In this guide, you will build a real-time due diligence and investment research agent in Snowflake with Tavily Web Search.

Tavily is a purpose-built web search API designed specifically for AI agents. Unlike traditional search engines, Tavily provides a fast, structured, and relevance-optimized web access layer that enables agents to retrieve high-signal, up-to-date information with minimal noise.

You will learn how to create a Financial Agent that analyzes ticker-level fundamentals stored in Snowflake and enriches them with timely insights from across the web. Using Tavily’s real-time search capabilities, the agent retrieves recent news, regulatory updates, litigation developments, executive changes, and emerging risk signals that may not yet be reflected in financial statements.

By combining trusted financial reference data in Snowflake with Tavily’s up-to-date web intelligence, the agent performs contextual risk and opportunity analysis that goes beyond static datasets. This approach helps reduce blind spots between earnings cycles and surfaces early warning signals for buy-side, private equity, and corporate development workflows.

By the end of this guide, you will have a working Financial Agent that demonstrates how Snowflake and Tavily together enable intelligent, real-time financial analysis across structured and unstructured data sources.

By the end of this guide, you will have a working Financial Agent capable of analyzing a company’s fundamentals and enriching that analysis with real-time developments from across the web.

This Financial Agent can support a variety of investment and corporate analysis workflows, including:

By combining trusted financial reference data in Snowflake with Tavily’s real-time web intelligence, you reduce blind spots in traditional analysis and enable faster, more informed decision-making across structured and unstructured data sources.

Follow the steps below to configure Tavily Web Search within Snowflake.

1. Install Tavily from the Snowflake Marketplace

   • Navigate to Snowflake Marketplace.
   • Search for Tavily Search API.
   • Click on 'Get' and follow the prompts to install it in your account.
   

2. Provide Your Tavily API Key

   • After installation, click on 'Open'

     
   • When prompted, enter your Tavily API key to enable real-time search functionality (If you do not have an API key, you can create one at: https://tavily.com/).

     

   • External API access must be enabled in the API configuration settings to allow outbound calls (You should find it below the API config field).

3. Validate the Configuration

   • Once the API key is configured, click on 'Open Worksheet'.
     
   • Then run the default query by selecting your appropriate warehouse and ensuring the Database is set to TAVILY_SEARCH_API and the Schema is set to TAVILY_SCHEMA, as shown below.
     
   • You should see the query results displayed in the output console, similar to the example shown above.

You can follow the steps in this video for a quick setup:
Watch the setup video

There are two ways to bring financial fundamentals into Snowflake for use with the Financial Agent.

The most production-ready approach is to integrate structured financial datasets directly from the Snowflake Marketplace, such as LSEG Financials. However, for this demo, we will use a smaller custom dataset (created using the Alpha Vantage API) to keep the environment lightweight, transparent, and easier to understand while focusing on agent orchestration and tool integration.

For a production or enterprise use case, the recommended approach is to subscribe to a structured financial dataset directly from the Snowflake Marketplace.

Example dataset: LSEG Financials on Snowflake Marketplace

Steps:

This approach provides:

For this guide, we will use a smaller custom dataset to:

Steps:

1. In the Snowflake UI, navigate to the AI & ML tab and select Agents.

2. Click Create Agent, then provide a name, description, and relevant example questions for your agent.
   

   
3. Navigate to the Tools tab and add the Cortex Analyst tool.
   

   

   Add the Cortex Analyst Tool
   

   
4. Create a new custom tool for the Tavily Search API and configure its required parameters.
   

   

5. Click Save Updates to apply the updates.

6. Launch Snowflake Intelligence and verify that the agent has access to both configured tools.

   

User Prompt:

1. Snowflake Intelligence calls tavily_web_search first

   • The agent identifies that the question requires recent, time-bound external developments.
   • It invokes the Tavily Web Search Tool to retrieve regulatory and legal news.
   • Tavily’s fast and efficient search ensures:
     • Up-to-date information (last 30 days)
     • Relevant filtering around investigations and enforcement
     • Reduced hallucination risk by grounding responses in real web sources

2. Snowflake Intelligence then calls the Cortex Analyst tool

   • The agent generates SQL dynamically to retrieve financial fundamentals.
   • It queries revenue, net income, and other structured data from your Snowflake database.
   • The SQL execution establishes the financial baseline.

3. Synthesis
   • External regulatory findings (Tavily)
   • Structured financial metrics (Cortex Analyst SQL)
   • Combined valuation-aware reasoning

Without Tavily, the agent would lack fresh regulatory intelligence.
Without Cortex Analyst, it could not contextualize events against actual financial scale.

User Prompt:

1. Cortex Analyst Tool is invoked
   • The agent dynamically generates SQL to query key fundamentals (e.g., TTM revenue, market cap, net income).
   • It establishes a structured financial baseline directly from Snowflake tables.
   • SQL execution is visible in the reasoning trace.

2. tavily_web_search is triggered

   • The agent searches for recent business developments tied to forward cash flow drivers, such as:
     • Supply chain or logistics disruptions
     • Labor-related news (staffing, wage pressure, store operations)
     • Pricing strategy changes
     • Earnings guidance updates or major business announcements
   • Tavily delivers high-signal, recent results in real time.

3. Combined Financial + Business Intelligence

   • Financial scale and baseline metrics (from structured data)
   • Recent operational and business drivers (from Tavily search)
   • Synthesized implications for forward cash flow

Tavily’s fast and efficient search:

This transforms Snowflake Intelligence from a SQL-driven assistant into a contextual financial reasoning engine.

Congratulations on completing this exercise!

By integrating Tavily Web Search with Cortex Analyst, Snowflake Intelligence becomes a multi-step reasoning system capable of:

The Financial Agent moves beyond simple querying and becomes a grounded, decision-support system for due diligence and investment research.

You have now implemented a Financial Agent that bridges structured Snowflake data with live external intelligence — enabling real-time, grounded financial analysis.