Overview

AI assistants deliver significantly better results when they can access high-quality premium databases.
This reduces hallucinations and improves reliability for domain-specific answers.

Two paths for databases

In meinGPT, databases are intentionally separated from general connectors.

1. Custom Database (lightweight)

For fast access to existing SQL databases with a read-only focus.

• direct access to external databases
• quick schema exploration and SQL querying
• best for fast analysis and operational questions

See: Custom Database

2. Dataset Manager (beta)

For deeper structured-data workflows with a dedicated dataset layer.

• structured data ingestion and introspection
• documented schemas, stored queries, and search options
• suitable for stable, reusable agent setups

See: Dataset Manager (beta)

Queryable data sources

Besides classic SQL databases, this section also includes query-first data connectors:

• Commercial Register
• Context7
• Google Maps
• EU Custom Tariffs (TARIC)