What You Get with KeepUrAi
    Features
    Overview
    SME AI
    Getting Started

    What You Get with KeepUrAi

    March 28, 2026Oliver Glas

    What You Get with KeepUrAi

    KeepUrAi is a self-hosted, private AI platform built for organizations that need full control over their data, models, and users. Once deployed, your team gets a complete AI workspace — no subscriptions to cloud providers, no surprises, and, in the on-premise editions, no data leaving your servers.

    Here is everything that comes with KeepUrAi.

    Feature Overview

    • AI Chat — three modes in one Ask general questions, query your own documents, or switch to a developer-focused mode. All three run on your locally hosted language model.

    • Document management and RAG Upload PDFs, Word files, or plain text. KeepUrAi indexes them with semantic search (pgvector) so the AI can answer questions grounded in your actual content.

    • Image generation Generate images from text prompts via an integrated ComfyUI backend. Images are stored server-side with shareable links.

    • Mobile client for Android A full-featured Android app with streaming chat, image generation, file upload, and QR-code pairing — no VPN configuration required for users.

    • Agent task queue Submit tasks from the web or mobile interface. External AI agents (e.g. Claude Code via MCP) pick them up, process them, and return answers — all tracked in real time.

    • Role-based access control Three roles (User, Power User, Admin) backed by 19 granular permissions. Admins manage users and permissions directly from the interface.

    • SSO and OAuth2/OIDC login Connect your existing identity provider — Keycloak, Google, Azure AD, or GitHub. Standard JWT authentication is also available for self-contained deployments.

    • Moderation service Can be enabled in log-only mode or enforce mode to detect and block abusive content before it reaches the model.

    • API gateway with MCP support CAPI sits in front of your backend services and can automatically expose REST APIs as MCP tools for LLM agents — zero code changes on the backend.

    • Binary mobile gateway A lightweight Go gateway handles all mobile traffic over an encrypted channel (WireGuard/Tailscale). Includes per-token rate limiting and a worker pool with priority lanes.

    • Debian and RPM packages Every component ships as a standard Linux package with systemd units. Install, update, and manage KeepUrAi like any other server software.

    • Multi-model support via Ollama Switch between language models (LLaMA, Mistral, Qwen, and others) without redeploying. Embedding models are configured separately for retrieval tasks.

    • Whisper speech-to-text An optional whisper.cpp server integration enables voice input across clients that support it.

    • Audit-ready deployment In the on-premise editions, all data stays on your infrastructure — no telemetry, no external calls, no third-party AI providers involved in your queries. The hosted edition (qcg-ai.ch) is a Swiss-operated service that uses external GPU providers (Ollama Cloud, Comfy Cloud) for chat and image inference.

    Who It Is For

    KeepUrAi is designed for teams and organizations that:

    • Handle sensitive or regulated data (healthcare, finance, legal, public sector)
    • Want AI capabilities without sending data to external providers
    • Need a single platform that works across web, mobile, and automated agent workflows
    • Prefer standard Linux deployment over managed SaaS

    Ready to explore? Check the documentation, try the hosted edition at qcg-ai.ch, or contact us to discuss your own deployment.