Task Status Updates

Dart holds the current status of every task a team is working on, from open to in review to done. A Beam agent reads status-relevant signals, such as a linked pull request closing or a checklist being completed, and applies the team's approved rule for what that means for the task's status. Once the rule clearly applies, the agent updates the task in Dart. Signals that could support more than one status, or that touch a task marked high priority, are routed to a person to confirm before anything changes.

AI-Assisted Prioritization

Dart uses AI to suggest which tasks in a backlog deserve attention first, based on signals like due dates, dependencies, and how long a task has sat untouched. A Beam agent reads Dart's suggested priority order alongside the team's approved criteria, such as never bumping a task past one already marked urgent, and applies the reordering only where those criteria clearly hold. It notifies the task owner whenever a priority changes so nothing shifts silently. Suggestions that would reorder tasks across different owners or teams are routed to a person rather than applied automatically.

Milestone Progress Updates

Dart groups related tasks under milestones so a team can see how a larger piece of work is progressing. A Beam agent reads the completion state of tasks under a milestone and, once every task tied to it is done, updates the milestone's status and notifies the owner. Where a milestone is close to its due date with open tasks remaining, the agent flags it rather than changing the status. Milestones with conflicting task statuses, or ones close to deadline, are routed to a person to look at.