Interoperable Europe logo

BASTI

BASTI

Version
0.1.5
Categories
knowledge-management
document-management
workflow-management

On-premises AI assistant with chat, knowledge spaces, RAG, and optional speech features.

Source catalogue
Open Code
0
0
0
0

Vitality

46%

Quick links

Source code repository

Detailed information

Release date
Supported languages
German,
English
Development status
Beta
License
EUPL-1.2
Platforms
web,
linux
Maintenance type
[internal] Internally maintained by the repository owner
Software type
standalone/web
Used by
Bundesanstalt für Straßen- und Verkehrswesen (BASt)
Technical Contacts
Boguslaw Jablkowski
publiccode.yml is valid

Description

BASTI is an on-premises AI assistant for organizations that need to run LLM-based capabilities entirely inside their own infrastructure. It combines a web-based chat interface, document-backed knowledge spaces, retrieval-augmented generation, knowledge graph visualization, document ingestion, and optional speech features in a layered architecture built from a UI, a gateway, and Python services.
The project is designed for data sovereignty, predictable operations, and practical reuse in technical or knowledge-heavy environments. BASTI runs on Linux with Docker Compose, expects OpenAI-compatible inference endpoints, and can be operated without cloud dependencies in restricted networks.

Features

  • WebSocket streaming chat
  • Knowledge spaces with RAG and graph exploration
  • Document ingestion for PDF, DOCX, PPTX, email, and more
  • Gateway-owned authentication and authorization

Intended audience

science-and-technology