The lakehouse with free queries.
Columnar analytics over your AI-native data. Run unlimited SQL over your Parquet — pay only for storage. Everyone else bills your compute. We don't.
Everyone else bills your compute
Our in-memory DuckDB runs per request — no standing cluster — so queries are free.
| Provider | Query / Compute | Storage |
|---|---|---|
| AINative Lakehouse | Free — unmetered | $0.15/GB |
| MotherDuck | $0.60–36/hr compute | $0.04/GB |
| Snowflake | $2–4 / credit | ~$23/TB |
| Databricks SQL | ~$0.70 / DBU | + storage |
| BigQuery | $6.25/TB scanned | — |
How it works
A fresh DuckDB engine per request reads your Parquet directly. No cluster to provision or pay for.
Every query is confined to your project partition (tenant={project_id}/). You only ever see your own data.
POST /database/lakehouse/query with SELECT SQL. Ingest with /lakehouse/ingest. That's the whole surface.
Query in one call
curl https://api.ainative.studio/api/v1/projects/$PROJECT_ID/database/lakehouse/query \
-H "X-API-Key: $AINATIVE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"sql": "SELECT status, count(*) FROM events GROUP BY status"}'Your agent writes once. You get everything.
One write becomes a row (ZeroDB), a bulk-queryable record (Lakehouse), a semantic memory, and a node in your knowledge graph — no ETL, no syncing, no separate systems. Competitors need Postgres + Snowflake + Pinecone + Neo4j and pipelines between them.
OLTP tables and vector search your agents provision themselves — zero human setup.
/api/v1/projects/{id}/database/*Free, unmetered SQL over your Parquet. In-memory DuckDB per request — no cluster.
/database/lakehouse/querySemantic recall plus an entity graph — centrality, contradictions, relationships.
/api/v1/public/memory/v2/*Simple, honest pricing
Free unlimited queries. Pay only for storage.
Free queries
10 GB included
Queries always free
Free queries
25 GB included
then $0.15/GB
Free queries
100 GB included
then $0.10/GB
Frequently asked questions
No. Queries are free and unmetered — an in-memory DuckDB engine runs per request over your Parquet, with no standing cluster and no compute bill. You pay only for storage ($0.15/GB on Pro, 10 GB free).
MotherDuck, Snowflake, Databricks, and BigQuery all bill compute (DBUs, credits, hours, or TB scanned). AINative bills nothing for queries — possible because it runs per-request in-memory DuckDB instead of a persistent cluster.
Queries are always free. Storage: 10 GB free, then $0.15/GB (Pro, 25 GB included) or $0.10/GB (Enterprise, 100 GB included).
POST SELECT SQL to /api/v1/projects/{project_id}/database/lakehouse/query with your X-API-Key. Every query is confined to your own project partition.
They are one platform. One agent write becomes a row (ZeroDB), a bulk-queryable record (Lakehouse), a semantic memory, and a knowledge-graph node — no ETL between separate systems.