TLDR Data Reader Offer · $100 Free Credit

An open lakehouse.
Any engine. Your cloud or ours.

Apache Iceberg + Parquet under the hood. Trino, Spark, or DuckDB over the top. One governance layer across all of them. SaaS or self-install in your own Azure tenant. Databasin One writes schema-aware SQL with Claude + GPT — built in, not an add-on SKU.

Claim $100 Credit — No Card See the stack

$100 for TLDR readers · $50 public · no commit · no reservation · cancel anytime

TLDR × Databasin
$100 $50 public

2× the standard signup credit, applied automatically when you come in from this page. Compute billed by the minute. AI billed by the token. Stop using it, stop paying.

Start with $100 →
Open · Flexible · Secure · Intelligent

Four things the hyperscalers won't give you.

Every pillar is a structural property of the platform, not a pricing tier. No add-on SKUs, no "contact sales" for the interesting bits.

Pillar 1 · Open

Open tables. Open engines. No dialect lock-in.

Your data lives in Apache Iceberg + Parquet — readable by anything that speaks Iceberg. Three engines, one lakehouse, zero lock-in.

  • Apache Iceberg + Parquet — vendor-neutral open formats
  • Apache Trino · Apache Spark · DuckDB — all read/write the same Iceberg tables
  • One governance layer — row/column policies, audit, lineage — enforced across every engine
  • No proprietary dialect. No egress fees to get your data back.
Read the architecture →
Pillar 2 · Flexible

SaaS in our cloud — or self-install in yours.

Run Databasin as SaaS for fastest time-to-value, or deploy the same platform into your own environment. Zero vendor logical access. Same APIs, either way.

  • SaaS: start in minutes, no infra to stand up
  • Self-install: Databasin deploys into your own cloud subscription
  • Your data never leaves your tenant — we can't see it
  • Switch modes later without re-architecting — same platform, same UI
See the platform →
Pillar 3 · Secure

Built inside a hospital. HIPAA is table stakes.

Databasin was co-created at Washington University School of Medicine. Security wasn't retrofitted — it's the design constraint.

  • HIPAA-compliant by default · BAA available
  • PHI never leaves your environment — by architecture
  • Row- and column-level policies enforced at the governance layer
  • Full audit trail of every query, export, and credential use
Our origin story →
Pillar 4 · Intelligent

Databasin One — schema-aware AI, built in.

Not a chatbot stapled to a docs page. An agentic analyst that indexes your catalog, picks the right gold tables, writes the SQL, runs it, and ships the chart.

  • Claude + GPT — multi-model routing for planning vs. generation
  • Reads the semantic model, not just the schema — metric definitions respected
  • Discover mode for exploratory analysis against your own files
  • Zero add-on SKU. Zero "contact sales."
More on Databasin One →
The stack, end to end

Point it at your lake. Pick your engine. Query.

The governance layer is the same regardless of which engine is executing. Switch from Trino to Spark for a heavy batch job, to DuckDB for a local dev loop — without moving or re-formatting a byte of data.

  • Storage: Apache Iceberg + Parquet
  • Compute: Apache Trino · Apache Spark · DuckDB
  • Governance: row/column policies · audit · lineage · masking
  • Medallion: bronze (raw) → silver (transformed) → gold (governed)
  • AI layer: Databasin One · Claude + GPT · schema-aware SQL · chart generation
  • Deploy: SaaS or self-install in your own cloud
Technology · deep dive

Every architectural decision has a reason.

Why open formats. Why three engines instead of one. How the governance layer enforces policies across all of them. Why PHI never leaves your environment by architecture — not configuration.

Read the architecture → See the four modules →
Databasin One

An analyst that knows your schema on day one.

Databasin One is not a copilot bolted onto a text editor. It reads your semantic model, routes between Claude and GPT based on the task, generates governed SQL, runs it against your gold layer, and returns a chart you can pin to a dashboard.

  • Multi-model. Claude for reasoning over schema; GPT for fast refinement and code.
  • Semantic-aware. Uses your defined metrics, not guessed aggregations.
  • Governed. Only queries gold views. Row/column policies always applied.
  • Discover mode. Point it at a spreadsheet or folder of files — it figures out the shape.
Show me last quarter's top 10 facilities by net revenue, excluding write-offs.
Databasin One · Claude Joining gold.facility with gold.fact_encounter_financials on the semantic relationship you defined. Net revenue excludes adjustment codes 100–199. 
SELECT f.name, SUM(r.net_revenue) AS revenue
FROM gold.fact_encounter_financials r
JOIN gold.facility f ON f.id = r.facility_id
WHERE r.quarter = CURRENT_QUARTER() - 1
  AND r.adjustment_code NOT BETWEEN 100 AND 199
GROUP BY f.name
ORDER BY revenue DESC LIMIT 10;
Pay for what you run · Every rate published

Billed per minute. No reservation. No seats. No commit.

Compute rounded up to the minute. Stopped clusters don't bill. Every rate is on the pricing page — not behind "contact sales."

Lakehouse Compute
$2–6
/ node / hour
Pipelines & Automations
$5
/ hour
Databasin One
$15
/ 1M AI tokens
Storage
$24
/ TB / month
No annual commit No DBUs · No CUs · No credits-that-aren't-dollars $100 TLDR credit on signup
See every rate → Pricing
TLDR × Databasin

$100 in credit.
No card required.

Credit applied automatically when you sign up via this page.