Retrospective AI margin diagnostic
Find margin leaks from the AI logs you already have.
Upload existing usage exports, choose a workflow, and see which customers, retries, failed runs, or pricing gaps changed the economics. The output is a concrete recommendation, not another token dashboard.
Analyze existing exports
Upload AI usage logs or CSV
Import workflow, customer, cost, retry, status, and price data from what you already have. The file is parsed locally in the browser and does not need provider keys.
Recommended decision
Fix workflow reliability before changing pricing.
Document Q&A is leaking spend through retries or failed outcomes. A price increase may hide the symptom while the workflow keeps wasting provider cost.
Includes successful, failed, and retried runs in the workflow baseline.
This is the spend that aggregate token dashboards usually hide.
Power users can be retention-positive and margin-negative at the same time.
The diagnostic also shows which fields are missing before this can be monitored reliably.
Events from CSV or integration
Imported log rows and future integration events use the same event shape, so the diagnostic can compare successful outcomes, retries, failed runs, and plan snapshots.
What this demo includes
Existing export input. Mock event preview.
The import path works with exported usage data. The tracked-event preview uses a deterministic mock provider to show the optional monitoring path without API keys.