Control access
Central login, multi-factor authentication, roles and permissions control who can access which knowledge assets.
ImplementedAuthentik · OIDC · SSO · MFA
Technology at a glance
Open architecture.
Interchangeable components.
praivate separates access, application, knowledge processing, as well as data and operations. The components are documented and can be selected or replaced to suit the infrastructure, protection requirements and use case.
← Back to the landing page01Architecture and implemented software
Four layers. One controllable architecture.
praivate connects clearly separated components through defined responsibilities and interfaces. Models, storage and the operating environment can be configured to suit protection requirements, hardware and the use case.
Manage workflows
The web application combines document management, search and source-based answers in a shared interface.
ImplementedFastAPI · Uvicorn · Jinja2 · HTMX · Server-Sent Events
Find relevant information
Term-based and semantic search, embeddings and reranking identify the passages that are relevant to a query.
ImplementedQdrant/BM25 · bge-m3 · ms-marco-MultiBERT-L-12
Operate models. Control data.
Models, persistent data, session state, reverse proxy and automated setup remain clearly separated and can be configured to suit the operating environment.
ImplementedOllama · configurable approved models · PostgreSQL · Redis · Caddy · Ansible
Models and hardwareModel files are not a fixed part of praivate. Language, embedding and reranking models are selected to suit the hardware, language, licence and use case. Openly licensed models can also be operated entirely within your own environment.
LimitsLocal operation and free software create control, but guarantee neither correct answers nor automatic compliance with data protection requirements. Data quality, configuration and organisational processes remain decisive.