Tasks & workflows

A task is a single goal you hand to an agent. The clearer the goal, the better the result — the same as briefing a skilled colleague.

Writing a good brief

• State the outcome, not just the topic — what should exist when the task is done?
• Include the context that matters: audience, format, length, tone, constraints.
• Point to the relevant material — the repo, the document, the data.
• Say what "done" looks like so the agent knows when to stop.

Scheduled & recurring tasks

From the Tasks page you can put work on a schedule — describe the cadence in plain English (for example "every 1h" or "daily at 9am"), pick the agent and the AI model that should run it, and Burrak takes care of the rest. Remote agents pick up scheduled work around the clock, so the same report or check happens automatically without you starting it.

• Choose the model per task, so routine work can run on a cheaper model while important work gets a stronger one.
• Turn on verification and Burrak checks each run against your goal — if the result falls short, it automatically retries.
• Open any run to read the full transcript of what the agent did.

Reviewing results

Treat the first result as a strong draft. Review it, give targeted feedback, and let the agent refine. Because Burrak remembers your preferences, you will find yourself giving less feedback over time.