SpeakesQuery – Splunk-style search over local Parquet, with LLM pipes (github.com)

SpeakesQuery has introduced a powerful local-first data querying tool that allows users to perform Splunk-style searches over local Parquet and SQLite databases using a unique query language called SPQL. This system enables users to ingest data on a schedule, conduct semantic searches, and utilize Large Language Models (LLMs) in data processing—all without the need for cloud dependency or telemetry. By marrying local analytics with AI capabilities, the software aims to enhance data insights while maintaining user privacy and control. Significantly, SpeakesQuery provides a range of built-in features such as automatic semantic ranking of data, customizable LLM integration (supporting both third-party APIs and local models), and a comprehensive set of connectors for various data sources—without requiring API keys for numerous operations. The design ethos emphasizes transparency and non-rent-seeking functionality, which could inspire a new wave of community-driven AI/ML tools. Users can quickly set up the platform with Docker and begin querying data, emphasizing ease of use and accessibility, ultimately enabling analysts to leverage AI on their terms while keeping operational costs predictable and manageable.