Task assignment to AI workers

BurstyAI runs tasks through configured AI workers set up to handle a specific piece of small business work. A Beam agent creates a task when a trigger fires, such as a new lead or a recurring weekly job, assigns it to the relevant worker, and reads back the result once the worker completes it. Where the output matches the rule the customer has approved, the agent updates the linked record and moves on. Tasks where the worker's output looks incomplete or does not match the expected format are routed to a person to review before anything downstream is updated.

Task status and completion tracking

A Beam agent reads the status of tasks running through BurstyAI, whether queued, in progress, completed, or failed, and keeps other systems in sync with that state. On a schedule, it checks for tasks that have finished and applies the agreed rule, such as marking a linked record complete or notifying the person waiting on the result. Tasks that stay in progress well beyond their expected time, or that fail outright, are flagged to a person rather than left to resolve on their own, since a stalled task usually needs a human decision to move forward.

Worker output review routing

Since BurstyAI's workers produce output that feeds into other small business processes, a Beam agent can read that output and check it against a defined rule before it is used elsewhere, for example confirming a generated summary or draft meets a minimum length or contains the required fields. Output that passes is passed along automatically to the next step or system. Output that fails the check, looks incomplete, or falls outside the expected format is routed to a person to review and correct before it is used, keeping unreviewed AI output out of downstream records.