Supernova

Supernova collapses the modern data stack into one product: ingestion, an encrypted Apache Iceberg lake and a query engine, plus an MCP server so an AI assistant can ask questions of the data without a BI seat in between. It is aimed at teams that need analytics but do not have a data team to assemble and operate the usual four-vendor pipeline.

What is in the box

• Syncs from 40+ sources in real time, including Stripe, HubSpot and PostgreSQL.
• An encrypted Iceberg data lake with table-level encryption. Because storage is open Iceberg, any engine that reads the format can read it too, and an existing warehouse can be attached with a zero-copy connection instead of a migration.
• TypeSQL, a schema-aware SQL layer that autocompletes across joins and type-checks queries before they run.
• Dashboards and models under Git version control, reviewed like code rather than edited in a web UI.
• A CLI for running data operations from a terminal or CI.

The AI angle

Supernova exposes the warehouse over MCP, so Claude or Codex can be pointed at it and answer questions in natural language against real tables — the distinguishing feature versus a conventional lakehouse, and the reason the schema-aware layer matters: the model gets types and join paths rather than guessing at column names.

Pricing

Usage-based, with no per-seat and no per-feature charges: compute at $0.15 per GB-hour, storage at $0.05 per GB-month, reads at $0.40 per million and writes at $5 per million, with AI tokens passed through at 1.5x cost.

Who it is for

Startups and small analytics teams that want a warehouse, a query surface and an AI entry point without stitching together an ELT vendor, a warehouse, a semantic layer and a BI tool — and who would rather pay for what they run than per analyst.