Database chat

Database chat is AI powered chatbot, which knows everything about your database. It understands both schema and database data. With your explicit permission, it could even execute SQL queries for retrieve data.

DbGate - database chat

Basic concepts

Natural conversation – ask questions in any natural language and get instant answers or SQL queries ready to run.

Context-aware intelligence – the model understands your database schema and live data, enabling accurate suggestions for joins, filters, and aggregations..

Iterative workflow – review results and make changes with simple commands like “add a date filter” or “group by customer” - the query updates instantly.

You stay in control – SQL queries are only executed with your explicit approval

Use any language, which is supported by OpenAI (almost all languages - English, French, German, Spanish, Portuguese, Italian, Japanese, Chinese, Korean, Czech, etc.)

Execute SQL queries - safety first

Database chat could execute SQL queries, to obtain data needed for answering your questions. There are several modes, how this feature works:

  • Ask before execute SQL (default) - DbGate will ask you before each query, which should be executed

  • Execute SQL automatically - DbGate will check and then execute queries without confirmation. Before executing, DbGate checks this query, and if it is not pure SELECT query, it requires your confirmation

  • Never execute, only suggest SQL - chat will show you query, you could open it in new SQL tab and execute by own. In this model, no data from your database are sent to LLM, so chat could not generate responses from your data

Error processing

LLM models are not ideal, so they sometimes suggest query, which is not valid or doesn’t work. It this case, DbGate shows error, and this error is also sent to LLM as the result of the query. In most cases, LLM could automatically analyse error message and suggest new correct query.

Use chat as consultant

Database chat could work also with definitions of your stored procedures and views. So you could ask questions like:

  • Explain procedure proc_ImportProjectFromHelios
  • Which procedure is used for importing into table Project
  • Which tables are affected by procedure proc_ImportUsersFromAD

DbGate also shows your, which views and stored procedures were analysed to respond your question.

Used LLM models

We use OpenAI GPT models (currently, you could choose between GPT-4o mini and GPT-4.1 mini), we plan to support more models in future.

Database Chat is available for all Premium and Team Premium users, just start chatting, no API keys or other configuration needed.

DbGate uses its AI gateway, which forwards requests to OpenAI API.