⚡ClickHouse

ClickHouse is built for blazing-fast analytical queries on huge datasets, and iDBQuery is built to ask those queries in plain English. Connect to ClickHouse Cloud, a self-hosted cluster, or a single-node instance over the native protocol (port 9000) or HTTPS (port 8443). Provide host, port, database, user, and password, and the schema across your databases is introspected in seconds. The AI sees tables, materialized views, the engine each table uses (MergeTree, ReplicatedMergeTree, AggregatingMergeTree, etc.), partition keys, sort keys, and projections — and uses that map to write SQL that hits the right indexes and the right rollup tables. Every chat question becomes a safe parametrized SELECT, returned as a chart, table, or dashboard. ClickHouse-specific features — array functions, `ANY`/`ALL` joins, window functions, `arrayJoin`, sampling clauses, materialized views as query targets — are all in the model's vocabulary. Connection details are encrypted at rest with strong symmetric encryption, used only at query time, and never returned by any API. The free tier — 1M tokens per month, three sources, five reports, no credit card — lets you point iDBQuery at a real ClickHouse instance and prove the workflow before paying. iDBQuery makes a billion-row OLAP store feel like a chat window.

• ClickHouse Cloud, self-hosted clusters, single-node
• Native protocol (9000) or HTTPS (8443)
• Schema introspection: tables, materialized views, engines, partition/sort keys
• AI prefers materialized views and rollups when they exist
• Array functions, window functions, sampling clauses supported
• Read-only enforcement at the connector layer
• Multi-database introspection on a single cluster
• Cross-source joins with spreadsheets or other databases