Time-Off Coverage: Every Request Checked Again When the Rules Change

Easy to approve one at a time, hard to approve as a set

The coverage policy in Google Drive says how many people on a team can be out on the same day, what balance a request needs, and when a request is waitlisted for a coverage conversation instead of approved. Requests live in a Notion database, the system of record, in the order they were submitted. Calendly shows the customer and candidate meetings people have already booked on the days they want off.

Most mistakes come after the first pass. A summary export leaves a request out, and nobody notices because the export looks complete. Leadership declares a blackout week after people have already asked for those days. Payroll corrects a balance, two people withdraw, and then the Sales blackout is lifted. If verdicts are written only for requests that change, the audit cannot show the others were checked at all.

Who works in the room

Most messages are about a single request, and those go to Ryan, the default agent. The export, the weekly audit and the replies belong to Fin. Ryan reads Calendly as evidence only and creates approved-leave events without editing existing ones. Notion fields are written under a lasting grant from the People Ops lead, after the lead approves each status. When a notice lands, Ryan re-checks the affected requests and mentions Fin with the ones that changed.

A coverage cycle, request by request

Requests are walked in submission order, because coverage is first come, first served under the policy, and every re-check keeps that order.

A blackout after the fact, then an exception to it

A blackout declared after requests are in changes the rules for a window that is already partly approved. Ryan re-checks every request that touches the blackout week from the top of the rule list, not just the approved ones, because a waitlisted request may now be declined and a declined one may need a different reason. Each result names the rule that decided it.

When the Sales blackout is lifted, the four Sales requests are re-checked without that rule and every other request keeps its status. The weekly audit still writes a verdict for all thirty, in order, whether or not anything changed, so the record shows every request was looked at. Fin drafts the reply on the Sales thread in Gmail, naming all four requests and their new statuses, and the People Ops lead sends it.

What the checks catch

• A request missing from the summary export that still needs a review flag
• A request that looked fine until the blackout week was declared, re-checked from the full rule list
• A customer meeting booked through Calendly on a day the requester wants off
• An audit that skips unchanged requests, written in full instead
• A reply that names only the Sales requests that changed, when all four belong in it

What stays with people

• Approving each status before it is written to Notion
• Deciding coverage with the team, which the manager does in the coverage conversation
• Declaring a blackout or an exception to one, which is leadership's call
• Sending every reply and signing the cycle closeout

Integrations in this workflow

Kylon is not an HR information system, and it does not replace the one that holds your leave balances. It works across the requests, calendars and policy the team already keeps, and keeps each request's reasoning in one place.

• Calendly
• Gmail
• Google Calendar
• Google Drive
• Notion
• Slack

• Notion holds the time-off database, the system of record for every request and its status
• Calendly shows meetings people have already booked, read as evidence only
• Google Calendar gets approved-leave events and coverage conversations, which agents create and never edit
• Google Drive keeps the coverage policy, the summary export and each notice from leadership and payroll
• Gmail carries the threads where teams ask for confirmation, answered with drafts a person sends
• Slack is where managers raise a coverage concern early, and Fin brings it into the request's thread

Where to start

Pick one month of requests. Connect the Notion time-off database, give Ryan read access to Calendly for the teams in scope, and add the rule the audit depends on to the room guidelines: requests are walked in submission order, every audit writes a verdict for every request, and the People Ops lead approves each status before it changes.

“Every request gets a verdict in every audit, so nobody has to ask whether it was checked.”

Set it up with your agent

Open a new Room, add your agent and paste this prompt. It connects the tools, builds the App with you and asks for your rules one step at a time, and nothing goes out until a person approves it.