Get hands-on experience with 20+ free Google Cloud products and $300 in free credit for new customers.

Vertex AI agent builder -- unable to get datastore working

So I'm having trouble trying to display anything from the datastore.

After I have built the agent, here are the instructions I've given it:

 

- Greet the users, then ask how you can help them today.
- Summarize the user's request and ask them to confirm that you understood correctly.
    - If necessary, seek clarifying details.
- Use ${TOOL: datastore} to look through the source code and file names. It is a datastore.

 

Then I tick the datastore to be used.

The datastore tool specifically is referencing the datastore I've built. The information i've imported into the datastore is a collection of filenames and their source code, here's an example:

 

{"id": "document393", "name": "BruteForceCounter.java", "content": "package com.thing.common.storefront.security;\npublic interface BruteForceAttackCounter\n{\nvoid registerLoginFailure(final String userUid);\nboolean isAttack(final String userUid);\nvoid resetUserCounter(final String userUid);\nint getUserFailedLogins(final String userUid);\n}", "metadata": {"file-name": "./thing/bin/ext-thing/thingcommon/thingcommonweb/commonweb/src/com/thing/common/storefront/security/BruteForceCounter.java"}}

 

The above seems to have been imported correctly into the datastore as it's mentioned it was successful.
However when I try and ask the agent "list entries in the datastore" or really anything, I get responses like "I cannot do that" etc... looking at what the agent did specifically, it looks like it's sending a query, but the return is always an empty 200?

Screenshot 2024-04-17 at 14.20.40.png

Anyone got any ideas? Why it's not working? Is the datastore just empty? If it's empty though, why won't it work with entries like "is it empty" either?

Screenshot 2024-04-17 at 14.22.25.png

Is there a way to just manually check if there is any values in the datastore?

3 REPLIES 3

Hello, I have ran into similar issues.  First thing I would check is:

a. Did the datastore get successfully indexed.  It says "Indexed" with the checkmark

b. Second this video (https://www.youtube.com/watch?v=cQfl3Nl3RXA) Shows how to use Gemini Pro to help create accurate 'goals' and 'instructions' for your LLM

c. I would test the datastore in Dialogflow CX or another app (search, chat, etc.) to see if it works there.

d. Lastly, I am in the same boat as you and hope Google provides more analytics for this new platform

Hello @deama ,

I agree with @Ian_Stack 's response that the data store must be successfully indexed. As quoted here in the documentation, be aware that it can take up to 4 hours for the agent to be ready for use while your domains, URLs, or documents are being indexed.

Also, you may want to check if you have hit any of the limitations below:

  • This feature currently only supports select languages in GA. See the data store column in the language reference.
  • Only the following regions are supported: global, us multi-region, and eu multi-region.
  • The only structured data store type supported is FAQ.
  • Apps with both chunked and non-chunked data stores are not supported.

Hope this helps.

i have the complete opposite problem. the output is empty and the bot is responding random stuff