you intend to use image recognition). I prefer to use GPT 4.1 mini because it does not hallucinate much, has been reliable in my tests, is efficient, relatively cheap and supports both vision and tool calling.
- Can I use AWS or locally hosted models ?
- Not yet but it is on the roadmap
- Are there any features on the roadmap ?
- Using locally hosted models using Ollama and AWS hosted models like Claude are on roadmap as well.
- When can I build microflows with this ?
- Once the microflows extensibility APIs are available, I can start building the capability
- Is this free ?
- The extension is free and does not record any information. You will be charged for your AI model usage by the provider.
Key Information
The extension is currently built using the .Net extensibility API and supports limited features and hence is in Beta. It needs to be rebuilt with the Web Extensibility APIs to make it future proof and introduce Pages and Microflow buidling.
Feel free to reach out to me for support, contributions and feedback.
Releases
Version: 1.0.0
Framework Version: 10.12.0
Release Notes: first beta vesion - only works with OpenAI models