A demonstration of Google’s Generative AI based Search Engine using VertexAI Agent Builder
The purpose of this demonstration is to showcase three different types of search engines created using Google’s Vertex AI Agent Builder and Google’s Cloud Storage service.
Search Engine 1: Public Website Data Source: This engine retrieves results from cloud.google.com, a public website.
Search Engine 2: Structured Data Source: This search engine accesses a dataset of movies stored in Google Cloud Storage.
Search Engine 3: Unstructured Data Source: This search engine searches through unstructured data in Google Cloud Storage, specifically earnings report PDFs from the Alphabet investor site Additionally, we also present the analytics of the search engines to evaluate performance and user engagement.
For additional information, please see our blog: Harnessing the Power of Generative AI for Next-Gen Search Experiences using Google Vertex AI Search
A Voice and Text Enabled Chatbot using Google’s Vertex AI Agent Builder and Dialog-flow Messenger
The purpose of this video is to showcase two different types of chatbots created using Google’s Vertex AI Agent Builder and Dialogflow Messenger.
Chatbot 1 is a Google Store Product Assistant that assists customers with inquiries about various products and devices available in the Google Store. Handles queries regarding phones, watches, laptops, and smart home devices.
Chatbot 2 is a Blood Donation Eligibility Checker that aids individuals in determining if they meet the requirements for donating blood.
Both chatbots support Voice and Text interaction modes and can engage in conversation either by calling a dialed phone number or through a chat widget embedded in a website.
We also present the conversation history and analytics of the chat bots to evaluate performance and user engagement.
For additional information, please see our blog: Enhancing Conversational AI with Dialogflow and Vertex AI: A Deep Dive into Vertex AI Agent Builder
Enterprise grade Multi-Modal Question Answering System
Typically, a Q&A system involves building a knowledge base, retrieving relevant knowledge from the knowledge base and using that knowledge to answer the user’s query. In this 5-minute video, we demonstrate the abilities of one such Q&A system to query PDFs built using OpenAI’s GPT-4 model, LangChain, Milvus Vectorstore (to store the text vectors) and Google Cloud’s DocumentAI (to parse the uploaded PDFs). As such, it is able to answer queries from several different types of PDFs with tables, images and very large amounts of text and is able to do so with a high level of accuracy. For additional information and access to source code, please see our blog: Building a production grade Q&A System for Knowledge Graphs and PDFs with LangChain Agents, Neo4j and Large Language Models