Skip to main content

Handover fails on the pending items, not the obvious ones

Handover is done verbally and under time pressure at the end of a strenuous shift. What gets communicated is whatever is foremost in mind, and what is foremost in mind is simply the most recent thing that happened.

The items that are missed are the quiet ones: something expected, a call to be returned, and a form to be signed before discharge. None of them is urgent at handover. All of them are important by morning.

Written and spoken forms are complementary, not alternatives. A written administrative list lets the conversation focus on what actually needs discussion: judgment rather than status.

The clinical handover is still a conversation between clinicians. This flow produces the administrative layer for that conversation: outstanding tasks, awaited items, and time-bound actions. Nothing in it is a clinical assessment.

How the ActionFlow runs on the canvas

The shift change determines the schedule. Just before it, a Start Scheduled node runs so the list is available when the conversation begins, not after.

The following sources are processed in parallel: an HTTP Request node for outstanding orders and results awaiting treatment, a PostgreSQL node for administrative tasks, and a Jira or Linear node for tracking operational work.

A Merge node combines them. A Filter node keeps only items that are still open, because a handover list that includes completed work is one people stop reading.

The Date node identifies anything with a time associated with it: a due result, an expected call, or a deadline for the next shift. The Sort node then prioritizes these items.

The Generate Object node groups the remaining items by area and presents them as a structure. It leaves the awaited-items group separate, because that is the group that is forgotten.

A Generate Text node writes the handover note within the length limit. A Slack or Teams node posts it to the incoming shift's channel.

The HTTP Request node also records it. If a question is asked at three in the morning, a written answer already exists.

The note contains no clinical information, and no information is generated.

Nodes this ActionFlow uses

  • Start Scheduled: runs shortly before shift change so the list precedes the conversation.
  • HTTP Request (Util): reads outstanding orders and awaited results.
  • PostgreSQL (Data): supplies administrative tasks. Linear covers operationally tracked work.
  • Merge (Control) plus Filter (Data): assemble every source and keep only open items.
  • Date (Util): identifies time-bound items so they lead the note.
  • Generate Object (AI Core): groups by area and keeps awaited items as their own section.
  • Slack (Communication): posts to the incoming shift. Teams fits the same slot.

Frequently asked questions

Start building AI workflows

Create a free account, open a template or a blank canvas, and run your first ActionFlow.

Newsletter

Get product updates

New nodes, agents, and product notes. We send mail only when we have something worth opening.

Unsubscribe at any time.