AM Pilot | Ops

Health Center

Tenant-wide runtime posture for admin and customer apps, with live incidents, workspace health, and client + backend telemetry. Last sync Pending.

API

Checking Waiting for /api/health

Ops Feed

Checking Waiting for internal health payload

Signals

No data No client or backend signals captured yet

Tenants

— Loading

Attention

— Checking

Incidents 24h

— Waiting

Backend 24h

— Checking

SLO

API availability

Checking API availability SLO.

Waiting for request samples across tenant apps.

SLO

Latency objectives by route

Waiting for route latency samples.

Per-route p95 latency versus objective will appear here.

Attention

Needs action

Waiting for tenant posture data.

Health issues will appear here.

Incidents

Live incidents

Waiting for incident events.

Client and backend failures will appear here.

Control plane

Infrastructure checks

Waiting for dependency probes.

Postgres, storage, slicer, and incident feed health will appear here.

Backend hotspots

API routes under stress

Waiting for backend diagnostics.

Top failing and slow API routes in the last 24 hours will appear here.

Tenant hotspots

Incident concentration

Waiting for incident concentration data.

Tenants with the highest 24h incident volume will appear here.

Tenant matrix

Admin and customer app posture

Tenant	Admin app	Customer app	Activity	Issues

Error Logs

Unified client + backend incident stream with tenant scope, request tracing, and source diagnostics.

Incident stream

— Loading

Client 24h

— Loading

Backend 24h

— Loading

Slow routes 24h

— Loading

Tenant incident log

Recent retained client and backend incidents with request trace diagnostics.

Time	Tenant	App	Event	Request

Customers

Customer workspace matrix with usage, orders, and revenue across all tenants.

Sort by

Customer matrix

Commercial and adoption view by tenant: accounts, orders, revenue, and latest activity.

Customer	Accounts	Orders	Revenue	Activity

Provisioning

Workspaces, domains, first-admin access, and platform fee settings.

Workspaces

Customer list

Provision and manage customer workspaces for the platform.

0 active workspaces

Create

New workspace

Create a tenant workspace and invite the first administrator in one pass.

Tenant slug * Company name

Communication email Brand logo URL

Quote domain Admin domain

Admin email * Admin full name

Platform fee (%)

My account

Password

Update your platform-owner password for the provisioning surface.

Estimator Lab

Keep this surface focused on shadow slicing, component errors, confidence bands, and release gates. It should explain whether the estimator is trustworthy enough to ship.

Shadow mode first Target fidelity: 90%

Shadow labels

Capture ML vs native slice deltas

Per upload, persist the resolved recipe, ML prediction, native result, slicer version, and estimator version.

Error segmentation

Break time and grams by component

Model time, support time, travel time, part grams, and support grams need separate quality visibility.

Release gates

Promote only when regimes improve

Validate by printer, material, layer height, support usage, and job size before any model promotion.

Before charts

Data model that has to exist

Resolved recipe snapshot shared by ML and native slicing
Per-job labels for print time and grams from both engines
Versioning for estimator, native slicer, and config bundle
Offline retraining flow instead of direct online learning

Boundary

What does not belong here

Generic runtime incidents should stay in Health Center
Tenant onboarding should stay in Provisioning
Release actions should wait until shadow validation data is real