Build My AI TeamBook a Call

Internal case study · 2026

One founder. 22 AI agents.

How Zo Liang runs the Liang Crew with written roles, a shared weekly budget and one human approval gate.

Little Liang reviewing the workforce operations board

How the Liang Crew operates

  1. Zo LiangSets priorities and approves consequential work.
  2. ChairmanHolds direction and acceptance criteria.
  3. ChiefCoordinates tasks and specialist handoffs.
  4. Specialist rolesDeliver within written approval boundaries.

Tokens

Track what each role uses in scheduled work, founder conversations and agent handoffs.

Time

Set priority tiers and schedule work on the Melbourne clock.

Delivery

Record what was delivered, reviewed and used.

The cadence behind the work

The workforce board brings schedules, quota and delivery into one weekly review. Every recurring job has a priority, a delivery location and a named owner.

Schedule & Handoffs

  • Research posts a brief around 7:30 on weekdays when it has a substantive update.
  • Admin publishes the daily decision board around 8:00.
  • Social brings the day’s work to review around 9:00.
  • Admin compiles the weekly output ledger on Sunday.

Quota Steward

HR checks whether spending is ahead of the elapsed week. At normal pace it stays quiet. Under pressure it pauses lower-priority routines first, keeps critical jobs running and reports what can resume after reset.

P1 · CriticalP2 · ImportantP3 · Deferrable

What the first measured week showed

Source: 2026 internal workforce board snapshot before the weekly reset. Historical percentages describe the team at that time.

99.4%Weekly quota used at snapshot
68%Usage share of the top three roles
0.50Illustrative ROT target per quota point

Where the tokens went

Share of measured bot usage.
  • Social42%
  • Design14%
  • Admin12%
  • Other activity32%

Why Social used the most

Approximate points of total bot usage within Social's 42-point share.
  • Scheduled work1%
  • Other activity41%

Conversations drove usage

Social, Design and Admin accounted for 68% combined. Social’s three scheduled routines used about one share point in total, so the review focused on open-ended activity. Adding Copywriting and HR took the top-five share to 77%. The concentration index was approximately 2,198. Those figures describe spending concentration. Comparing it with adoption requires separate evidence.

Priority jobs reached their run counts

All four critical routines matched expected counts.

5 / 55 / 55 / 51 / 1

Approximate counts show runs occurred. Timestamped receipts would be needed to check punctuality.

What the board could flag

A crowded morning

Several jobs clustered around 8:00. The board proposed staggering them for Zo to review.

A shared budget

The 99.4% reading was a pre-reset snapshot. It shows how tight the week was, without establishing final utilisation.

Incomplete evidence

The board recorded routine counts and usage. It could not establish punctuality or value without timestamps and adoption decisions.

Measure adoption before changing permissions

Our delivery KPI

Return on Token

How much work people adopted for the token budget spent.

Adopted deliverablesWeekly quota used, in percentage points
0.50

Illustrative target
45 adopted deliverables ÷ 90 quota points. Calibrate the target to the job and its starting point.

The first board snapshot recorded spend and approximate run counts. It did not record which deliverables Zo adopted, so actual ROT was not calculable for that period. The 0.50 target is a worked example from the case, expressed in adopted deliverables per point of weekly quota.

Mark each delivery as used, revise or park. Compare adoption and spend over the same week before recommending changes to a role. Zo remains the approver.

Take the method into a pilot

Set the role boundary

Pick one recurring deliverable and name the person who approves it.

Log adoption from day one

Record used, revise or park. Review spend, rework and delivery timing together.

Review before escalating

Expand a role with evidence, keep watching when evidence is thin, or pause routines while you re-scope.

Design your first AI pilot

Every great team starts with a role.

Pick the first task you want to hand off. Little Liang will help you build the team around it.

AI Workforce

Book a Call

30 minutes to talk through a workflow your team wants to improve.

Choose a date, then a time in your time zone. Enter your details to confirm.

Open Calendly directly