Spice

Spice is an open-source SQL query and AI compute engine, built in Rust, that lets developers query operational and analytical data without traditional ETL pipelines. It federates queries across more than 30 data sources, including databases, data warehouses, and data lakes, and accelerates them through local materialization and caching. Alongside SQL, Spice supports hybrid search that combines vector, full-text, and keyword retrieval in a single interface, and it can invoke local or hosted large language models directly from a query for tasks like retrieval-augmented generation and in-line analytics. The engine runs anywhere from a laptop to a Kubernetes sidecar, an edge microservice, or a fully managed Spice.ai Cloud Platform, giving teams flexibility in how they deploy it. It also offers AI sandboxing, restricting applications and agents to least-privilege access over specific datasets. Spice.ai, Inc. was founded in 2021 in Seattle by Luke Kim and Phillip LeBlanc, both former Microsoft engineers, and has raised early-stage funding to build out the platform. Spice is aimed at enterprise data teams and AI developers building data-grounded applications and agents, particularly in cybersecurity, financial services, and SaaS, where fast, unified access to distributed data matters. One limitation is that fully realizing its capabilities, such as the 100TB+ of preloaded datasets, requires the managed cloud tier rather than the open-source version alone.