Documentation

Overview

The Destination Query feature lets users explore and test queries against their connected destinations (e.g., data warehouses) through the AI Agent before building reports, models, or downstream transformations. It provides a fast, straightforward way to:

• Explore available schemas, tables, and fields in a connected destination
• Generate destination-specific queries (e.g., SQL)
• Execute queries using your existing destination credentials and review results
• Iterate quickly by refining queries with the AI Agent using returned errors and outputs

How to use the API Discovery feature

Prerequisites

To use the API Discovery feature, you'll need:

• Active connection credentials for the data source you want to explore
• Access to the AI Agent

Guide

1. Open the AI Agent chat interface and type your request about exploring a specific API (e.g., "Query campaign_a table in Google Big Query destination").
2. When prompted, select the destination you want to query (if multiple destinations or environments are available).
3. Select one of your existing destination connections when prompted to use its credentials for query execution.
4. The I Agent retrieves relevant destination metadata and generates an appropriate query for your request, then executes it using the selected credentials.
5. Review the query output (or any returned error) and continue iterating—ask the AI Agent to refine filters, joins, date ranges, aggregations, or limits as needed.
   

The AI Agent will guide you through the process by suggesting relevant tables/fields, generating destination-specific queries, and executing test runs based on your inputs.

Technical details

The Destination Query feature utilizes several key tools and components:

• Metadata & schema retrieval: Pulls the destination catalog (schemas/tables/columns) and keeps it current for accurate query generation.
• Credentials manager: Provides a list of available credentials for active destination connections and securely applies them to query execution.
• Query execution engine: Submits queries to the destination, captures results, and returns errors/warnings to help the AI Agent refine the query.