Live demo · sample data. A fictional engineering org so you can see the spend-review loop end to end. Review and scan actions are disabled here.

Org overview

AI coding spend review

A guided demo on sample data: what shipped, what is still in flight, and what did not tie back to a PR.

Captured AI spend

$980

63.5M tokens · list-price estimate

Merged PR spend

$542

5 PRs shipped

In flight

$234

Open PR or inside attribution window

Unshipped — review

$119

Past attribution window; no matching PR

Install coverage

0/5

Active installs / expected

Dollars are a conservative list-price estimate from captured usage — real billed cost is reconciled from your invoice. Token counts are new work (input, output, and newly-cached context); repeated cache reads are excluded from tokens but still counted in billed cost.

Where the spend went

Spend states stay aggregate-first. Open a team or engineer only when you need diagnostics.

Merged PRs

54%

34.3M tokens · $542

Tied to merged PRs

In flight

24%

15.2M tokens · $234

Open PR or inside attribution window

No matching PR

13%

8.2M tokens · $119

Past attribution window; needs review

Non-PR work

6%

3.5M tokens · $52.21

Research, support, debugging, or setup

No git context

1%

900.0K tokens · $10.15

Captured outside a git tree

Other repo

2%

1.4M tokens · $22.70

Outside your tracked repo

Spend by model

Which Claude model versions the spend ran on, how much of each shipped, and the right-sizing headroom.

Opus 4.8

$794

81% · 43.4M tokens

54% shipped · tied to merged PRs

Sonnet 4.6

$167

17% · 14.9M tokens

61% shipped · tied to merged PRs

Haiku 4.5

$19.57

2% · 5.2M tokens

61% shipped · tied to merged PRs

Right-sizing headroom: up to $317. Opus spend ($794) would be ~$476 at Sonnet rates — review which flagship tasks could run on a smaller model.

Recommendations

What the agent flags for a manager to review — unshipped spend, budget spikes, or cap risk. 2 open flags.

  • Model right-sizingPlatform$255

    $255 model right-sizing headroom — Platform

    Platform ran $636 on Opus in the last 30 days; the same usage at Sonnet rates would be ~$382. Review whether these tasks needed the flagship model.

    AI spend tied to merged PRs (PR-only scope) · billing: metered API · 30-day attribution window · ≥3 contributors (k-anonymous) · metadata only — no prompts, code, or transcripts stored

    Period 5/8/20266/7/2026 · last scan 6/7/2026

  • Unshipped — reviewPlatform$119

    $119 unshipped — review

    Platform: 3 session window(s) in the last 30 days burned $119 with no matching merged PR past the attribution window. Could be abandoned work, research, or waste — review.

    AI spend tied to merged PRs (PR-only scope) · billing: metered API · 30-day attribution window · ≥3 contributors (k-anonymous) · metadata only — no prompts, code, or transcripts stored

    Period 5/5/20266/7/2026 · last scan 6/7/2026

Teams

2 teams
TeamEngineersTokensAvg tokens/PREst. spendShipped %
Platform342.8M6.8M$70548%
Billing220.7M6.6M$27567%

Each row is team totals · Shipped % = share of tokens tied to merged PRs · per-engineer diagnostics are admin-only in the full product.