The stack · open source
On these shoulders.
Everything behind this demo is open source and self-hostable — 34 building blocks, not a single corporate subscription, no lock-in. If you want to leave, you take your data with you.
How it all fits together
Data flows from the left (machines, files, web) via gateway & streaming buffer into the backbone — time-series DB, data lake, ERP — and on into BI, twin & AI up to the public face. All secured underneath by backups & offsite.
Web & front
ERP & operations
Odoo ERP: order→invoice LGPL-3 see live ↗ ERP ⌕ 
PostgreSQL database PostgreSQL ERP ⌕ Data, BI & data science
Metabase BI dashboards AGPL-3 see live ↗ BI & analytics ⌕ 
Grafana monitoring AGPL-3 see live ↗ BI & analytics ⌕ 
JupyterLite notebooks in browser BSD-3 see live ↗ Material analysis ⌕ 
Meilisearch full-text search MIT see live → AI search ⌕ Plant, twin & visualisation
MO model-viewer 3D in browser Apache-2.0 see live →
Mermaid diagrams/plans MIT see live → AI & search
Mistral language model (answers) open weights / API see live → Artificial intelligence ⌕ 
n8n automation fair-code Collaboration
Nextcloud files & office AGPL-3 see live → Knowledge & kiosk ⌕ 
Collabora office online MPL-2.0 see live → Knowledge & kiosk ⌕ 
Matrix / Element chat & video (E2EE) Apache-2.0 / AGPL-3 see live → Knowledge & kiosk ⌕ XW XWiki wiki & knowledge base LGPL-2.1 see live →
Infrastructure & sovereignty
Docker containers Apache-2.0 
Proxmox VE virtualisation AGPL-3 see live → Backup & restore ⌕ NE NetBird mesh VPN BSD-3
RustDesk remote support – self-hosted AGPL-3 Resilience & backup ⌕ Operations, security & audit
WA Wazuh SIEM: intrusion detection & host audit GPL-2.0 see live →
Zabbix infra metrics & alerts AGPL-3 see live → Compliance & security ⌕ 
Graylog central log aggregation SSPL-1.0 see live → Compliance & security ⌕ Grafana live cockpits & alerts AGPL-3 see live ↗ BI & analytics ⌕ Data backbone & resilience
TimescaleDB Time-series DB: millions of sensor points, analysed in a flash — feeds historian & twin. Apache-2.0 see live → Historian ⌕ OB Object-Store / Data-Lake (S3) The catch-all store: raw files, images, reports, measurement artefacts — cheap, practically endless. Apache-2.0 see live →
Kafka Streaming buffer: decouples ingest spikes — only needed at high throughput (see scaling). Apache-2.0 see live → Historian ⌕ Backups & offsite (PBS) Snapshots, offsite copy, restore tests — so nothing is ever lost. Visible on the status board. AGPL-3 see live → Backup & restore ⌕ Search, archive & AI bridge
Full-text search (all company data) One search across documents, mails, notes, wiki and machine data — typo-tolerant and meaning-aware. MIT see live → AI search ⌕ EM Email archive Tamper-evident, searchable email archive — everything stays findable, even after years. AGPL-3 see live →
MC MCP search server (AI bridge) Model Context Protocol: gives the AI direct, permission-checked access to search — it answers with real company data. open see live →
The value isn't in locking you in — it's in the assembly: proven open parts, cleanly integrated, run in your own house.