Supernova home

Everything in one place.

Sign upConnect your data to Claude
MCP

Made for Claude and Codex

Get set up quickly — connect to mcp.supernova.ai/mcp and every table you sync is one question away.

claude mcp add --transport http supernova https://mcp.supernova.ai/mcp
Titan

A powerful data stack

The lake, the catalog, and the engine — one system, built to be handed the keys.

Encrypted Iceberg lake
State of the art open tables in object storage — encrypted end to end, never locked in.
Zero copy connections
Any engine that speaks Iceberg reads the lake directly. No exports, no drift, no second copy.
Time travel
Every table is versioned. Query last Tuesday the same way you query right now.
Rust, all the way down
Our own writer, catalog, and query runtime. Cold starts in milliseconds, not minutes.
CLI

A Swiss army knife for your data

One binary: connect sources, run queries, tail changes, wire up Claude — from your laptop or from CI.

TypeSQL

What if someone made a TypeScript for SQL? We have.

Schema-aware SQL that autocompletes across joins and type-checks before it runs.

select c.email, sum(ch.amount) as revenue
from hubspot.contacts c
join stripe.charges ch on ch.email = c.email
where ch.status = 'succeeded'
group by c.email
order by revenue desc
inferred { email: string | null; revenue: Money<"usd"> }[]
where ch.stauts = 'succeeded'
column 'stauts' does not exist on stripe.charges — did you mean 'status'? caught before it ran
emailrevenue
[email protected]$48,120.00
[email protected]$31,876.50
[email protected]$27,433.25
[email protected]$19,204.00
[email protected]$14,882.75
0.09s · 5 of 3,411 rows

Every column in your lake is typed — including money, in its currency. Queries type-check against the live schema as you write them, autocomplete across joins, and refuse to run when they can't return what you asked for.

FTL

Connectors that keep up

Thirty-plus sources today, written in a language built for exactly this — with Postgres and MongoDB streamed in real time.

Postgresreal-timeMongoDBreal-timeStripeHubSpotSalesforceShopifyNetSuiteMySQLGoogle AdsGoogle AnalyticsGoogle SheetsFacebook AdsMicrosoft AdsSnapchat AdsReddit AdsIntercomZendeskHelp ScoutJiraKlaviyoMixpanelAmplitudePendoGainsightAirtableApple App StoreSurveyMonkeySendGridPostmarkPandaDocPardotCanny

Each connector is a few hundred lines of FTL, our flat table language — small enough to read in one sitting, precise enough to capture every field. When an API changes upstream, the fix ships in hours. When you need a source we don't have yet, so does the connector.

Pricing

Transparent pricing

Pay for what you actually use. Nothing per seat, nothing per feature.

Rows moved$1 per million
Query compute$0.15 per hour
Storage$0
Seats$0
B2B SaaS, ~60 people
9 sources · 120M rows/mo
before$4,700/mo
with Supernova$610/mo
Fintech, Series B
14 sources · 300M rows/mo
before$12,400/mo
with Supernova$1,900/mo
E-commerce brand
5 sources · 40M rows/mo
before$2,100/mo
with Supernova$240/mo
Full pricing →
FAQ

Questions

Where does my data live?
In an encrypted Iceberg lake in object storage. Point any engine that speaks Iceberg at it, or take it with you — it was never locked in.
Do I need a warehouse?
No. Supernova is the pipeline, the lake, and the query engine in one. If you already have a warehouse, zero copy connections let it read the lake directly.
How does Claude connect?
Over MCP. Add mcp.supernova.ai/mcp to Claude or Codex and your data is queryable in about a minute.
What does it cost?
You pay for rows moved and compute used — no per-seat fees, no platform fee. See the pricing page for the full rate card.
Sign upConnect your data to Claude