Spool.
Supervised AI work
See every AI task, result and decision together.
One of fifteen tools the studio built for its own work. We build the same kind of app for clients, shaped around the job they need doing.
- Made for
- Putting AI to work as a supervised team, not a chat window
- Shape
- Native Mac app
- Data
- Stays on the machine
What Spool does.
Spool keeps our AI work in one place. Each task has a priority and a project; an AI assistant picks it up, works in a separate copy and returns a complete record of what it did.
A person chooses the work and decides what happens to each result. Nothing is released automatically. A finished task can also trigger a separate review, so checking keeps pace with the work.
Let AI work through routine jobs overnight while your team sets the priorities and approves anything that matters. Every job leaves a clear record.
Why we built Spool.
Once several AI assistants are doing useful work, keeping track can become a job of its own. We were checking progress in one window, reading results in another and writing the same update in several places.
Spool brings every task, assistant and result into one place, with a person deciding what enters the queue and what happens to each result.
How Spool earns its place in a working day.
Put a task on the queue
A plain-English instruction, a priority, and a target project. Similar tasks already queued are flagged, so nothing is done twice.
An assistant starts work
The work happens in a separate copy, away from the live product. Priorities, schedules and dependencies set the order.
The work reports back
Every task ends with a complete record: what was done, how long it took, whether it succeeded. Anything that fails can be sent back to the queue with one click.
The checking keeps pace
When a task finishes, a written review of the work can queue itself automatically, so the checking never falls behind the doing.
Each result is recorded once
Finished work updates the task record, daily notes and items awaiting review without anyone copying the same update three times.
Overnight became working hours
Routine jobs go on the queue in the afternoon and are done by morning, each with a full record waiting.
Failures stopped hiding
A failed job is a red number on a dashboard, not a surprise discovered a week later.
Made to fit the work it needs to do.
Spool connects several of our other tools (the automatic reviewer, the working notes, the shared memory), but every connection is optional, off by default, and a missing tool never breaks another. That restraint is the design: each piece works alone, and works better together.
It is also a useful pattern for clients: a queue a person controls, AI assistants that report what they do and records that update automatically.
Spool sends each task to the chosen AI assistant, follows its progress and keeps a full record. A failed task only runs again when a person asks.
Could your business use something like this?
Tell us the job in plain English. We'll say what would solve it, what it would cost, and whether a desktop app is the right shape for it.