How to Build an Out-of-Office Coverage Handoff Flow in CogniAgent

Jun 26, 2026
4 min read

Most coverage handoffs happen as a rushed Slack message the day before someone leaves — or don’t happen at all, and the backup person finds out about an open task when a client follows up. This is a small, sequential flow: collect what’s open from the person leaving, then hand it to whoever’s covering, with enough context that they don’t have to ask follow-up questions.

The architecture at a glance

This is fully sequential — no branching needed, since every leave request goes through the same two stages. If you later want different handling for, say, leave longer than two weeks vs. a single day off, that’s where an Activation Condition would come back in.

Step 0: Add the channel in Flow Settings

Open Flow Settings -> Channels -> Add Communication Channel. Custom Slack Bot or SlackBot (CogniAgent) is the natural fit here — both the person going on leave and their backup are almost certainly already in Slack, and this flow benefits from feeling like a quick conversation rather than a form to fill out.

Step 1: Build the Handoff Intake Actor

This actor’s whole job is asking the person leaving a short set of questions before they go.

  • Context: company name, team structure, tone of voice — set here since it may be inherited by child actors, so the Backup Notification Actor doesn’t need it re-defined.
  • Instructions: “Ask the person what’s currently open or in progress, who should cover each item while they’re out, and anything time-sensitive that needs attention during their leave. Ask naturally, one or two things at a time — don’t present this as a form.”
  • Definition of Done: “Open tasks, assigned backup per task, and any time-sensitive items have all been captured.”

Keep this actor narrow — it collects, it doesn’t notify. That’s the next actor’s job, and splitting it keeps each actor’s Instructions simple.

Step 2: Build the Backup Notification Actor

  • Inherit Context: check both Global flow context and Collected information so far — this actor needs everything the Intake Actor just gathered, not a fresh start.
  • Instructions: “For each backup person identified, send them a clear summary of what they’re covering, any time-sensitive items, and how long the person is out. Confirm each backup person has seen the message.”
  • Definition of Done: “Every named backup person has received their assigned context and acknowledged it.”

If there are multiple backups for different tasks, this actor needs to handle sending separate, relevant context to each person rather than one blanket message to everyone — nobody wants to read about three other people’s tasks to find the one line that’s theirs.

Step 3: Set Focus Mode

  • Handoff Intake Actor: Persistent is a good default — you want it to stay on task collection even if the person starts chatting about their vacation plans, without being so rigid it feels robotic.
  • Backup Notification Actor: Flexible or Auto works fine — it’s mostly a one-way delivery, low risk of drift.

A note on what NOT to do

Don’t skip Definition of Done on the Intake Actor — without a clear signal that all open items and their backups have been captured, the Notification Actor will fire on incomplete information, and you’ll end up with a backup person missing context on the one thing that actually mattered. And don’t let the Notification Actor batch everyone’s tasks into a single broadcast message — that defeats the purpose of having captured who covers what in the first place.

Why this pattern matters 

Same two-stage shape — narrow intake actor collects structured information, a second actor distributes the right slice of it to the right people — works for shift handoffs, project handoff between team members, or any “before you go, here’s what needs to carry over” situation.

Check our other guides to learn how you can enrich the logic and build fully automated processes!

Other Guides