Release 1.0.0

[v1.0.0] - 2026-03-27

Added

• AI Agents - Hooks for using genai to perform tasks on VectorStore results.
• Hooks Framework - new framework for hooks to support premade and custom hook development.
• Server Class Features:
  • new methods for instantiating the FastAPI application and/or routing.
  • allows middleware to be used, or the routing to be attached to another FastAPI service.
• Documentation - new QuartoDocs documenting the ClassifAI package and new demo notebooks.
• Partial String matching - reverse search VectorStore method now does optional partial matching.
• Vectoriser Class - More options for instantiating HuggingFace models.

Changed

• Datasets - updated dataset column names for v1.0.0
• Documentation - better docstrings and updated demo notebooks.
• Dataclasses - updated for more intuitive dataframe column naming.
• Server Class Refactor:
  • expanded scope of features.
  • renamed start_api method to run_server.

Fixed

• Server hook data - hook metadata now returned in FastAPI responses.
• Reverse Search results - fixed issue where max_n_results defaulted to None causing errors.

Release 0.2.1

[v0.2.1] - 2026-02-06

Fixed

• Dependency Bug - Removed old dependency code for data processing.

Release 0.2.0

[v0.2.0] - 2026-01-28

Added

• Improved Vectorisers - More ways to use GcpVectoriser.
• Hooks - User defined hook functions for custom input and output VectorStore logic.
• Dataclasses - Data objects for interfacing with core VectorStore methods.
• Documentation and Demo - README updates, Jupyter Notebooks for running server, hooks and more.

v0.1.0 pre-release

[v0.1.0] - 2025-11-04

Added

• Vectoriser Class - Abstract base class, and GCP, HuggingFace, Ollama Vectorisers
• Vector Store Class
• REST API - FastAPI served with Uvicorn
• Documentation and Demo - README and Jupyter Notebook minimal demo with fake dataset.