Configuring a generative AI secret

To activate the generative AI feature in Automation Decision Services, you must create a Kubernetes secret.

About this task

You do this configuration to connect to a watsonx.ai service that hosts the large language models (LLMs). Automation Decision Services calls this service when generation is triggered in a decision.

Procedure

Create a Kubernetes secret to activate the generative AI feature:
```
apiVersion: v1
kind: Secret
metadata:
  name: ads-genai-secret
type: Opaque
stringData:
  genAI: |- 
    { 
      "name": "GENAI",   
      "type": "WML",   
      "description": null,   
      "updateTime": 0,   
      "authInfo": { 
        "authType": "IAM",
        "apiKey": "<watsonx_ai_api_key>",
        "authUrl": "https://iam.cloud.ibm.com/identity/token"   },   
        "providerAdditionalInfo": {     
            "providerType": "WML",
            "mlUrl": "https://<location>.ml.cloud.ibm.com/",     
            "projectId": "<project_id>"
          },   
      "version": "V1" 
    }
```
where <location> in mlUrl is:
- us-south for Dallas
- eu-gb for London
- eu-de for Frankfurt
- jp-tok for Tokyo
- au-syd for Sydney
Also, where <watsonx_ai_api_key> and <project_id> are retrieved from your IBM watsonx account:
1. Create an IBM Cloud API key.
  
  i. Log in to IBM Cloud and select Manage > Access (IAM) > API keys.
  
  ii. Create an API key for your own personal identity.
  
  iii. Copy the key value, and save it in a secure place.
2. Create an IBM Cloud service. For more information, see Creating and managing IBM Cloud services.
3. Create a new project.
4. Associate your IBM Cloud service to the project. For your information, see Adding associated services to a project.
5. Go to Project > (project_name) > Manage and copy the project ID.
The secret must be referenced in spec.ads_configuration.genai_secret_name.