Summarize system status
Explain service health, backups, storage conditions, recent alerts, and notable events in plain language without sending the owner through multiple dashboards.
Most AI sends your secrets to a big server to train a corporate model. Ours stays in your house. It knows your system. It obeys only you.
The operator is there to reduce dashboard sprawl, make the stack legible, and help with approved actions across configured services. It should feel like a practical front door to the environment, not a generic chatbot bolted onto it.
Explain service health, backups, storage conditions, recent alerts, and notable events in plain language without sending the owner through multiple dashboards.
Combine Frigate detections, Home Assistant alerts, and recent device changes into one clear summary with clips or follow-up context where available.
Provide a conversational front door into files, photos, monitoring, smart-home state, backups, calendar, contacts, tasks, and other configured services.
Trigger bounded actions such as report generation, diagnostics, file sharing, thermostat changes, scheduling helpers, onboarding checklists, and configured workflows where appropriate, with explicit confirmation where needed.
Where configured, the operator can help with files, photos, email, calendar, contacts, tasks, smart-home control, camera activity, backups, and general system questions. The goal is not artificial autonomy. The goal is one reliable operator over the private stack, with one place to ask what happened, what matters, and what needs attention.
These are the kinds of daily workflows that make the operator matter. They are customer-facing, bounded, and tied to the services already deployed in the stack. Each one works by combining signals from multiple parts of the system into one useful answer or one approved action path.
Answer “What happened while I was gone?” by checking Frigate events, motion or door activity, network status, and server alerts, then returning unusual events, device changes, and clips worth reviewing.
Summarize important emails, calendar items, top tasks, server health, backup status, Home Assistant alerts, and recent camera events into one morning brief with the few actions worth attention.
Answer “How private is my setup right now?” by reviewing VPN status, public shares, DNS and ad-blocking, remote exposure, backup encryption, cloud integrations still active, and unknown devices.
Answer “Would I be okay if the server died today?” by checking RAID health, last local and off-site backups, restore test status, critical services covered, and missing backup paths, then giving a direct confidence answer.
Answer “Are our photos safe?” by checking Immich upload status, which devices are synced, storage health, backup coverage, and off-site copy status where enabled.
Answer “Summarize camera activity from yesterday” by filtering Frigate events into meaningful detections, object types, times, and clips, including “nothing unusual” when that is the honest answer.
Handle “Set up my new phone” with a step-by-step checklist for VPN, Nextcloud, Immich, Home Assistant, and sync verification so the new device lands cleanly in the private stack.
The operator should be usable in ordinary language. The value comes from pulling together the right systems behind the scenes and returning one clear answer or one bounded next step.
The operator does not invent authority, improvise unsupported changes, or claim powers it does not have.
Actions are reported with explicit status, scope, and outcome so the owner knows what did and did not happen.
The baseline design keeps critical operator functionality local whenever practical instead of routing trust back through the cloud.
If a workflow could affect security, access, data, locks, or other sensitive behavior, it should require explicit approval.
8-16 GB
VRAM target
Core system summaries, basic workflows, and lightweight local assistance.
16-24 GB
VRAM target
Stronger responsiveness, richer context handling, and more capable local workflows.
24-48 GB+
VRAM target
Advanced private operator behavior for larger deployments and more demanding local models.
64-100 GB+
VRAM target
Heavy-duty AI that works for you, not for a subscription.
Each tier changes the local compute envelope and workflow capacity, but the governing principles stay the same: local-first operation, approved workflows, truthful reporting, and one clear operator experience over the stack.
Entry local model capacity for conversational status, summaries, and essential operator tasks.
Balanced performance for clients who want broader day-to-day usefulness from the operator.
Higher-capability local stack for complex environments or more capable private operator behavior.
Dedicated premium infrastructure for heavier private AI workloads, coding agents, and advanced assistance.