Real-time Analytics
Fireblocks needed to query billions of rows across 15 data domains, unify customer and internal analytics, and expose digital-asset insights securely without relying on manual analyst workflows or risking data leakage.
Reported impact
Fireblocks now handles 40-50% of analysis queries through AI agents, saves the equivalent of two full-time analysts per month, gives customers secure self-service answers about wallets and assets, and strengthens retention with more personalized analytics.
Fireblocks rebuilt its analytics stack on Snowflake running on AWS so both internal teams and customers can query digital-asset data at scale without exposing one another's data. On-chain telemetry, wallet and asset activity, revenue data, Salesforce analytics, and Zendesk insights are loaded into Snowflake across 15 data domains to create one governed data foundation. On top of that foundation, the team built Fire Genie in a 45-day hackathon so customers can ask natural-language questions about wallets and assets through an API; RBAC and row-level security restrict each tenant to its own data. Fireblocks then reused the same approach internally with Cortex Agents and semantic views so product teams can go from a Jira ticket to the relevant product metrics, while Streamlit apps publish live shared reports instead of screenshots. The result is an AI-assisted analytics environment that scales to billions of rows, preserves security, and frees analysts to focus on deeper insight.
Asset, wallet, revenue, Salesforce, and Zendesk data from 15 domains flows into Snowflake on AWS, where it is secured with RBAC and row-level security. Cortex Agents with semantic views power Fire Genie and an internal natural-language analytics agent, while Streamlit surfaces live shared reports and APIs return customer-specific answers about wallets, assets, and product behavior.
At first, I didn’t think we’d be able to do this without a massive risk of data leaks.
But then in a 45-day hackathon, we developed our AI agent, Fire Genie, on Snowflake.
Fireblock’s internal agent now powers over 2,000 analysis queries per month, roughly 40-50% of all queries.
This saves the equivalent of two full time analysts’ worth of capacity every month.
With billions of rows of data across our platform, asking even basic queries was taking up to 20 minutes.
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Snowflake Sample Archived · Oct 8, 2026 Mercury replicate the company’s critical operational database into Snowflake fast enough to support trusted analytics, product decisions, and risk use cases without overloading production systems. 0 Replication incidents since the migration Snowflake · Snowflake Openflow Snowflake Sample Archived · Sep 1, 2025 RavenPack Delivering multi-petabyte financial data to clients across regions and clouds through FTP and ETL pipelines was too slow, siloed, and expensive for RavenPack's growing analytics and AI business. 14x lower egress costs AWS Bedrock · Bigdata.com · Egress Cost Optimizer Snowflake Sample Archived · Sep 16, 2024 Open Insurance replace time-consuming, siloed insurance data management with a secure self-service platform that could scale analytics, mask sensitive data, and accelerate partner and acquisition integration. 1 month to integrate 16 data sources from an acquired company ChatGPT · Snowflake · Streamlit 