Conversational Agent - Data Store/Knowledge Access

[NeilA]

Hi all,

I've been experimenting with creating a Conversational Agent (very cool) and added a csv of FAQ's as a datastore, but they're not available when accessing the agent via Web Chat or the Phone number integration.

They do however work in the Simulator client, which makes me think it's a permissions issue.

I'm new to learning GCP and DialogFlow and can't find any knowledge articles on it, I'm assuming there's a permissions issue and the Chat/Phone call are running as a different user than the Admin in the test client, but I don't know where to start with fixing it.

Anybody got any advice?

...added three photos below:

1. Simulator client where it works

2. Deployed Chat button where it doesn't (same on a phone call)

3. Conversation history of 2, showing the same as 1 except now the data store doesn't return an answer

[MarvinLlamas]

Hi @NeilA,

Welcome to Google Cloud Community!

You are correct; the simulator client has different authorization than the deployed integrations. Furthermore, Dialogflow CX uses your service accounts to manage API calls for various integrations. During your simulations, it likely uses your personal Google Cloud Platform (GCP) account's credentials. For your Web Chat or Phone integration deployments, the agent operates under a service account linked to the Dialogflow CX project. Additionally, your data store, which is part of the Google Search and Conversational AI API (Discovery Engine), must grant explicit access to this service account.

Here are potential ways that might help with your use case:

• Identify the Dialogflow CX service account: Make sure to identify your service account in Dialogflow CX. Once you have identified it, copy its email address. This email address is the identity of the service that needs to be granted permission. For more information, you may refer to the following documentation.
• Grant Permissions to the Service Account: To grant access to your service account, you may refer to the following documentation.
• Correct Data Store ID: Make sure that the data store ID in your agent is correct.
• Data store indexing: Verify that your data is successfully indexed in your Discovery Engine data store.
• Logging: You may want to turn on Google Cloud Logging for your Dialogflow agent and the Discovery Engine if it still doesn't work. This will help you identify the source of permission or other configuration issues.

For more information about Dialogflow, please refer to the documentation.

In addition, you may try Google Cloud Skills Boost to help you explore and understand the use of GCP.

Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help.