Grounding Your Large Language Model in Data – Mendix Cloud GenAI

Last modified: April 4, 2025

Introduction

This document explains how to add data to your smart app to integrate with a Large Language Model (LLM). To do this, you can use your existing app or follow the Build a Smart App from a Blank GenAI App guide to start from scratch.

In this document, you will:

  • Learn how to ground your LLM in data within your Mendix application using the Mendix Cloud GenAI Resource Packs.
  • Discover how to integrate GenAI capabilities with a knowledge base to effectively address specific business requirements.

Prerequisites

Before implementing this capability into your app, make sure you meet the following requirements:

Grounding Your LLM in a Data Use Case

Choosing the Infrastructure

Since this document focuses on the Mendix Cloud GenAI Resource Packs, ensure that you have the Mendix Cloud GenAI Connector installed.

Follow the Navigate through the Mendix Cloud GenAI Portal instructions to collect the resources keys.

Creating Domain Model Entity

Since your application needs to store information, you must create attributes for the knowledge you want to save. In this example, based on the demo data mentioned below, a Description attribute of type String is created.

Demo Data

You can upload your custom data into the knowledge base. However, for this example, a modified and streamlined version of the demo data from the GenAI Showcase App is used. This demo data includes a Description attribute that provides information on resolving basic IT support issues. The following details are provided:

  • A JSON file containing examples of IT support solutions, such as “If the software crashes every time you try to save your document, first ensure you have the latest updates installed. Try…”
  • An Import Mapping that maps the JsonObject into the corresponding domain model entity.

Loading Data Into the Knowledge Base

To start, create a microflow that allows you to upload data into your knowledge base.

Loading Microflow

  1. Create a new microflow, for example, ACT_TicketList_LoadAllIntoKnowledgeBase.

  2. Add the Retrieve Objects action. You can configure it as follows:

    • Source: From database
    • Entity: Select the entity that contains your knowledge, which in this example would be the MyFirstModule.Ticket
    • Range: All
    • Object name: TicketList

  3. Next, add the Chunks: Initialize ChunkCollection action. You can keep the Use return variable as Yes and object name ChunkCollection.

  4. As shown in the image above, include a loop where the iterator has Loop type For each (item in the list), the Iterate over is the List retrieved in the above step, which in this case is named TicketList (List of MyFirstModule.Tickets). Lastly, you can add a Loop object name as IteratorTicket. After saving these settings, add a Chunks: Add KnowledgeBaseChunk to ChunkCollection action inside the loop. Here you can configure it as follows:

    • Chunk collection: $ChunkCollection
    • Input text: edit the expression to use the iterator object from the loop with the desired attribute, which in this case is $IteratorTicket/Description
    • Human readable ID: empty (optional)
    • Mx object: Select the loop’s iterator, such as $IteratorTicket
    • Use return value: No
    • Metadata collection: empty (optional)

  5. After the loop, add a Retrieve action to retrieve a MxCloudKnowledgeBase. In this example, the first entry found in the database is used.

    • Source: From database
    • Entity: MxGenAIConnector.MxCloudKnowledgeBase
    • Range: First
    • Object name: MxCloudKnowledgeBase

  6. Next, add the Connection: Get action from the Mendix Cloud Knowledge Base category:

    To edit the parameter value for MxCloudKnowledgeBase, double-click its type, select Variable, and assign it the value MxCloudKnowledgeBase. Similarly, for CollectionName, double-click its type, select Expression, and assign it the value TicketSolutions.

    You can keep the Use return variable as Yes and the object name MxKnowledgeBaseConnection.

  7. Add the Embed & Repopulate Collection action to insert your knowledge into the knowledge base:

    To edit the parameter value for Connection, double-click its type, select Variable, and assign it the value MxKnowledgeBaseConnection. Similarly, for ChunkCollection, double-click its type, select Variable, and assign it the value GenAICommons.ChunkCollection.

    You can keep the Use return variable as Yes and the variable name IsSuccess.

  8. Next (optional), include a decision:

    • Caption: for example, Replace Success
    • Decision Type: Expression
    • Expression: $IsSuccess

    If the decision is true, an End event action can be added where a microflow return value to true. You may add a message to inform the end user that the insertion was successful.

    If the decision is false, an End event action can be added where a microflow return value to false. You may add a message to inform the end user that the insertion failed.

You have successfully implemented the knowledge base insertion microflow! If you do not have any data available in your app yet, you need to create a microflow to generate the dataset, as described in the Data Set Microflow section below.

Data Set Microflow

This microflow first checks whether a list of tickets already exists in the database. If not, it imports a JSON string as described in the demo data section above.

