What is a managed task, and why we charge per task, not per hour
AI does the work, a person is accountable. What a managed task is, what its five parts are, and why charging per task lets you do the math before you start.
Signing up for AI today is easy. Anyone can pay for a subscription and start using it.
The hard part is something else: keeping it working month after month, having someone review what it does, and having someone take responsibility when it gets something wrong.
That is a managed task: AI does the work, a person is accountable.
The five parts of a task
Before a task can be managed, it has to be well defined. Every task has five parts:
- Input. What it receives: an email, an Excel file, an audio recording, a document.
- Output. What it delivers: a record entered into the system, a summary, a reply.
- Frequency. Once, every day, every week, every month.
- Acceptance test. How we both know it works.
- Price. Agreed on before the work begins.
If any of the five is unclear, the task isn't ready yet. And a task that isn't ready doesn't get automated: it gets organized first.
An example
A distributor receives orders through WhatsApp. Someone copies them by hand into a spreadsheet. Every day, for several hours.
- Input: order messages on WhatsApp.
- Output: each order entered in the sales spreadsheet, with customer, product, and quantity.
- Frequency: daily.
- Acceptance test: a weekly sample of orders with no data-entry errors.
- Price: monthly retainer.
AI reads the messages and enters the orders. The Router reviews the unclear cases, corrects them, and adjusts the task so the error doesn't happen again. Each month, the owner receives a report: what was done, what failed, what was corrected, and how much time was saved.
Why per task, not per hour
Billing per hour rewards whoever takes longer. Billing per task rewards whoever gets it done.
And there is a more important reason: with task-based pricing, the owner can do the math. They know how many of their team's hours the task takes today, what those hours are worth, and what they gain if we handle it. If the math doesn't work, it isn't worth doing. And we tell them.
What happens when AI makes a mistake
At some point, it will make a mistake. That is why every task has a human checkpoint written into its acceptance test.
That is what you pay for with a monthly retainer: not the AI itself, but a named person who is accountable for it. At Inbucle, we call that person the Router: they decide which path each task takes, whether it is handled by AI, a tool, or a person, and they are accountable for the outcome.
What a managed task is not
- It isn't 24/7 support. It has set hours and written response times.
- It isn't magic. If the input data is disorganized, it gets organized first.
- It isn't forever. If a task stops making sense, it is retired.
If you want to see what the plans look like, they are on the managed tasks page.