Private AI · On-Premises or Hosted · Swiss-made

    KeepUrAi

    Your company's own AI — on your server, Swiss-made, no per-user fees.

    Why KeepUrAi

    The middle ground

    Cheap AI tools are neither private nor tailored to how you work. A custom build from scratch can cost tens of thousands of francs. KeepUrAi sits in between: private AI software, ready to run on your infrastructure or hosted in Switzerland — without the custom-build price tag.

    Made in Zürich by INSS — founded by an engineer with 25+ years in Swiss banking IT.

    Features

    Built for private AI, on your terms

    Everything you need to deploy, manage, and scale KeepUrAi within your own infrastructure.

    Complete Data Sovereignty

    On the on-prem editions, your data never leaves your infrastructure — full control over where models run and how data is processed.

    On-Premises Deployment

    Deploy on bare metal, virtual machines, or Kubernetes. Air-gapped deployment is possible with the on-prem editions.

    GPU-Optimized Performance

    Automatic hardware detection and optimized inference on your existing GPUs.

    Security by Design

    Role-based access control, SSO via OAuth 2.0 / OpenID Connect, and audit logging built into every layer.

    Model Flexibility

    Run open models via Ollama, such as Gemma and Qwen. Switch or upgrade models as your needs evolve — no vendor lock-in.

    Compare Solutions

    Public vs Private AI Hub

    Understand the key differences and discover why companies choose private AI infrastructure for their sensitive workloads.

    Public AI Hub

    Cloud-based solutions

    Private AI Hub

    On-premise deployment

    Data Location
    Cloud servers (3rd party)
    Your infrastructure
    Data Privacy
    Data leaves your network
    Data never leaves
    Latency
    Network dependent
    Local, ultra-low latency
    Compliance
    Limited control
    Full regulatory control
    Customization
    Vendor restrictions
    Complete flexibility
    Cost Model
    Pay-per-use (unpredictable)
    Fixed infrastructure cost
    Availability
    Vendor dependent
    You control uptime
    Integration
    API limitations
    Deep system integration
    Keep your data secure with a Private AI Hub
    Security

    Security by Design

    Security isn't an afterthought — it's built into every layer of KeepUrAi.

    • Role-based access control with fine-grained permissions
    • SSO via OAuth 2.0 / OpenID Connect (e.g. Keycloak, Microsoft Entra ID, Google, GitHub)
    • Audit logging for access and changes
    • Network isolation and private subnet deployment

    Built for compliance

    Designed with these requirements in mind

    GDPR

    Built for its requirements

    revDSG

    Swiss data protection law

    Air-gapped

    Possible with on-prem editions

    Audit log

    Included

    Zero Trust Architecture: Every request is authenticated and authorized, regardless of network location.

    INSS currently holds no security certificate. We only say "certified" once a certificate exists.

    Deployment

    On-site installation

    Simple deployment process with powerful defaults. No complex configuration required.

    01

    Kubernetes

    Get the AI you need in your Kubernetes infrastructure.

    02

    On-prem Linux server

    Install all AI components on your own server.

    03

    Docker

    Run all AI components in Docker containers on your own infrastructure.

    04

    AI workstation

    Get your own AI workstation.

    Hosted edition

    KeepUrAi Hosted — try it now

    The hosted edition runs at qcg-ai.ch — a Swiss service operated by INSS. No installation, no server to maintain.

    Data flow, honestly: GPU inference for chat and images runs at external providers (Ollama Cloud, Comfy Cloud). This is not covered by the on-prem "your data never leaves your servers" claim — for that, choose an on-prem edition.

    Editions

    Three tiers. All on your infrastructure.

    Core

    Everything you need to get started with private AI.

    • On-premises AI inference with Ollama
    • RAG pipeline with pgvector semantic search
    • Document ingestion with Apache Tika
    • REST API with JWT authentication
    • Single-node deployment
    See prices in the shop

    Professional

    Core plus multi-user and single sign-on.

    • Everything in Core
    • Multi-user RBAC with fine-grained permissions
    • SSO: OAuth 2.0 / OpenID Connect (e.g. Keycloak, Microsoft Entra ID, Google, GitHub)
    • Image generation with ComfyUI
    • Priority support
    See prices in the shop
    Enterprise

    Enterprise

    Full feature set with custom SLA and priority support.

    • Everything in Professional
    • Kubernetes Helm charts
    • Custom integrations on request
    • Custom SLA options
    • On-site deployment assistance
    See prices in the shop

    Available as: Debian package · Docker Compose · Kubernetes

    FAQ

    Frequently Asked Questions

    Everything you need to know about deploying KeepUrAi on your own infrastructure, or trying it hosted.

    Download

    Try the Mobile App

    Download the KeepUrAi mobile app for Android and connect to your AI server from your pocket.

    KeepUrAi Mobile (Android)

    Early-access build. Requires a KeepUrAi server on your network.

    Download APK
    Demo invite QR code for the KeepUrAi mobile app

    Scan this QR code with the app to connect to the demo server.