Self-Hosted Support Desk in Docker: When Customer Data Cannot Leave the Building

For some organizations, "the cloud is fine" is not a sentence anyone can say out loud. Government agencies, banks, healthcare providers, subsidiaries under strict group policies, trading and export firms with data residency obligations, and any team whose network is segmented or air-gapped - they need the same support software as everyone else, on infrastructure they control. Self-hosting a modern support desk is no longer a science project. Here is what it actually involves.


WHAT SELF-HOSTED MEANS HERE

The same software as the cloud version, running on your own LAN or in your own cloud account. Data never leaves your network; the vendor has no access to it. You provide the AI model endpoint and the storage - which is exactly why there are no AI fees and no storage fees on this track. Installation, software updates and technical support are included in the license.

WHAT IT COSTS

- $15 per seat per month, billed annually, 10 seats minimum - $1,800 per year to start

- The same price on your own hardware or in your own cloud

- No AI fees, no storage fees: you supply the model and the disks

- Installation service, updates and support included

- A 60-day trial of the full product before any payment; licenses are final sale afterwards, so the trial is where evaluation happens

WHAT IT REQUIRES


- A 64-bit Linux VM with Docker Engine 24 or newer, and a domain with TLS (or internal DNS)

- Sizing by team: up to ~10 users, 2 vCPU / 4 GB RAM / 40 GB disk; up to ~50, 4 vCPU / 8 GB / 100 GB; up to ~200, 8 vCPU / 16 GB / 250 GB

- Disk grows with attachments only - text conversations are negligible

- A license file, bound to your domain and capacity tier, loaded at first startup

THE DEPLOYMENT, IN FOUR STEPS

1. Place the license file and environment configuration in the deployment directory and set your domain and timezone.

2. Bring the stack up with Docker Compose - a single image set with no external dependencies.

3. Confirm health with a simple HTTPS check against your domain.

4. Create the admin account, connect channels and invite users - identical to the cloud setup.

AI IN A SELF-HOSTED DESK

Three options, all supported: bring your own API key against an OpenAI-compatible endpoint; point the assistant at a model served inside your network; or run fully air-gapped with AI enabled only when an internal endpoint is available. The core software runs completely disconnected. Credit accounting works the same as cloud, but no credit packs are sold on this track - the model is yours.

BACKUPS AND UPDATES

Backups are scripted and encrypted: a scheduled job writes an archive containing settings, conversations, documents and contact data, with a matching restore command - and the documented routine is to run it daily via cron. Upgrades are backward compatible within the supported major version: pull the new images, bring the stack back up. With a license, updates are part of the plan and upgrade paths are tested before you apply them.

WHO SHOULD STAY ON CLOUD

If your constraint is not data residency, the cloud track is free for three seats and takes 15 seconds to start - most teams should begin there and move later if requirements change. The export/import path between tracks is supported, so that decision is reversible.

The self-hosted overview, including the requirements form and pricing, is at https://onsitesai.com/self-host.html, and the full deployment guide - requirements, license, compose file, AI endpoints, backups and upgrades - is at https://onsitesai.com/docs/self-host-deployment.html.

#SelfHosted #Docker #HelpDesk #DataResidency #Compliance

Comments

Popular posts from this blog

Why Employer Need Monitor Software?

The Benefit of Using Computer Monitoring Software