Agents Builders

Task Intelligence

How work flows through the pipeline — stage speed, cycle time, token and dollar spend, and how our sealed-bid estimates hold up against what shipping actually took.

Tasks tracked

510

Shipped

464

Total spend

$17,509.23

Tokens burned

23,843,446,962

Avg cycle time

6.4h

Avg cost / task

$49.46

Stage events

4,198

Estimate hit rate

1%

Slowest stage right now: Shipped (avg 22.01h before moving on).

Stage speed

Average vs median hours a task waits in each stage before moving on.

Loading...

Cycle time by task

Hours from created to shipped, longest first (top 12).

Loading...

Cycle time trend

Average task cycle time (hours) by the week it shipped.

Loading...

Tokens per task

Total tokens (in + out) reported per task (top 12).

Loading...

Tokens by stage

Where the tokens burn — total reported per stage worked.

Loading...

Tokens over time

Tokens reported across all stage events, by week.

Loading...

Cost per task

Total agent dollar spend per task (top 12).

Loading...

Spend over time

Dollar spend across all stage events, by week.

Loading...

Estimate vs actual

Sealed-bid sizes vs what shipping took, in points (1 S · 2 M · 3 L · 4 XL). Sparse until tasks ship with a derived actual size.

Loading...

Model mix

Dollar spend share by model.

Loading...

Priciest tasks

Most expensive work, by total agent spend.

Task Tokens Cost
Secure learning-loop grade endpoints 791,976,194 $491.30
Redact secrets from capture hook 753,213,381 $463.58
Root merge gates at worktree 726,653,997 $444.43
Rolio demo polish rehearsal 570,791,320 $432.71
Correct stale ship spec header 685,867,308 $420.01
Scope span invariant per agent 673,547,183 $412.64
Feed banked insights forward 630,692,083 $388.65
Agent span grading endpoint 593,592,010 $368.98
Generate insights doc from bank 540,953,928 $339.44
Make grading actions public 449,195,989 $283.25
Distillation pipeline three-column page 419,982,979 $267.33
Fix deployment duration metrics 363,098,882 $243.25

Biggest estimate misses

Dev estimate vs actual size, largest gap first.