Vertex AI AgentBuilder blank response via console and DialogFlow

[joaovitorAlvesR]

Hey,

I have some agents developed in Vertex AI AgentBuilder, new agents and some old ones, and I use their integration via DialogFlow.

However, for a few days now, I have been receiving the following error: AgentBuilder simply does not return a response, both in the console and in DialogFlow.

Does anyone know if there is any instability on the platform?

I thought it could be an error in the DialogFlow API, but even within GCP, in the console, where the agent is created, I'm not getting a response.

This example is the same error in two different agents.

1 - This example shows the AgentBuilder console, where as no response was received the agent shows the message "Sorry something went wrong, can you repeat?".

2 - This example shows the integration with dialogflow, where as the information was not received in the message, selected for the red box, it does not display any response, but it does display the matches found through the question I asked.

When asking again, the agent displays the answer, but as shown in the green box, a response message was retrieved and the matches were made.

[ibaui]

Hi @joaovitorAlvesR,

Welcome to Google Cloud Community!

Regarding your current issues, here are some strategies you can consider to potentially resolve it:

1. Check Google Cloud Status Dashboard: The first step is to check the Google Cloud Status Dashboard, which reports known outages and service disruptions. Pay particular attention to any issues related to Vertex AI, Dialogflow, or other related services in your region.

2. Examine AgentBuilder Logs: Check the logs associated with your AgentBuilder agent. These logs can offer detailed error messages that help identify the source of the issue.

3. Review Dialogflow Logs: Similarly, inspect the logs in Dialogflow for errors related to fulfilling requests from AgentBuilder.

4.  Simplify the Query: Try a very simple query to your agent (e.g., a single, common keyword) in both the AgentBuilder console and through Dialogflow. This helps isolate whether the problem is specific to complex queries or a general issue.

5. Check AgentBuilder Configuration: Carefully review the configuration of your AgentBuilder agent. Verify:

• Your data store connections are valid and working.
• Your data store schemas are compatible with your AgentBuilder agent.
• All necessary APIs are enabled for your project.
• Permissions are correctly set for your AgentBuilder service account. 

6. Review/Simplify Dialogflow Intent Matching: Examine the Dialogflow intents that trigger the AgentBuilder integration. Make sure the intents are correctly configured and that there are no conflicting intents that might be interfering. Consider simplifying the intent matching (e.g., using simpler training phrases) to see if that resolves the issue.

7. Monitor Resource Usage: Visit the Google Cloud Console and review your resource usage for Vertex AI and Dialogflow. Look for any indications of resource exhaustion, and check your quotas and request limits.

8. Isolate the Problem:

• Create a Minimal Example: Try creating a very simple AgentBuilder agent with minimal data and a simple query. See if that works. If it does, the problem is likely in the configuration or data of your more complex agents.
• Test Different Agents: You mentioned you have multiple agents. Does the problem occur with all of them, or only specific ones? This helps determine if it's a general platform issue or specific to certain agent configurations.

9. Consider Regionality: Verify that your Vertex AI Agent Builder agents and Dialogflow agents are in the same region. If not, there may be latency and/or compatibility issues.

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.

[joaovitorAlvesR]

The problem was not resolved, I did all the checks.

[tomasbarak]

I'm having the exact same issue. I don't know how to solve it yet, do you?