Overview
Platform health and key metrics at a glance.
Active Tenants
–
Active Users
–
Total Clients
–
across all tenants
Inquiries
–
all time
Emails Processed
–
by sentinel AI
System Health
Platform DB
Checking…
Email Sentinel
Checking…
AI Provider
Checking…
Background Jobs
Checking…
Storage
Total Data
–
Tenant DBs
–
Disk Free
–
Largest tenant databases
| Tenant | Size |
|---|---|
| Loading… | |
Tenant Activity Snapshot
| Tenant | Product | Users | Clients | Inquiries | Emails | Status |
|---|
Tenants
Companies and subscriptions on this platform.
Add New Tenant
Users
All platform accounts — agents, admins, and superadmins.
Add New User
| Name | Role | Tenant | Joined | Last Login | Transcript Quota | Status | Actions |
|---|
Email Sentinel
Per-tenant AI email monitor. Configure IMAP credentials in the Tenant manager.
All Tenant Sentinels
| Tenant | Product | Status | Processed | Last Poll | Error | Actions |
|---|---|---|---|---|---|---|
| Loading… | ||||||
Email Summaries
| Received | Sender | Subject | Category | Priority | Summary | Status |
|---|
LLM Usage
Token usage and estimated cost across every AI feature — sentinel, proposals, chat, client profiles.
Spend Today
–
Spend This Month
–
Calls (period)
–
Spending Cap
Currently Active
–
This is what's actually enforced right now — the fields below are just for editing
and don't take effect until you click Save.
Set either to 0 for no limit. Once monthly or daily spend reaches its
limit, AI features automatically switch to the free local model (Ollama) instead of
the paid one — nothing stops working, it just stops costing money until the period
resets or you raise the cap.
Spend Over Time
| Purpose | Calls | Cost | Success Rate | Avg Latency |
|---|---|---|---|---|
| Loading… | ||||
By Customer
| Tenant | Calls | Cost |
|---|---|---|
| Loading… | ||
AI Agent QA Pipeline — Recent Evaluations
A separate Judge model (always Claude/Anthropic) grading a Worker's real output
against an expected answer — see TODO.md's "AI Agent QA Pipeline" section.
| When | Worker | Reference | Score | Judge | Rationale |
|---|---|---|---|---|---|
| Loading… | |||||
Agent Dashboard
AI Agent QA Pipeline, Phase 4 — real Actor accounts that log in and drive the app on
their own. Every run below costs real money (Actor's own LLM call + the Worker feature
it exercises + the Judge grading it) — see TODO.md.
Actor accounts
–
Runs so far
–
Total spent (Actor runs)
–
Travel personas
| Persona | Type | Last run | Last score | Last cost | |
|---|---|---|---|---|---|
| Loading… | |||||
Legal personas
| Persona | Type | Last run | Last score | Last cost | |
|---|---|---|---|---|---|
| Loading… | |||||
Legal RAG Effectiveness
How often the jurisdiction-law RAG actually retrieves articles, aggregated from the
per-chat logs — counts only, never message content. Moved here from the lawyer's own
chat UI 2026-07-27 (the per-message article chips were removed as noise for the lawyer;
the signal is more useful here, in aggregate).
Chat queries (30d)
–
Retrieved articles
–
Avg articles / legal query
–
Direct vs semantic
–
Business Metrics
Internal-only, never customer-facing — real numbers from real traffic, measured
from
llm_usage_logs and real pipeline timestamps, never estimated. See
TODO.md's "Business Metrics / KPI Dashboard" section.
Avg. inquiry → ready proposal
–
Median
–
–
Time saved (vs. your estimate)
–
Manual time (min):
Per-feature real cost, volume & latency
Real product features only — test/diagnostic traffic and the AI Agent QA Pipeline's
own calls are excluded. "Your estimate" is a manual input (how long this takes a
person by hand) — "Time saved" is only ever as good as that estimate, never a
measured number itself.
| Feature | Calls | Avg. latency | Real cost | Your estimate (min) | Time saved |
|---|---|---|---|---|---|
| Loading… | |||||
Failures
Every non-2xx API response across both products — management-only, never seen by travel/legal users.
Total (period)
–
Top Failing Paths
| Path | Status | Count |
|---|---|---|
| Loading… | ||
Recent Failures
| When | Method | Path | Status | Product | Tenant | Error | ms |
|---|---|---|---|---|---|---|---|
| Loading… | |||||||
Usage
Which pages get used, how often, and roughly how long people stay — management-only, never seen by travel/legal users.
Events (period)
–
Sessions (period)
–
Pages — views & avg. time spent
| Page | Views | Avg. dwell | Total dwell |
|---|---|---|---|
| Loading… | |||
Top Clicks
Only buttons explicitly marked for tracking show up here — add
data-track="some_label"
to any button in travel/app or legal/app's index.html to start tracking it.
| Label | Count |
|---|---|
| Loading… | |
Audit Log
Who changed what, and when — every role change, activation toggle, tenant reassignment, password reset, and account creation.
Recent Actions
| When | Actor | Action | Target | Detail |
|---|---|---|---|---|
| Loading… | ||||
Setup
One-time platform initialisation and maintenance utilities.
Seed Base Products
Creates the BarkerSuite-Travel and BarkerSuite-Legal products in the platform database if they don't exist yet. Safe to run multiple times.
Platform Info
Platform DB data/platform.db
Tenant DBs data/tenants/{id}.db
API Base /api/admin/…
Auth JWT Bearer token