Building Real-Time Pricing Intelligence in Snowflake with Upriver + Nimble

Imagine you're managing pricing for a consumer packaged goods brand with 200+ SKUs sold through dozens of authorized distributors. Your products are listed on Amazon, Walmart.com, specialty retailers, and direct-to-consumer channels. Each distributor sets their own prices and runs their own promotions.

The problem: You have no centralized, real-time view of what your products actually cost across the market. Your team manually checks a handful of distributor sites each week, but with hundreds of SKUs and dozens of channels, you're seeing less than 5% of your market exposure. The data you need—SKU catalog, distributor relationships, product attributes—already lives in your Snowflake warehouse, but it's static.

What if you could automatically enrich that warehouse data with live market intelligence?

Real-time access to pricing, discounts, inventory availability, and product listings across all distributors would enable you to:

• Optimize pricing dynamically — Adjust prices based on real-time competitor moves and market conditions
• Enforce MAP policies — Instantly detect unauthorized discounting or price violations
• Protect margins — Identify arbitrage opportunities where distributors undercut your own channels
• Respond to stockouts — Reroute marketing spend when key distributors are out of stock
• Benchmark performance — Compare promotional effectiveness across different retail channels
• Feed pricing models — Provide live market data to machine learning algorithms for automated pricing

In this post, we'll walk through how Upriver's AI agent, integrated with Nimble's real-time web data, can build this entire pipeline within your existing Snowflake warehouse—no manual scraping, just automated market intelligence flowing into your data infrastructure.

Let's dive into how it works.

Starting State: Your Existing Data Warehouse

Your Snowflake warehouse already contains thousands of tables from ERP systems (SAP, Oracle), CRM platforms (Salesforce), and other business systems. It's your single source of truth for product data, customer records, sales transactions, and supply chain operations.

Buried within this warehouse are two tables:

These tables tell you what products you sell and who's authorized to sell them. What they don't tell you is what's actually happening in the market right now—the prices customers see, the discounts being offered, or whether products are in stock.

The challenge: Among thousands of tables, how do you identify the right data, join it intelligently, and enrich it with live market information? This is where Upriver's AI agent comes in.

The Solution: Automated Enrichment with Upriver + Nimble

By connecting Upriver to your Snowflake environment, it learns your data context—understanding your existing tables, schemas, and relationships the way a seasoned data engineer would. This means you can use a single natural language prompt, and Upriver handles everything else – let’s see how that would work in this example:

"Build a continuously updating dataset that shows the current price each distributor is charging for our products."
‍

By connecting Upriver to your Snowflake environment, it learns your data context—understanding your existing tables, schemas, and relationships the way a seasoned data engineer would. This means you can use a single natural language prompt, and Upriver handles everything else: finding the right tables, mapping relationships, building transformations, and deploying the pipeline.

Upriver's AI agent analyzes your Snowflake warehouse, identifies the relevant INVENTORY.SKUS and INVENTORY.DISTRIBUTORS tables, and determines how to join them based on product relationships. Using these tables, upriver autonomously builds a pipeline that knows how to enrich data in real-time using Nimble's APIs

Upriver Writes the Enrichment Function Using Nimble to Collect Live Web Data

Upriver automatically generates a Snowflake User-Defined Table Function (UDTF)—a custom function that runs inside your data warehouse and integrates Nimble's real-time web scraping capabilities.

Normalizing the data

Before enrichment can happen, the SKU and Distributor tables need to be properly joined. Upriver's agent automatically analyzes the schema relationships and determines how to connect products with their distributors—mapping product names, UPCs, or other identifiers across both tables. It then sets up the necessary intermediate tables and transformation steps to create a clean, joined dataset ready for enrichment. Only after this foundation is in place does Upriver generate the UDTF and integrate it into the pipeline, ensuring that live web data flows into properly structured tables.

Enriching the tables

The UDTF that Upriver writes includes python code that calls Nimble's search tool with the appropriate parameters (product name, UPC, target distributor website). Behind the scenes, Nimble's Web Search Agents autonomously:

1. Navigate to each distributor's website and locate the relevant product pages
2. Extract current pricing, discounts, and availability without requiring custom scrapers
3. Return structured data that integrates directly into your Snowflake tables

No configuration. No script maintenance. Just live market data flowing into your warehouse.

How It Runs

Once Upriver deploys the UDTF, it can be:

• Triggered on-demand — Run manually when you need immediate pricing updates
• Automated in a pipeline — Upriver also creates a distributor-prices pipeline that joins your SKU and Distributor tables, calls the UDTF for each product-distributor combination, and materializes live results into an enriched output table

The Result

With Upriver and Nimble, a task that would previously take days instead takes minutes. You get a continuously updated table that combines your internal data with live market intelligence:

ENRICHED.DISTRIBUTOR_PRICES

Your static warehouse data is now enriched with real-time market reality—all running automatically within your existing Snowflake environment.

From Static Data to Dynamic Intelligence

What you've seen here is a fundamental shift in how retail and CPG companies can approach market intelligence. Instead of manual price checks and disconnected data sources, you get:

• Automated enrichment — Live pricing, discounts, and availability data flowing continuously into your warehouse
• Built in minutes — An AI agent constructs the entire pipeline without custom development
• Always current — Real-time market data updates automatically, not weekly spreadsheets
• Actionable insights — Feed dynamic pricing models, detect MAP violations, optimize channel strategy

The combination of Upriver's intelligent data pipeline builder and Nimble's real-time web data transforms your static product catalog into a living, breathing view of your market position. You're no longer reacting to pricing changes days after they happen—you're seeing them as they occur and responding accordingly.

Your data warehouse already contains the foundation. Now it can tell you what's actually happening in the market, in real-time. Unlike traditional data pipelines that capture a snapshot in time, this table continuously updates as the market evolves—capturing price changes, new promotions, and inventory fluctuations as they happen.

This example shows the power of automating UDTF development with Upriver and web data collection with Nimble – to try it yourself, check out the code we open sourced to begin automating your data enrichment.

‍