  1. Create a new microflow, for example, Tickets_CreateDataset.

  2. Add a Retrieve action:

    • Source: From database
    • Entity: Select the entity that contains your knowledge, which in this example is MyFirstModule.Ticket
    • Range: First
    • Object name: Ticket

  3. Include a decision where:

    • Caption: Tickets?
    • Decision Type: Expression
    • Expression: $Ticket = empty

    If the decision is false, an End event is added, as importing tickets is not required.

    If the decision is true, continue to the next step.

  4. In the true path, add the Create Variable action, where the String value includes the JSON text mentioned in the demo data. Use TicketJSON as the variable name.

  5. Next, add the Import With Mapping action with the following configurations:

    • Variable**: TicketJSON created in the previous step
    • Mapping: Use the mapping mentioned in the demo data section
    • Range: All
    • Commit: Yes without events
    • Store in variable: No (optional, not needed here)
    • Variable name: (optional) only when stored in a variable

With both microflows created, they must be combined and added to the homepage to populate the knowledge base.

Joining the Microflows

  1. Create a new microflow ACT_TicketList_CreateData_InsertIntoKnowledgeBase.

  2. Add a Call Microflow action where you call the MyFirstModule.Tickets_CreateDataset microflow created above.

  3. Next, add a Call Microflow action where you call the MyFirstModule.ACT_TicketList_LoadAllIntoKnowledgeBase microflow created above. For the Use return variable, select No.

You have successfully added the logic to insert data into the knowledge base!

Chat Setup

To use the knowledge in a chat interface, create and adjust certain microflows as shown below.

  1. Search for the pre-built microflow ChatContext_ChatWithHistory_ActionMicroflow in the ConversationalUI > USE_ME > Conversational UI > Action microflow examples folder and copy it into your MyFirstBot module.

  2. Search for the pre-built microflow ACT_FullScreenChat_Open in the ConversationalUI > USE_ME > ConversationalUI > Pages folder. Copy the microflow into your MyFirstBot module. Right-click on the copied microflow and select Include in project.

  3. In the ACT_FullScreenChat_Open microflow, change the parameters of the New Chat action:

    • Set the Action microflow input parameter as your new MyFirstBot.ChatContext_ChatWithHistory_ActionMicroflow from your MyFirstBot module.

    • Set the System prompt input parameter as a prompt that fits your use case. For example, You are a helpful assistant supporting the IT department with employee requests. Use the knowledge base and previous support tickets as a database to find a solution to the user’s request without disclosing sensitive details or data from previous tickets.

    • The Provider name input parameter can be modified to a more purpose-specific text, such as My GenAI Provider Configuration.

    With the MyFirstBot.ACT_FullScreenChat_Open microflow configured, the MyFirstBot.ChatContext_ChatWithHistory_ActionMicroflow can now be adjusted to handle user-submitted messages in the chat interface.

  4. Open your MyFirstBot.ChatContext_ChatWithHistory_ActionMicroflow microflow in your MyFirstBot module.

  5. After the Request found decision, add a Retrieve action. In this example, the first entry found in the database is used, just as in the insertion microflow.

    • Source: From database
    • Entity: MxGenAIConnector.MxCloudKnowledgeBase
    • Range: First
    • Object name: MxCloudKnowledgeBase

  6. Add the Tools: Add Mendix Cloud Knowledge Base action with the settings shown in the image below:

The rest of the actions can remain as they are currently set. Now that everything is implemented, you can test the chat with enriched knowledge.

For the application to function as expected, ensure that the following microflows can be called from the navigation menu or homepage:

  • Chatbot: Add the MyFirstModule.ACT_FullScreenChat_Open microflow which was created in the Chat Setup section.

  • Create Demo Data and Populate KB: Add the MyFirstModule.ACT_TicketList_CreateData_InsertIntoKnowledgeBase which was created in the Joining the Microflows section.

  • Mendix Cloud Configuration: If you started from a Blank GenAI App, the configuration page should already be included. In case you started from your application, add the Configuration_Overview page.

  • Ensure that your admin role has the following module roles assigned: MxGenAIConnector.Administrator, ConversationalUI.User, and MyFirstModule.Administrator.

Testing and Troubleshooting

Before testing, ensure that you have completed the Mendix Cloud GenAI configuration as described in the Build a Chatbot from Scratch Using the Blank GenAI App, particularly the Mendix Cloud GenAI Configuration section.

To test the Chatbot, click on the Create Demo Data and Populate KB option to populate the knowledge base and go to the Chatbot icon to open the chatbot interface. Start interacting with your chatbot by typing in the chat box something related to your knowledge base. For example, My computer crashes every time, what can I do?

Congratulations! You grounded your LLM in data and your chatbot is now ready to use.

If an error occurs, check the Console in Studio Pro for detailed information to assist in resolving the issue.