Ops-console entry

Auditable support operations before the agent exists

A static FieldOps Copilot shell for ticket triage, run timelines, source evidence, approvals, tool calls, audit trails, evals, and observability.

Open tickets
7 need review
24
Approval waits
3 high risk
6
Automation confidence
MVP demo target: 80%
78%
Audited actions
0 duplicate side effects
148
TanStack Start
Dashboard shell, static mock screens, future BFF server functions.
Railway FastAPI
Future OpenAPI contract, ticket/run/approval APIs, no code added now.
Postgres and Redis
Future source of truth, queue coordination, checkpoints, and locks.
LangGraph and LangSmith
Future agent workflow, traces, evals, and run explainability.
Run timeline
Mock LangGraph node state with latency, cost, confidence, and approval wait markers.
Ticket normalized
Complete

Tenant, customer, severity, product area, request_id, and idempotency scope were attached to the run packet.

Latency
212 ms
Cost
$0.000
Confidence
100%
Classified as deploy regression
Complete

Matched Field API, production, P1 severity, and likely schema compatibility risk.

Latency
1.4 s
Cost
$0.012
Confidence
88%
Retrieved evidence
Complete

Selected a runbook, deployment changelog, service status snapshot, and one similar resolved ticket.

Latency
2.8 s
Cost
$0.006
Confidence
84%
Risk and action plan
running

Preparing a customer update plus a Jira escalation draft. External side effects are blocked until approval.

Latency
running
Cost
$0.014
Confidence
82%
Active work
High-signal ticket records with run and trace identifiers exposed.
TCK-1048P1

Production API calls started failing after yesterday's deploy

Needs review
TCK-1047P2

Technician app cannot sync attachments over LTE

Running
TCK-1042P1

Portal shows stale outage status for field coordinators

Escalated
Trace handoff
Future LangSmith links should be persisted with run IDs and exposed here.
LangSmith trace placeholder
Mock trace ID trace-demo-3f82 is displayed without linking to a real external workspace.