Data Talks
    Integration

    Talk to any SQL database

    If it speaks SQL, Data Talks can talk to it. MySQL, SQL Server, MariaDB, SQLite, Snowflake, Redshift — same connector pattern, same conversational UX.

    Get Started

    Most BI tools force you to pick a stack. Data Talks doesn’t care: the same workspace can have a MySQL replica for product analytics, a Snowflake warehouse for marketing, and a SQLite file for a one-off audit — all queryable in the same conversation.

    Each connector handles dialect differences automatically. The agent generates SQL Server T-SQL when the source is SQL Server, MySQL syntax when it’s MySQL, and so on. You don’t need to know the dialect — just ask the question.

    Setup

    Add a SQL source

    1. 1

      Pick the database type from the source picker (MySQL, SQL Server, MariaDB, SQLite, Snowflake, Redshift…).

    2. 2

      Provide host, port, credentials, and database name (or upload the .sqlite file).

    3. 3

      Data Talks introspects schema and prepares the agent.

    4. 4

      Start a conversation scoped to that source — or any combination of sources.

    Examples

    Questions across SQL flavors

    • › [MySQL] What were our top 5 best-selling SKUs last month?
    • › [SQL Server] Show me failed login attempts grouped by IP for the last 24 hours.
    • › [Snowflake] Compute monthly retention cohorts for users acquired in 2025.
    • › [SQLite] Summarize this audit log file by event type and severity.
    Highlights

    Why one tool for all your SQL

    Dialect-aware SQL generation: T-SQL, MySQL, ANSI, Snowflake-specific.

    Connect heterogeneous sources to one workspace and join across conversations.

    Schema introspection works the same across vendors — connect once, ask anything.

    Read-only credentials enforced at the source level.

    Pin charts from any SQL source to the same dashboard.

    100% self-hosted — no vendor sees your connection strings.

    FAQ

    SQL FAQ

    Which databases are supported?

    MySQL, MariaDB, PostgreSQL, SQL Server, Oracle (community), SQLite, Snowflake, Redshift, and Amazon Athena via SQL endpoint.

    Can a single conversation query two databases?

    Yes — the agent can be scoped to multiple sources. Cross-source joins happen in-memory after each query runs against its native source.

    How are credentials stored?

    Encrypted at rest with the workspace key. They never leave your self-hosted instance.

    Does it support stored procedures?

    Yes for execution; the agent prefers pure SELECT generation but can call SPs you expose explicitly as tools.

    Connect your SQL database

    Open source, self-hosted, free. Same conversational UX, every dialect.

    Get Started