Can I add a database not on the list?

Yes. The Enterprise plan includes the development of dedicated extensions for proprietary or legacy engines not yet supported.

Are generated queries always parameterized?

Yes. The AI always produces parameterized SQL/NoSQL — no string concatenations, no SQL injection. Values are bound at execution time.

How do you handle engine-specific errors?

Native error codes are normalised into uniform messages (permission denied, constraint violation, timeout, etc.), so feedback is consistent regardless of the underlying engine.

Are credentials safe?

Credentials are encrypted at-rest with enterprise-level encryption and read only at execution time. Never sent to the AI, never logged in plain text.

Supported databases

9 databases. One interface.

Every dialect is an independent plugin with schema snapshot, executor and dedicated validator. No compromises on the SQL produced.

Get started for free Explore the product

PostgreSQL
SQL

The most beloved open-source relational database.

Full support for advanced SQL, custom types, full-text search and analytics and geospatial extensions.

versions

databases.details.postgres.versions
- Transactional OLTP
- Data warehousing
- SaaS applications
- GIS / geospatial
MySQL
SQL

The de-facto standard of the web.

Full MySQL dialect with window functions, recursive CTEs and search on JSON data.

versions

databases.details.mysql.versions
- Web applications
- E-commerce
- CMS / WordPress
- Operational analytics
SQL Server
SQL

The reference enterprise RDBMS for the business world.

Supported enterprise dialect: temporal tables, encrypted data and advanced features.

versions

databases.details.sqlserver.versions
- ERP and management software
- Business reporting
- Corporate data marts
- Enterprise stacks
Oracle
SQL

Mission-critical database for the enterprise world.

PL/SQL, sequences, multi-tenant schemas, partitioning and optimisation hints.

versions

databases.details.oracle.versions
- Core banking
- Insurance
- Public Administration
- Legacy data centres
MariaDB
SQL

The community-driven MySQL fork.

Compatible with the MySQL dialect with proprietary engine extensions.

versions

databases.details.mariadb.versions
- Open-source web applications
- Shared hosting
- MySQL drop-in alternatives
SQLite
SQL

The world's most widely deployed embedded database.

File-based, zero-config, perfect for prototypes, mobile apps and single-writer use cases. Full SQLite dialect support.

versions

databases.details.sqlite.versions
- Prototypes and demos
- Desktop / mobile apps
- Edge computing
- Automated testing
MongoDB
NoSQL

The most popular JSON document store.

Full aggregation pipeline, geospatial and full-text indexes. Support for standalone, replica set and managed cloud instances.

versions

databases.details.mongo.versions
- Content management
- Product catalogue
- Event logging
- Real-time analytics
Cosmos DB
NoSQL

The cloud-native multi-model database.

Support for multiple APIs (SQL and document), aware of partition key and consistency levels.

versions

databases.details.cosmos.versions
- Global applications
- IoT telemetry
- Real-time gaming
- Omnichannel retail
Redis
KV

The fastest in-memory key-value store.

Commands on data structures, streams, pub/sub and JSON / search modules. Support for clusters and high availability.

versions

databases.details.redis.versions
- Application cache
- Rate limiting
- Leaderboard / ranking
- Session store

Plugin architecture

One contract. Many dialects.

Each engine implements the same IDbPlugin contract. The AI produces dialect-aware SQL, the validator blocks invalid queries, the executor runs with bound parameters.

step 01

Schema extraction

The schema (tables, columns, indexes, foreign keys) is read and cached. The AI only receives this snapshot, never the credentials.
step 02

Dialect-specific validation

The AI-generated query is validated against the dialect grammar before any execution.
step 03

Controlled execution

Execution with always-bound parameters, configurable row limit and uniform error messages independent of the engine.

Frequently asked questions

Can I add a database not on the list?: Yes. The Enterprise plan includes the development of dedicated extensions for proprietary or legacy engines not yet supported.
Are generated queries always parameterized?: Yes. The AI always produces parameterized SQL/NoSQL — no string concatenations, no SQL injection. Values are bound at execution time.
How do you handle engine-specific errors?: Native error codes are normalised into uniform messages (permission denied, constraint violation, timeout, etc.), so feedback is consistent regardless of the underlying engine.
Are credentials safe?: Credentials are encrypted at-rest with enterprise-level encryption and read only at execution time. Never sent to the AI, never logged in plain text.

Can't find your database?

The Enterprise plan includes the development of custom plugins for proprietary or legacy engines.

Get started for free Contact Sales

9 databases. One interface.

PostgreSQL

MySQL

SQL Server

Oracle

MariaDB

SQLite

MongoDB

Cosmos DB

Redis

One contract. Many dialects.

Schema extraction

Dialect-specific validation

Controlled execution

Frequently asked questions

Can't find your database?