Basalt turns raw event streams into typed, joined, queryable tables in milliseconds. One layer for ingest, transforms and serving.
Every product runs on events. Clicks, payments, sensor ticks, model calls. Most of them cool slowly in a warehouse, hours late and full of cracks. Basalt sets them the moment they land: typed, joined, indexed and ready to query while they still matter.
-- materialized, incrementally maintained CREATE TABLE revenue_by_minute AS SELECT window(ts, '1 minute') AS minute, p.region, sum(p.amount_cents) / 100 AS revenue, count(DISTINCT s.user_id) AS buyers FROM stream.payments p JOIN stream.sessions s USING (session_id) WHERE p.status = 'succeeded' GROUP BY 1, 2;
Point Kafka, Postgres CDC, webhooks or the SDK at Basalt. Events land in an ordered, replicated log with exactly-once delivery, even when producers retry.
Write plain SQL. Basalt maintains every view incrementally, so joins, windows and aggregates stay current with each new event instead of on a nightly schedule.
Every table is an API the moment it exists. Read it from your app, dashboards or models with single-digit millisecond latency at p99.
Idempotent writes with per-key ordering. Retries, replays and network splits never double-count a payment.
Add, rename or retype a field without a migration window. Old readers keep working, new readers see the new shape.
Query any table as of any second in the last 30 days. Debug last Tuesday's numbers with last Tuesday's data.
Read replicas follow your users. Writes stay consistent, reads stay close, and failover takes under a second.
Write policies once in SQL. Basalt enforces them on every API, replica and export, keyed to your auth provider.
Rebuild a view across two years of history while the old version keeps serving. Swap atomically when it catches up.
Across all public regions, trailing 90 days.
Peak sustained load on one production cluster.
Financially backed on Scale and Enterprise.
From Reykjavík to Singapore, more every quarter.
Typed SDKs for TypeScript, Python and Go. Define a source, write a view, read the result. Basalt handles partitioning, retries, scaling and the 3 a.m. pages.
Read the docsimport { basalt } from "@basalt/sdk"; const db = basalt({ project: "acme-prod" }); await db.source("payments", { kafka: "payments.v3" }); await db.view("revenue_by_minute", ` SELECT window(ts, '1m') AS minute, sum(amount) AS revenue FROM payments GROUP BY 1`); const rows = await db.table("revenue_by_minute").last(60); // 3.1 ms
from basalt import Basalt db = Basalt(project="acme-prod") db.source("payments", kafka="payments.v3") db.view("revenue_by_minute", """ SELECT window(ts, '1m') AS minute, sum(amount) AS revenue FROM payments GROUP BY 1""") rows = db.table("revenue_by_minute").last(60) # 3.1 ms
$ npx basalt init acme-prod $ basalt source add payments --kafka payments.v3 $ basalt view apply ./views/revenue.sql $ basalt deploy --region eu-north-1
“We deleted fourteen pipelines and a Kafka team's worth of on-call in one quarter. Our fraud model now sees a payment four milliseconds after it happens.”
“Basalt is the first data tool our product engineers use without filing a ticket to the data team. They write SQL, they get an API, they ship.”
“Time travel paid for the contract in the first month. Finance asked why a number changed, and we answered with the exact rows from that morning.”
For side projects and prototypes that need real streaming.
For production workloads with SLAs and a team behind them.
For regulated data, dedicated clusters and private networking.
For side projects and prototypes that need real streaming.
For production workloads with SLAs and a team behind them.
For regulated data, dedicated clusters and private networking.
Free up to 10M events a month. No card, no sales call. Your first pipeline is live in about four minutes.
