Business Problem

• VantagePoint Seed Venture Fund specializes in investing in businesses and supporting founders who build and enable Intelligence-as-a-Service. With a vision to drive AGI adoption across various functions, VantagePoint Fund seeks to automate investment and startup intelligence for early-stage founders and portfolio investors. The goal is to leverage their extensive knowledge base and publicly available data to create a tool that super-serves questions related to scaling AI startups. This tool, named Addy, should be able to utilize existing investment videos, podcasts, and other resources to provide valuable insights through a simple yet interactive interface.

About VantagePoint

VantagePoint Venture Fund is a forward-thinking investor specializing in Intelligence-as-a-Service (IaaS) businesses and the visionary founders who build them. We believe in the transformative power of AI and are committed to driving its adoption across various industries.

Solution

To address VantagePoint Fund's requirements, GoML developed a comprehensive MVP of a Generative AI-powered solution leveraging the AWS technology stack and Lyzr SDKs. The solution encompasses the following key features and objectives:

Automated Query Resolution:

Designed to super-serve questions related to scaling AI startups, the solution uses advanced Generative AI models, specifically GPT-4, to provide detailed, contextually relevant answers. This ensures that founders and investors receive precise and actionable insights.

Knowledge Base Utilization:

The system effectively leverages VantagePoint Fund’s existing extensive knowledge base, which includes investment videos, podcasts, and other valuable resources. By indexing and processing this information, the solution offers comprehensive responses enriched with expert insights and historical data.

Interactive User Interface:

An intuitive and interactive user interface was developed to facilitate seamless interaction. Founders can easily navigate through the platform, ask questions, and receive instant feedback, thereby enhancing user experience and engagement.

State-of-the-Art Technology Standards:

Ensuring adherence to state-of-the-art technology standards, the solution was built with robust architecture and infrastructure design. It employs industry best practices in security, scalability, and performance, providing a reliable and efficient tool for users.

Architecture

• Data Integration: Connects to multiple data sources (YouTube, Crunchbase, PitchBook, blogs, and documents) using AWS Lambda and Lyzr-SDK Connector to fetch and preprocess data.
• RAG Process: Uses a scheduler and AWS Lambda functions to periodically update and retrieve relevant information from the data sources, storing it in Pinecone Vector Store for efficient access.
• User Interface: Provides a responsive interface with sign-in/sign-up pages and a chatbot canvas, built using React.js and powered by GPT-4 for interactive and intuitive user interactions.
• Query Processing: User queries are handled by an API Gateway and processed through the Lyzr-SDK Questions and Numerical Insights, utilizing a knowledge base and numerical data to generate accurate responses.
• Generative AI: Leverages GPT-4 and Lyzr-SDKs to super-serve questions related to scaling AI startups, ensuring responses are contextually relevant and enriched with expert insights.
• Data Storage and Management: Utilizes AWS services like S3 for data storage, ECS for containerized applications, and a datastore with a knowledge graph for managing relationships and training data, ensuring robust and scalable data management.