Enrich Your Mendix App with GenAI Capabilities
Last modified: November 19, 2024
Introduction
With the Mendix GenAI capabilities you can create engaging, intelligent experiences with a variety of AI models and your own data.
These pages cover modules which integrate with third-party generative AI tools. For running pre-trained Machine Learning (ML) models using the Mendix Runtime, please see Machine Learning Kit.
Typical Use Cases
Mendix supports a variety of generative AI tasks by integrating with tools such as Amazon Bedrock or Azure OpenAI. Typical use cases include the following:
- Create conversational UIs for AI powered chatbots and integrate those UIs into your Mendix applications.
- Connect any model through our GenAI connectors, or by integrating your own connector into our GenAI commons interface.
- Connect your own data to ground GenAI systems with data from inside your application and the rest of your IT landscape.
Getting Started
To start using the GenAI capabilities of Mendix, complete the following tasks:
- Familiarize yourself with concepts such as prompt engineering, Retrieval Augmented Generation (RAG) and function calling (ReAct).
- Select the right architecture to support your use case. For a full list of possibilities, see Architecture and Components.
- Obtain the required credentials for your selected architecture.
Architecture and Components
To help you get started, the following sections list the available GenAI components and models.
Mendix Components
| Asset | Description | Type | Studio Pro Version |
|---|---|---|---|
| AI Bot Starter App | Lets you kick-start the development of enterprise-grade AI chatbot experiences. For example, you can use it to create your own private enterprise-ready ChatGPT-like app. | Starter App | 10.12 |
| Blank GenAI App | Start from scratch to create a new application with GenAI capabilities and without any dependencies. | Starter App | 10.12 |
| Support Assistant Starter App | Learn how to combine common GenAI patterns, such as function calling and RAG to build your own support assistant. Connect it to a model like Anthropic Claude or Amazon Titan via Amazon Bedrock or use an (Azure) OpenAI subscription. | Starter App | 10.12 |
| GenAI Showcase App | Understand what you can build with generative AI. Understand how to implement the OpenAI and Amazon Bedrock connectors and how to integrate with the Conversational UI module. | Showcase App | 9.24.2 |
| Conversational UI | Create a Conversational UI for a new or existing app. | UI Module | 9.24.2 |
| OpenAI Connector | Connect to Azure OpenAI. | Connector Module | 9.24.2 |
| Amazon Bedrock Connector | Connect to Amazon Bedrock. Use Retrieve & Generate or Bedrock agents. | Connector Module | 9.24.2 |
| PgVector Knowledge Base | Manage and interact with a PostgreSQL pgvector Knowledge Base. | Connector Module | 9.24.2 |
| GenAI Commons | Common capabilities that allow all GenAI connectors to be integrated with the other modules. You can also implement your own connector based on this. | Common Module | 9.24.2 |
Available Models
Mendix connectors offer direct support for the following models:
| Architecture | Models | Category | Input | Output | Additional capabilities |
|---|---|---|---|---|---|
| Azure / OpenAI | gpt-3.5 | Chat completions | text | text | Function calling |
| gpt-4, gpt-4o, gpt-4o mini | Chat completions | text, image | text | Function calling | |
| DALL·E 2, DALL·E 3 | Image generation | text | image | ||
| text-embedding-ada-002, text-embedding-3-small, text-embedding-3-large | Embeddings | text | embeddings | ||
| Amazon Bedrock | Amazon Titan Text G1 - Express, Amazon Titan Text G1 - Lite, Amazon Titan Text G1 - Premier | Chat Completions | text | text | Document Chat (except Titan Premier) |
| AI21 Jamba-Instruct | Chat Completions | text | text | ||
| AI21 Labs Jurassic-2 (Text) | Chat Completions | text | text | ||
| Amazon Titan Image Generator G1 | Image generation | text | image | ||
| Amazon Titan Embeddings Text v2 | Embeddings | text | embeddings | ||
| Anthropic Claude v2.0, Anthropic Claude v2.1 | Chat Completions | text | text | Document Chat | |
| Anthropic Claude v3 Sonnet, Anthropic Claude v3 Haiku, Anthropic Claude v3 Opus | Chat Completions | text, image | text | Function calling, Document Chat | |
| Anthropic Claude v3.5 Sonnet | Chat Completions | text, image | text | Function calling | |
| Cohere Command | Chat Completions | text | text | Document Chat | |
| Cohere Command Light | Chat Completions | text | text | ||
| Cohere Command R, Cohere Command R+ | Chat Completions | text | text | Function calling, Document Chat | |
| Cohere Embed English, Cohere Embed Multilingual | Embeddings | text | embeddings | ||
| Meta Llama 2, MetaLlama 3 | Chat Completions | text | text | Document Chat | |
| Meta Llama 3.1 | Chat Completions | text | text | Function calling, Document Chat | |
| Mistral AI Instruct | Chat Completions | text | text | Document Chat | |
| Mistral Large, Mistral Large 2 | Chat Completions | text | text | Function calling, Document Chat | |
| Mistral Small | Chat Completions | text | text | Function calling |
For more details on limitations and supported model capabilities for the Bedrock Converse API used in the ChatCompletions operations, see Supported models and model features in the AWS documentation.
The available showcase applications offer implementation inspiration for many of the listed models.
Connecting to Other Models
In addition to the models listed above, you can also connect to other models by implementing one of the following options:
- To connect to other foundation models and implement them in your app, use the Amazon Bedrock connector.
- To connect to Snowflake Cortex LLM functions, configure the Snowflake REST SQL connector for Snowflake Cortex Analyst.
- To implement your own connector compatible with the other components, use the GenAI Commons interface.