Setting up K.ai
Setting up K.ai requires Administrator access
To get setup with K.ai you will need the following
API Key: provided by Kada or key for your own Azure Open AI model
URL: provided by Kada or the endpoint URL for your own Azure Open AI model
Context: You will need provide K.ai with context about your organisation to help it construct relevant descriptions for data and reporting assets. Your context will be used by K.ai to frame how it will respond to your requests to describe data assets.
We recommend the following pattern:
”You work at <enter details about your company>. <Company name> provides/supports …”
Configuring K.ai
Open K in your browser and log in
Go to Platform Settings and click on Integrations
Select Generative AI
Click New Integration
Add the URL
Add the API Key
Click Save
You will need to add the context and guard rails for K.ai to use.
Click the edit button to customise the properties to improve the quality of outputs from K.ai
Property | Description | Recommended actions |
|---|---|---|
Business context | Context about your business to help the LLM respond in relevant terms | Edit this property to include context about your business |
Guard rail for data and content objects | Instructions to guide the LLM when describing technical assets such as tables and reports | Adjust this property if the responses includes information that is not relevant |
Guard rail for collection objects (including terms) | Instructions to guide the LLM when describing business assets such as glossary terms | Adjust this property if the responses includes information that is not relevant |
Max response length (sentences) | How succinct the LLM response should be (in sentences). The lower the length, the more direct the LLM is. | Reduce or increase this property depending on the response preferred |
Persona for data objects | The type of persona the LLM should adopt when describing data objects such as tables and columns | Adjust this property to change how the responses are generated |
Persona for content objects | The type of persona the LLM should adopt when describing content objects such as reports and sheets | Adjust this property to change how the responses are generated |
Persona for collection objects | The type of persona the LLM should adopt when describing business objects such as terms | Adjust this property to change how the responses are generated |
Using K.ai
Now that the LLM connection details have been added, you will be able to generate descriptions for data and content assets like reports, sheets, and columns.
See https://kadaai.atlassian.net/wiki/spaces/KSL/pages/2073362433