REPORTS & AI COST

Know what happened and what it cost

Reports show the whole operation from live rows: conversations, containment, CSAT, responsiveness, queues, cases and the knowledge doing the work. Beside them, a durable usage ledger meters every AI turn, down to the conversation it happened in.

THE FULL PICTURE

The whole operation on one page

Pick a range and every panel follows: how many conversations came in, how many Shinobi resolved without a human, bot reply time with p95, CSAT with the rating count behind it, and the AI QA score from quality reviews. All live rows, nothing sampled, with a CSV export on the same range.

CONTAINMENT

Resolved without a human

Shinobi resolved versus human-handled, with escalations counted in range. The number your automation case rests on, computed from real closes.

RESPONSIVENESS

Handoff to human reply

Median, average and 90th percentile from escalation to the first human reply, plus time to solution from open to closed. Bucketed so the slow tail is visible.

TEAMMATES

Per-person, same bar

Assigned, closed, replies, CSAT, AI QA score, first reply and time to solution for every teammate, in one table.

QUEUES

Waiting, claimed, abandoned

What happened after every handoff, per queue: claimed within two minutes, claimed slower, never claimed, and who is waiting right now.

CASES

Reasons and pipeline

Cases opened in range grouped by reason, and the open, pending and solved pipeline beside them.

VOLUME

Per day and per hour

Conversations per day with the escalated share marked inside each bar, and an hourly distribution in your workspace timezone for staffing.

KNOWLEDGE GAPS

See what Shinobi could not answer

The report lists what Shinobi handed off, question by question, and ranks the articles it actually cited in replies, so you see both the knowledge doing the work and the gaps in it. Self-Healing reads the same evidence and turns the gaps into drafted fixes, one approval from live.

100%of rows are live aggregates from your workspace, never samples
1range picker drives every panel and the CSV export together
Everycited article is counted, so you know which knowledge earns its place
THE METER

Every AI turn writes a durable usage event

The meter is not an estimate. Each AI call appends an immutable usage event with its tokens, surface, provider and conversation, and every number on the usage page aggregates those events live: tokens per day, the by-surface ledger, the provider split and average tokens per conversation. Export the CSV and you hold the raw ledger for reconciliation.

PER CONVERSATION · LIVE
Spend you can point at

Live today: usage events carry their conversation, so you see conversations with AI activity, average tokens per conversation, and the exact rows behind any bill question.

BY SURFACE · LIVE
Where the tokens go

Live today: a ledger of calls and tokens per surface and provider, with proportional bars, so the expensive corner of the product has nowhere to hide.

BYO KEYS · LIVE
Your keys, $0 AI line

Live today: bring your own Anthropic, OpenAI, Gemini or Grok key and those calls bill through your provider account. The metered line on your Shinobi bill drops to $0.

BILLING PERIOD · LIVE
An honest month-to-date

Live today: Usage & Metering itemizes your spend line by line, plan, metered AI actions, apps and invoices, every dollar the meter times a published rate.

Want answers instead of panels? Scout takes open questions across the same conversations and replies with the transcripts linked.

QUICK DEMO

See it work

SIMULATION
Outcomes and the raw AI meter, live: containment, CSAT, and spend per conversation.Open the full demo →