Technology Overview#

The Knowlestry Platform is a modular system built around AI and web technologies. Its architecture is designed for flexibility, scalability, and seamless integration of various data sources and analysis pipelines.

AI and Knowledge Engine#

At the core of the platform lies an intelligent AI agent that uses a Retrieval-Augmented Generation (RAG) architecture. This enables the system to combine knowledge retrieval from structured sources with generative capabilities from large language models.

The platform is compatible with multiple LLMs and vector databases. By default, it integrates:

  • LLM: Azure OpenAI (GPT-4o by default)

  • Vector Database: Internally hosted ChromaDB

These defaults ensure powerful natural language processing and efficient semantic search within the organization’s domain knowledge.

Technology Stack#

The platform is implemented using modern technologies across its various layers:

  • Frontend: Built using React, providing a responsive and user-friendly interface for workspace navigation, knowledge interaction, and analytics.

  • Backend: Developed in Python, orchestrating all internal logic, AI components, and task management.

  • REST API: All interactions between frontend and backend are managed through a REST API. The specification of this API will be published soon for developers who wish to build custom integrations or automate workflows.

  • Data Analysis: The platform performs in-depth analysis of machine and time-series data using established Python libraries such as NumPy, SciPy, Pandas, and Matplotlib.

Storage Infrastructure#

The internal storage system is based on ownCloud, a flexible and secure file storage solution. It provides efficient file handling for each workspace, enabling uploads, structured organization, and persistent storage of analysis artifacts.

While ownCloud offers strong integration with the platform, there are some limitations to consider:

  • File access is scoped to individual workspaces and must be explicitly shared.

  • There is a size limit for file uploads, which depends on the organization’s storage quota.

  • Simultaneous access to the same file from multiple background tasks may lead to temporary locks.