Contextual AI

Contextual AI is a solution that demonstrates how to build interactive, AI-powered dashboards. It allows users to not only visualize data but also interact with it conversationally to gain deeper insights. The solution showcases how to integrate data APIs, AI agents, and dynamic query generation (like Natural Language to SQL) to create a rich, contextual user experience.

Architecture

The solution is composed of a frontend UI, a data access API, and an agent API powered by generative AI. It supports different types of widgets for data interaction.

Key Features

• Deterministic Widgets: Widgets that fetch data using pre-defined SQL queries via a Data Access API.
• Dynamic Widgets: Widgets that use an Agent API to generate SQL queries from natural language, allowing for free-form data exploration.
• AI-Powered Analysis: Users can select data points and ask questions. An AI agent fetches relevant data, analyzes it, and provides answers.
• Search Integration: A search widget integrates with Vertex AI Search to provide search and summarization capabilities.

Update 2026-03-04:

This project has been archived. You can still access the code by browsing the repository at commit dd333d5