AI & Automation

Devix Nexus

A private, on-prem AI workspace for local chat, knowledge bases, and admin RBAC — powered by Ollama with zero external telemetry.

Client: Devix — internal product
Completed: March 2025
Technologies: SvelteKitFastAPIPythonOllamaChromaDBSQLiteDocker

Devix Nexus gives enterprises a ChatGPT-like experience that runs entirely on their own infrastructure. Teams chat with local models via Ollama, build knowledge bases with RAG, and administrators manage users and access through RBAC — all offline-capable and free of external telemetry, so sensitive data never leaves the organization.

Ollama (local)

Inference

ChromaDB

Retrieval

None

Telemetry

Docker

Deploy

The challenge

Enterprises wanted the productivity of conversational AI without sending prompts, documents, or usage data to third-party clouds — and with proper admin controls for internal rollout.

The solution

A self-hosted AI platform: SvelteKit frontend, FastAPI backend, Ollama for local inference, ChromaDB for retrieval-augmented knowledge bases, and admin RBAC for user management — deployable via Docker with no outbound telemetry.

Architecture

SvelteKit UI over a FastAPI backend, Ollama for local LLM inference, ChromaDB + SQLite for knowledge and persistence, and Docker packaging for offline enterprise deployment.

Key features

Local model chat via Ollama
RAG knowledge bases with ChromaDB
Admin RBAC and user management
Offline-capable, no external telemetry
Docker deployment
Sensitive data stays on-prem

Have a similar problem to solve?

Tell us what you're building — we'll tell you honestly whether we're the right team for it.

Start a conversation