Conversational Search agent - Search Employee data

[balajit]

I work for a Enterprise. I'm in charge of developing applications related to HRIS applications within the company.

Recently i did a POC in Dialogflow CX to automate some process.

Now i'm keen in using AI or related tools in GCP to acheive the following

1. A chatbot or similar where anyone within the company will be able to search for employee information

2. Eg., Like i might expect someone to ask who is the Supervisor of X, some might ask where is person X located. 

3. What i want is that instead of typing in Employee firstname or lastname or employee number, i wish to provide the possibility for someone to search based on natural language search

i came across something called as datastore in Vertex agent, but i find no real example or walkthrough of fetching from a database. 

I would like the query in natural language to hit a database(any form) and show info pertaining to it.

I would like to build more on security and authorization, but want to focus on getting the base program before getting into that.  I certainly need to keep in mind the sensitivity of PII and other stuff, which i will want to work further based on the solution i get

Please suggest solutions based on Vertex or any others that i'm not aware of

[ruthseki]

Hi @balajit,

Welcome to Google Cloud Community!

I understand that you want to know how to retrieve data from the database in a Vertex AI Agent. Additionally, you need recommendations on building natural language-based employee information chatbot, while taking data sensitivity and access controls into account.

For information on fetching data from the database in a Vertex AI Agent, you may consult this documentation.

You may create your chatbot using Dialogflow CX, a natural language processing engine. Here’s the summary of Dialogflow CX key components that you may utilize for building your chatbot:

Intents: Define the types of questions your chatbot will understand.

Entities: Represent the data types you'll extract from the user's query (e.g., "employee name", "location").

Dialogflow Fulfillment: Write code to connect your chatbot to your database.

You may want to check this example for guidance on working with Dialogflow CX.

With regard to Implementing security measures in your application, you can utilize Identity and Access Management (IAM). Control access to your database and Vertex AI Agent using Google Cloud's robust IAM system. If you expose employee information in the chatbot responses, consider data masking techniques to protect sensitive PII such as hiding parts of the employee's email address, showing only the first few digits of a phone number. Lastly, use secure protocols like HTTPS for all communication with your database and chatbot.

Please feel free to check some of these documentations that may help you build your project.

• Dialogflow CX Quickstarts
• Dialogflow CX Tutorials and Samples
• How to keep sensitive data out of your chatbots

I hope the above information is helpful.