Sigma Computing

Sigma's homepage doesn't tell you who built it or why. There's no founder letter and no origin anecdote, just an About page link. What the company does reveal is the bet it's making: you shouldn't have to copy data out of your warehouse to analyze it or build on it. Its line "Go build it. IT approves." sums up the problem it targets. Business teams want to build their own tools, and data teams want governance to hold. Sigma tries to give both sides what they want by querying your cloud warehouse directly, so "security and governance always hold at the source." That bet is the closest thing to an origin story the page offers, and it explains most of the product.

The pitch has moved well beyond dashboards. Sigma now calls itself "the runtime for AI Apps and agents on governed warehouse data," with one control plane for permissions, audit, lineage and change management. You can build in chat, a familiar spreadsheet interface, SQL or Python. The spreadsheet UX supports write back, which matters if you want finance or ops people entering data rather than just reading it. On top of that sits an agent layer that can "trigger actions in external systems," such as updating a revenue forecast when an order tops $10K or alerting a CSM team about medium churn risk. It connects to Snowflake, Databricks, ClickHouse, AWS, Azure and Google Cloud. The compliance list covers SOC 2 Type II, HIPAA, GDPR, and SSO with SCIM.

The honest read: this is an enterprise tool built for companies that already run a serious cloud warehouse. Sigma says it's trusted by 2,000+ enterprises, and its examples include an FP&A portal running variance math on Databricks Genie and an executive briefing workbook with a natural-language interface. If your data still lives in a pile of SaaS tools and CSVs, the warehouse-first design gives you nothing to stand on. There's a free trial, but the scraped pricing page is a contact form with no tiers, so expect a sales conversation before you know what it costs. If you already have Snowflake or Databricks and your analysts are tired of rebuilding the same extract for every team, Sigma deserves a serious look.