How this studio is built

What the studio costs, and how we trim it

Usage & Optimization

Every message resends a session's whole history, so a long session costs more on every turn. The studio keeps sessions short, runs one model at low effort by default, paces itself loosely by the day, and saves quota-hungry work for the end of the week, when unused quota would otherwise expire.

Usage & Optimization: how it runs, as a flowchart1A task is assignedBy The Owner or the Project Managerthe task, to one Agent as active work2Work at low effortthe default model at low effort, raised only if the task needs itThe Agent3Checkpointa work-log entry, an updated state file and a commitThe Agent4Clear the sessiona fresh session for the next taskThe Owner5Watch the pacethe day's usage against a loose 15% guideProject Manager6Spend the surplusbriefed surplus-queue work, dispatched before the weekly quotaresetsProject ManagerUsage & Optimization: how it runs, as a flowchart1A task is assignedBy The Owner or the Project Managerthe task, to one Agent as active work2Work at low effortthe default model at low effort, raisedonly if the task needs itThe Agent3Checkpointa work-log entry, an updated state fileand a commitThe Agent4Clear the sessiona fresh session for the next taskThe Owner5Watch the paceBy Project Managerthe day's usage against a loose 15% guide6Spend the surplusBy Project Managerbriefed surplus-queue work, dispatchedbefore the weekly quota resets

The pieces

Checkpoint and clear
Finish the task, write down where you are, start a fresh session.
One default model
Every Agent runs the same model at low effort unless a task needs more.
A loose daily pace
About 15% of the weekly quota a day, watched by the Project Manager, not by every Agent.
Three kinds of work
Active work now, a backlog for later, and a surplus queue for spare quota at the end of the week.
Measure round trips, not words
Short messages save almost nothing; an extra round trip costs a whole re-read.

Where this was worked out