Sovereign AI
Optional private AI platform for EchoDesk and GLPI
AI for GLPI, running on your hardware, under your control. No per-token bills. No data leaving your network.
What is Sovereign AI?
Sovereign AI is a self-hosted, on-premises private AI platform. Members run open-weight large language models inside their own infrastructure — with no dependence on third-party cloud AI providers.
The stack is the open-source combination of Open WebUI (chat, RAG, assistants, multi-user RBAC), Ollama (local model runtime with an OpenAI-compatible API), and curated open-weight models (Llama, Qwen, Mistral, Phi, Gemma and others). Existing GLPI AI plugins connect to a local endpoint instead of a cloud API key.
Why Members Choose It
Data sovereignty
Ticket content and user data stay on your network. Optional fully air-gapped mode for the highest sensitivity estates.
Cost certainty
No per-token or per-seat cloud AI invoices. Models are free for commercial use; you fund the hardware you already control.
No lock-in
OpenAI-compatible API surface. Swap models without new contracts or re-architecting your plugins.
GLPI-first
Drop-in endpoint for Echo-9 AI plugins — ticket drafting, KB search, categorisation and summaries stay on-prem.
Web chat and RAG
Open WebUI gives multi-user chat, document upload, grounded answers with citations, and named assistants.
RBAC and audit
Invite-only sign-up, roles, optional OIDC/LDAP, usage analytics and configurable retention.
How It Fits EchoDesk
Plugin flow
GilpAI, ProblemAI and other OpenAI-style GLPI plugins call your local /v1/chat/completions endpoint. Ticket context stays in GLPI.
Web chat flow
Staff use Open WebUI over HTTPS for chat, document RAG and assistant templates.
What we deliver
Hardened reference deployment, automation, GLPI integration guidance, monitoring and support — not another SaaS AI tenancy.
What it is not
Not a cloud AI service. Not a fine-tuning platform on day one. Not a GLPI plugin itself — it is the AI backend those plugins consume.