5.0 Insights

See what merged, how fast, and at what cost.

One dashboard for all the work: what merged, how long it took, where it waits and how often a person had to step in. Rules watch the numbers and raise a flag, and every change to your process is marked on the charts.

The Performance dashboard: changes merged over seven days, arrivals against finished work, cycle time at p50 and p90, the outcome of finished work per day (clean, after rework, failed, cancelled), review passed first time, checks passed first time and re-gates per day.

5.1 Dashboard

What is moving, what is stuck, and what it costs

Start with Overview for what is happening now, then go deeper. Each view answers one question, from what is moving this minute to what changed in your process last week.

The Overview dashboard: work by stage from intake through agents, checks and the merge gate to the merge queue, work in flight and merged, and the last 24 hours: main green, cycle time, autonomy, decisions asked and cost per merge.
01

Overview

Health, and the work in flight by stage: intake, agents, checks, merge gate and merge queue.

02

Performance

Merged against your goal, cycle time at p50 and p90, review and checks passed first time.

03

Bottlenecks

Where the time goes, so you know what to fix first.

04

Merge

How changes get through the merge gate and the queue into main.

05

Health

Which rules fired, how fresh each source is, and main over time.

06

Activity

What changed in your process and when, next to the numbers.

5.2 Measures

Speed, quality and cost of every change

The delivery numbers your team already knows, plus the two that agents add: what each merged change cost and how often a person was asked.

The Bottlenecks dashboard: an agent account’s five-hour and weekly usage against its reserve line, node load, queue depth against the limit on work in flight, time from ready to main at p50 and p90, and merge gate passes per merge by reason.
  • Changes merged per week
  • Merged against the goal
  • Cycle time, p50 and p90
  • Where time goes
  • Review passed first time
  • Checks passed first time
  • Main branch green
  • Outcome of finished work
  • Cost per merged change
  • Decisions asked of people
  • Arrivals and completions
  • Age of open work

5.3 Signals

When a number drifts, you hear about it

About 20 rules watch the numbers, each with a threshold you set. When one fires, a flag lands in one Attention inbox: the build on main broke, work has waited too long for a person, throughput dropped, cost drifted, a test file got noisy, a source went silent.

MAIN BRANCH WAIT FOR A PERSON THROUGHPUT COST PER CHANGE TEST FILE FAILURES SOURCE ACTIVITY THRESHOLD RULES ABOUT 20 · THRESHOLDS YOU SET BUILD BROKEN COST DRIFTED ATTENTION ONE INBOX FOR PEOPLE

5.4 Feedback loops

Problems become work, not noise

What goes wrong, or turns up along the way, comes back to the same queue as new work. Agents file what they find outside their task as separate work, and review remarks that do not block become follow-up work. A broken build on main becomes repair work and an alert becomes an investigation. The factory also files its own incidents and alerts you.

MAIN QUEUE AGENT WORKS REVIEW BUILD BREAKS REPAIR WORK FOUND OUTSIDE ITS TASK FOLLOW-UP WORK NON-BLOCKING REMARK FOLLOW-UP WORK ALERT INVESTIGATION FILED AS NEW WORK BACK IN THE SAME QUEUE

5.5 Process changes

See what each process change did to the numbers

Each new version of a workflow or of a project’s rules is marked on the charts. You read the numbers on either side of the mark and decide whether the change helped. The factory shows the evidence; a person makes the call.

01

Marked on the charts

Each change shows on the Performance charts at the moment it went live.

02

Listed in Activity

Every change with its version and the note written with it.

03

One step back

If it did not help, activate the previous version. Work in flight stays on the version it started with.

Keep your process.
Add the agents.