Whatnot details AI-powered observability and analytics on Snowflake

To overcome bottlenecks in data access, Whatnot adopted AI-powered tools, evolving from a Slackbot to integrating semantic views with LLMs. Their latest phase uses Snowflake Cortex Agents (Hex Threads) to provide a conversational interface for data analysis, allowing employees to query data as easily as typing.

Whatnot is extending its AI analytics capabilities to external sellers through the Whatnot Seller Hub. This feature embeds Cortex Agents, ensuring data privacy with row-level security, allowing sellers to receive instant updates via natural language text queries.

Snowflake now offers AI-assisted observability workflows, allowing users to create system alerts via conversational requests within Snowsight. This simplifies the process, enabling users to define alerts for warehouse performance anomalies or cost spikes without writing extensive SQL code.

To address the challenges of real-time monitoring in a decentralized environment, Snowflake enhanced its native telemetry engine with next-generation event tables via Snowflake Trail. This update makes event ingestion 10x faster, reducing the cost and delay associated with comprehensive logging.

Whatnot, a live-shopping platform, outlined its strategy for managing hyper-growth using Snowflake for data observability and AI-driven analytics. They achieved this by decentralizing their data infrastructure using IaC, enabling business units to manage their own Snowflake warehouses and pipelines.

Within 90 days of launch, over 80% of Whatnot's employees adopted the agentic AI solution, with 17 departments achieving 100% utilization. This shift allows teams to perform complex analysis and rapidly iterate on business and product strategies without lengthy ad hoc data requests.