Overview
Your AI spend right now, from your providers' own usage reports.
Daily spend by source
at list prices · click a day for the breakdownTop actionable insights
found by nightly checks and AI review, per monthCost per task
first week → last week of the periodSpend by model
API keys
"save up to" is the biggest open no-tradeoff fix on each keyMonth-end forecast
Where your savings came from
verified per key against your first auditLive
Spend as it happens, from your providers' per-minute usage reports. Providers publish usage about 5 minutes after each request.
Spend per minute, last 60 minutes
Model mix today
share of spend since midnightTop API keys, last hour
re-ranks as traffic shiftsLatest minutes
newest firstCost explorer
Daily API spend at list price. Tap or hover a bar for the day's breakdown.
Spend per minute
Daily cost
Breakdown
Key hygiene
Recommendations
Actionable insights from your usage: savings, risks and optimizations, each sized in dollars per month with the exact fix. The low end uses cautious assumptions. Set a status to track what has shipped.
How these are found
Nightly, automated checks run on your providers' usage and cost reports and work out every figure with plain arithmetic, so each one can be traced back to your data. An AI review then reads summary figures (never your prompts) and adds insights the checks don't cover; those are labeled Found by AI review. To get any fix explained in plain words, ask the TraxAgent AI assistant (powered by Claude).
Top 5 to do first
the biggest open fixes, per month · the full list is belowDevelopers
Who spends what on coding agents and the API, and what it produces.
API keys
Every key at a glance: owner, spend and trend, open fixes, and how much of its spend is waste. Highest waste first.
Spend share by key
a month, at the current rateWaste to recover
biggest open no-tradeoff fix per key, $/moTeams
Each team's spend, its budget and pace, who owns it and what it could save. In your account you create the teams and assign keys and people on this page.
Manage teams
where each key's spend goes · read-only in the demoIn your account you add teams here, then send each workspace, project or key to one. Spend splits by team everywhere once it's assigned.
| API key | Workspace or project | Team | Spend |
|---|
Alerts
What TraxAgent emails, posts to Slack or sends to your webhook, and when. In your account you set each threshold and who gets it.
Recent alerts
what Demo Co would have received this monthCaps
A cap is a spend limit per person on AI coding tools (Claude Code, Codex and Cursor), per day, week or month. It answers two questions: how much AI should each engineer spend, and what happens if we cut it? Try an amount on Demo Co's last 30 days of sample agent spend to see who it would touch, how often, and how much spend was above it, then compare with the suggested cap for each team, which is set from how that team actually works. Before cutting, the cheaper-model table shows what moving routine work one tier down would save.
In your account you name your own caps and teams, get an email at 80% and 100% of a cap, approve requests for a higher one, and see how much of each cap people really use. TraxAgent reads usage only, so caps warn; they never block anyone's work.
What would a cap have done?
Demo Co, last 30 days| Team | People | Would hit it | Times over | Above cap | Suggested cap |
|---|
Your caps
samples · editable in your accountCheaper models before lower caps
same work, one tier down| Move | Spend | Would save | People |
|---|