Your AI agent,
on your servers
RAG Weaver deploys on-premise in 2 to 4 hours via Docker. Your documents, conversations and data never leave your infrastructure. 100% GDPR compliant, no DevOps team required.
Why choose on-premise deployment?
In our recent projects, 50% started on SaaS and 50% chose on-premise, always for the same reason: data privacy policy.
Data sovereignty
Your documents never leave your servers. Zero transmission to third parties, zero exposure to the US CLOUD Act.
Native GDPR compliance
No complex transfer impact assessment. Your DPO validates in a day. Data stays in Europe.
Private network integration
Accessible only from your VPN or internal network. Ideal for sensitive or regulated documents.
Local LLM option
100% air-gapped with local Llama 3 or Mistral. No external API calls, no data transmitted.
Full control
You manage updates, access and backups according to your internal IT policies.
Predictable cost
A fixed subscription. LLM costs stay with your provider, without markup or per-message fees.
How does the on-premise architecture work?
RAG Weaver on-premise runs on 4 components deployed via Docker Compose on your server.
Vector database (Qdrant)
Stores vector representations of your documents. Lightweight (8 GB RAM for 100,000 docs), open-source, European-built.
RAG pipeline
Automatically chunks, indexes and updates your document sources. Connected to SharePoint, Confluence, PDFs, Word and more.
LLM (external or local)
Mistral/OpenAI API for performance, or local Llama 3 for 100% air-gapped. You choose based on your constraints.
RAG Weaver interface
No-code interface, channel management (widget, Teams, WhatsApp), analytics and business rules. Accessible from your internal network.
Minimum technical requirements
Server
8 CPU · 16 GB RAM · 200 GB SSD (for 50,000 documents)
OS
Ubuntu 22.04 LTS or RHEL 8+
Runtime
Docker 24+ and Docker Compose
Network
Outbound HTTPS access on port 443 (for LLM APIs)
Local LLM (optional)
NVIDIA RTX 4090 or A10 GPU, only for air-gapped mode
Deployment time
2 to 4 hours with RAG Weaver support
Frequently asked questions
What is an on-premise AI agent?
An on-premise conversational AI agent is hosted directly on your company's servers. Your documents, conversations and data never leave your infrastructure, unlike cloud SaaS solutions.
Is an on-premise AI agent GDPR compliant?
Yes, it is the most GDPR-compliant configuration. Data stays on your infrastructure, eliminating risks related to the US CLOUD Act. Your DPO retains full control over data processing.
Do we need a DevOps team to deploy RAG Weaver on-premise?
No. RAG Weaver on-premise deploys via Docker Compose in 2 to 4 hours. Our team supports the initial deployment and trains you on day-to-day management.
What are the minimum server requirements?
A dedicated server or VM with 8 CPU, 16 GB RAM, 200 GB SSD is sufficient to index 50,000 documents. Ubuntu 22.04 LTS or RHEL 8+, Docker 24+ and an outbound HTTPS connection.
Can we use a local LLM on-premise?
Yes. RAG Weaver supports local LLMs (Llama 3, Mistral 7B) via Ollama. This configuration is 100% air-gapped with no external calls. It requires a dedicated GPU (minimum NVIDIA RTX 4090).