AI Cost Tracking per Crawl Run
AI Cost Tracking per Crawl Run
As of v0.1.40, the platform tracks estimated OpenAI API token usage and cost for every crawl run. This gives the internal team full visibility into AI spend as crawl volume scales.
Cost Breakdown
Each crawl run is broken down into three AI-powered stages. Token consumption and estimated cost are recorded separately for each:
| Stage | Description |
|---|---|
| Skim | Initial pass over directory listings to identify candidate products |
| Scoring | Generation of the composite opportunity score (replicability, demand, revenue, competitive gaps) |
| Dossier generation | Full product dossier including mission statement, feature list, and competitor weakness analysis |
The per-stage cost breakdown is visible on the crawl run detail page.
Monthly Spend Tracking
All AI usage is persisted to the ai_usage table, which maintains a cumulative monthly total across all crawl runs. This allows the team to monitor overall OpenAI spend over time without needing to sum individual runs manually.
Cost Threshold Alerts
If a single crawl run exceeds a configurable cost threshold, an in-app notification is raised immediately.
- Default threshold: $5.00 per run
- The threshold is configurable — contact your administrator to adjust it.
- Alerts appear as in-app notifications; no external integrations are required.
This safeguard prevents unexpectedly expensive runs (e.g. a very large directory) from going unnoticed.
Notes
- Cost figures are estimates based on token counts and current OpenAI pricing at the time of the run. Actual billed amounts may differ slightly.
- Token and cost data is stored per run and is available for historical review on each run's detail page.