Getting Started with Dify and OceanBase seekdb

Preface

This post walks you through connecting Dify to OceanBase's AI‑native database seekdb. Given that SeekDB supports unstructured data and hybrid search, and is an AI-native database, I chose to use SeekDB to power my chatbot.

Build AI Applications with Dify

This section describes how to use Dify and OceanBase seekdb to quickly build a basic AI application. If you are already familiar with Dify, you may skip this section.
Get an Azure OpenAI API Key
To get an API key from Azure, visit https://azure.microsoft.com/en-us/products/ai-foundry/models/openai/.

[!NOTE]
Here I use Azure OpenAI only as an example.
Configure the Model Provider and System Model in Dify
In the upper right corner of Dify's console, choose Settings > Model Provider.

Add the Azure OpenAI Service Model by entering your Deployment Name, Model Type, API Endpoint URL, Authentication Method, API Key, API Version, and Base Model.
Click System Model Settings, and configure your embedding model for later use.

Create a Knowledge Base
On the Knowledge tab page, click Create Knowledge.

Choose your data source and upload files.

Configure document processing options. For the index method, select High Quality. Then, choose the embedding model you configured above.

Document embeddings will be generated here.
After the knowledge base is created, click Go to document to view the list of documents in this knowledge base.

You can then test the retrieval quality.

Create a Chatbot
On the Studio tab page, choose Chatbot > Create from Blank.

Name your application.

Within the application, you can add the knowledge base you just created.


You can then start debugging and previewing.