Routing in Pylon
Two layers. Triggers send the issue to a team. The team's assignment strategy picks the person.
- Issue Routing (push): round robin with capacity, Active-only, and a fallback.
- Recurring (schedule): time blocks, no capacity. Being phased out.
- Pull queue: reps click Take.
- Direct: trigger assigns a person, account owner, or AI agent. Skips the team strategy.
Do
- Route to a team in triggers. Let the team pick the person. Hard-coded assignees cause overwrite fights.
- One master routing trigger, narrow branches above the fallback. Setters above dependents. Triggers run top to bottom and never re-run earlier ones.
- Turn on "Only assign to Active" and use support hours so status flips automatically.
- Fallback is "Leave in team queue". "Leave unassigned" never retries.
- Set capacity when adding someone to a team. Default is unlimited and will drain the queue.
- Turn on "Override existing assignee" for any team that takes handoffs. Otherwise the old assignee sticks and routing silently skips.
- To re-run routing, set team to No Team, then back. Routing only fires on team change.
- AI conditions: one trigger per destination, one true-only statement, structured filters first.
Don't
- Don't use Recurring for "just assign to people". Anything outside a block stays unassigned forever.
- Don't stack two assign triggers on the same kickoff. The first wins, the second silently no-ops.
- Don't expect round robin to run on reopen, unassign, or customer reply. Only on team change.
- Don't put capacity on the parent and expect the subteam to honor it. Everything is per team, no fallback to parent.
- Don't rebuild triggers live. Agree the flow, then rebuild in one pass.