Infrastructure direction

Where should the work live?

Hosting is a set of separate decisions: compute, model files, storage, networking, secrets, backups, and the interface people use. Make each layer explicit before moving a local experiment online.

Local

Private workstation

Best for sensitive inputs, frequent iteration, and predictable ownership. Limited by hardware, electricity, uptime, and local networking.

Hosted app

Managed service

Fastest route to a usable interface. Check data retention, export, model choice, rate limits, and whether the service supports your real workflow.

Cloud VM

Hetzner-style server

Good for a coordinator, API, database, dashboard, or always-on service. You own more operations: updates, firewall, backups, access, and monitoring.

Rented GPU

RunPod or marketplace

Useful for bursts of image or video generation. Treat storage, idle time, image persistence, and deletion behavior as part of the cost.

A simple architecture map

LayerQuestionFailure to prevent
InterfaceWho can reach it, and what can they do?An accidentally public admin panel
CoordinatorWhat receives jobs and reports status?Lost jobs or invisible retries
Model runtimeWhich model runs, with what limits?Unexpected quality, license, or GPU cost
StorageWhat persists, for how long, and who can delete it?Secrets or user media in an unprotected volume
OperationsHow are updates, backups, logs, and alerts handled?A service that works until the first restart

How to compare providers

MeasureRecord startup time, generation time, idle time, storage use, egress, and the cost of one representative job.
ProtectUse least-privilege credentials, a firewall, private storage by default, and a clear deletion path. Never place a secret in a page, prompt, or public repository.
RecoverTest a restart and restore before calling the service reliable. A backup that has never been restored is only a hope.
BoundSet spending limits, timeouts, concurrency limits, and a kill switch before connecting automation to paid compute.

Practical starting point

Keep the handbook and small coordinator local. Put only the expensive, bursty generation step on a rented GPU. Move an always-on public service to a managed or self-managed host only after the local workflow has a measured baseline.

Review hosting sources →