The Spend Controller

For whoever signs off on compute spend.

One signal · 12 specialists · one result

Supervises other agents1
Keeps knowledge1
Runs infrastructure7
Audits and guards3

How the work moves

Four steps, each done by one specialist

Starts with

A request to start a machine or AI worker

First

1Refuses any start the daily limit cannot cover, with the reasonCompute Budget Guardian 2Records the cost of each AI worker's model useAI Spend Recorder

Then

3Adds the day's storage cost into the daily budgetStorage Spend Reconciler 4Turns the totals into a cost report by projectEngineering Cost Analyst

Ends with

A daily limit held and cost reported by project

the signalthe resultEach step is done by one specialist on the crew

The story

The owner sets a daily limit, and only the owner can change it. Every request to start a machine or an AI worker passes the agent first. A request that would break the limit is refused with the reason and the numbers. The finance lead gets a cost report by project.

The crew

The specialists behind this role, grouped by the kind of work they do.

Supervises other agents

Work Budget SetterSets a limit on the number of steps an AI worker may take and reports when the worker reaches it.Also serves The AI Workforce Manager

Keeps knowledge

AI Spend RecorderIt reads the model usage from the session, removes duplicates, and records the cost against the task it belongs to.Also serves The AI Workforce Manager, The Model Broker

Audits and guards

Compute Budget GuardianHolds the line on the daily spend that only you set: it refuses any start the budget cannot cover and says why, warns you as spend climbs, and stops paid machines at the cap.Also serves The Software Department, The Bulk Run Manager
Engineering Cost AnalystAttributes every model call to the task, role and phase that spent it and turns the totals into cost and human-equivalent effort hours per project.Also serves The Model Broker
GPU Spend and Power GuardKeeps rendering inside the daily spend limit and the office's power limit: no job starts without budget, and only one graphics card runs flat out at a time.Also serves The Studio Engineer

Runs infrastructure

Cloud Compute DispatcherThis agent picks an image, checks the container settings, runs the job on local or rented GPUs, keeps an eye on it and shuts the container down when it is done.Also serves The Branch Office Engineer
Storage Spend ReconcilerIt syncs the day's storage cost into the daily budget so tomorrow's limit reflects real spend.Also serves The Archive Keeper
Compute Cost ReporterReports what compute and AI usage cost, broken down by machine, by dataset, by team role and by model, with today's projection and storage costs alongside.Also serves The Morning Brief, The Bulk Run Manager
Idle Machine DrainerSets how long a machine may sit idle, empties its queues when the time is up and shuts it down.Also serves The Branch Office Engineer
Cloud Price WatcherIt compares the new price with the ceiling the owner set, then moves the work or holds it.Also serves The Bulk Run Manager
Recovery Policy SetterSets for each job whether a lost machine is replaced automatically, moved to owned hardware, paused for a decision or allowed to fail, and how much extra price is acceptable.Also serves The Bulk Run Manager
AI Provider SwitchboardSends every agent's model calls to an approved provider, keeps each provider inside its rate limit, steps down to a cheaper tier when a quota runs out, and blocks traffic that a provider's terms do not allow.Also serves The Model Broker

Where this crew also works

Each bar counts the specialists this role shares with another.

Routes

Routes this role works in. Its place in each one is filled in.

A new model becomes a story

A new function becomes a staffed desk

The engineering behind this role

AI subscriptions as governed workersCapability Budgets and refusalsIdea One call to a GPU anywhereEngineering story What the work cost, to the pennyEngineering story