USAGE SOURCETeaching sampleexample numbers · not your account

EXPECTED MONTHLY PROVIDER BILL$640.94after $4.06 of assumed local savings$9.06 leftThis plan leaves almost no room for a spike.thin confidence · 3/30 days observedTune assumptions

MONTHLY AI COST PLAN

This example pace lands near $645.00 in monthly paid AI spend.

This is teaching data, not your account. That estimate extends 3 observed calendar days across a 30-day month. 27 days are still outside the evidence window, so confidence is thin. Nothing on this page changes usage, providers, score, or rank.

01 · OBSERVED$64.503 calendar days · 3 active · 0 idle

02 · CALENDAR PACE$21.50 / day2026-07-03 → 2026-07-05

03 · 30-DAY PLAN$645.00$64.50 ÷ 3 days × 30

example not account data2,400 accepted rows3 providers3 calendar days observed

OPERATOR BRIEF · READ THIS FIRST

The decision, without the dashboard jargon

observed $64.50 → projected $645.00 → planned $640.94

Budget status$9.06 left

99% of the $650.00 limit. One heavy day can erase the buffer.

Forecast confidencethin · 3/30 days

27 days are unobserved. The current pace is extended 10.0×.

Review firsthiggsfield

Highest observed spend. That makes it the first place to inspect, not proof that it is wasteful.

Do nextCreate a real buffer

Review higgsfield jobs first, then lower paid usage or choose a limit you can enforce.

Plain English. We saw $64.50 across 3 calendar days, extended that pace to a 30-day estimate of $645.00, then subtracted $4.06 for the 35% local-work scenario. The result is $640.94 of estimated paid-provider spend against a $650.00 limit.

Expected bill$640.94

Monthly limit$650.00

EVIDENCE APERTUREthin confidence

Coverage: 10%
Projection stretch: 10.0×
+10% local: −$1.16

Read: Most of the 30-day window is unobserved. Use the projection as a directional warning, not a bill prediction. 27 of 30 calendar days are outside the evidence window. The same-rhythm projection stretches the observed pace 10.0×. Even at 100% of eligible work moved local, the scenario floor is $633.40.

CHANGE THE ASSUMPTIONS

Choose your comfort level

Updates instantly

Monthly spending limitThe most you are comfortable paying for AI providers in one month.$650.00Eligible work moved localA scenario for repeat work that could run on local tools instead of paid providers.35% → save $4.06

This percentage applies only to $11.60 of estimated eligible work, not to the full $645.00 forecast.

OPERATOR METHODCalculation & confidence ledger

Open six calculation rows, provider review, evidence boundaries, formula, and CLI dry run.

READ THE CALCULATION

Where the result comes from

Observed source period3 calendar days · 3 active · 0 idle$64.50
Same-rhythm 30-day projection$64.50 ÷ 3 days × 30$645.00
Estimated local savings35% of eligible work−$4.06
Planned paid spendForecast after this scenario$640.94
Your monthly limitThe boundary you selected$650.00
Budget remainingclose to the limit$9.06

WHERE TO REVIEW FIRSThiggsfield

It has the highest observed spend in this example. That does not mean it is wasteful; check its high-cost jobs before changing providers.

CONFIDENCE BOUNDARY

What is known, estimated, and unchanged

Known from the sample$64.50 observed across 3 calendar days · 3 active · 0 idle, plus provider totals and accepted rows.

Estimated hereThe 30-day projection assumes the same calendar rhythm. Local savings are a separate scenario. Either can be wrong if work changes.

Never changed hereUsage totals, provider settings, public profile, score, and rank. This page writes nothing.

Show the formula and CLI dry run

$64.50 ÷ 3 × 30 − ($11.60 × 35%) = $640.94

npx vibetrack mission --budget 650Preview only. The command retains --dry-run.