How Fireblocks Uses Snowflake Cortex AI to Handle 40-50% of Data Queries Autonomously

Fireblocks provides the infrastructure behind much of the digital asset economy — its platform secures and moves $10 trillion in transactions across 550 million crypto wallets for hundreds of institutions and fintechs. The scale of data that generates is enormous, and it was growing faster than Fireblocks’ infrastructure could keep pace.

Before Snowflake, even basic queries across billions of rows were taking up to 20 minutes. The company operated 15 separate data domains, each with its own data and semantic layer, making cross-domain analysis nearly impossible. Sales reps couldn’t get revenue insights while engineers were pulling on-chain data — the systems didn’t talk to each other.

Fireblocks built a new data platform on Snowflake, running on AWS. With all 15 domains flowing into the AI Data Cloud — from Salesforce analytics to Zendesk insights to on-chain transaction data — any team can now query across the full data estate. During a 45-day internal hackathon, the engineering team built Fire Genie: an AI agent using Snowflake Cortex Agents and semantic views that lets customers query only their own data in a secure environment through a natural language API.

The results shifted how the business operates. Fire Genie now handles 40–50% of all data queries autonomously, replacing what previously required direct analyst involvement. The AI-powered efficiencies across the platform save the equivalent of two full-time analysts per month, freeing the team to focus on higher-value work.

With row-level security and RBAC enforced at the Snowflake layer, Fireblocks is able to offer customers genuine self-service — without the risk of data leakage between accounts. The platform now scales with transaction growth rather than headcount.