Vertex AI Chatbot Documents selection based on Customer

[shital1]

Hi

I have usecase where chatbot should respond with data for particular customer. So we have thousands of customers each with similar documents. Chatbot should take/filter input customer and give answers from documents of that particular customer ignoring other customers similar documents. How to develop this. Currently I am using vertex ai agent builder with chat option and datastore connecting to cloud storage bucket where I am storing documents. How to partition documents based on customer

[MJane]

Hi @shital1,

I understand that you want to develop a chatbot that can filter or provide answers by referencing the documents of individual customers.

Here are possible approaches you can try that might help you achieve using Google Cloud Services: 

• Document Organization and Storage - Since you're using a cloud storage bucket to store documents you might consider using a folder structure that partitions data by customer.
• Indexing and Search - You can use a search engine or indexing service such as Vertex AI Search to index the documents based on customer-specific metadata. Make sure to tag each document with customer IDs in the index to allow for efficient querying.
• Configure Vertex AI Agent - Adjust the chatbot’s logic to recognize and capture the customer context from user input. You can achieve this through initial questions, user authentication, or by using predefined customer identifiers. Then use the identified customer context to filter queries to the indexed documents or storage.filters and retrieves customer-specific data.
• Test with Sample Data - Test your chatbot with different queries to ensure it correctly filters and retrieves customer-specific data.

You can refer to this documentation for more detailed information regarding Vertex AI Agent Builder. 

I hope the above information is helpful.