Everything ships as Docker containers on your infrastructure or private cloud — the application, the databases, and the AI model itself. No prompts, no readings, nothing leaves the network.
One server, one Docker Compose file, the whole platform.
Security your IT/OT team can verify — because it all runs where they can see it.
Stored connection credentials are encrypted with AES-256 (Fernet). Secrets live on your server, protected at rest.
Token-based authentication with role-based permissions — operators, analysts and admins each see exactly what they should.
Your users log in with the directory you already run. No separate identity silo to manage.
Who connected what, who asked what, who exported what — a trail your compliance team can actually query.
Fine-grained, row-level permissions in the database — multi-site teams see their site, not everyone else’s.
The AI chatbot can query your data but can never modify it — a hard guard enforces read-only SQL, always.
Nerqon deploys with a single Docker Compose command — from bare server to running product in under 15 minutes, on hardware you probably already have.
A 14-billion-parameter language model plus local vector search run as containers next to your data. Prompts and answers never cross your firewall.
There is no OpenAI key, no cloud AI account, no third-party processor for your energy data — nothing to rotate, audit or worry about.
Ask a thousand questions a day. The marginal cost of an answer is your own hardware — not a metered API bill.
Every AI answer shows the exact SQL it ran. Grounded in your database — not the internet — and enforced read-only.
They’ll ask the hard questions about deployment, auth and data flow. We like those questions.