Skip to main content
HOME>Use Cases>On-call Rotation Planning: A Fair Schedule That Survives Leave and New Hires
Use Case8 min read

On-call Rotation Planning: A Fair Schedule That Survives Leave and New Hires

The on-call block is built date by date within caps, blackouts and the escalation policy, calendar events follow approval, and leave, new hires and policy changes reassign only the dates they touch. The engineering manager approves every change.

Get started
Kylon TeamProduct

A fair rotation is easy to build once and hard to keep fair

On-call planning starts with a block of dates and a few rules. Every date has a primary and a secondary, nobody goes over their shift cap for the block, nobody is scheduled on a blackout date, the secondary never comes from the primary's team, and approved swaps are honored. Getting a fair four weeks out of that takes an afternoon. Keeping it fair through leave, a new hire and a policy change is where rotations go wrong.

Shift records live in Airtable, swap and blackout requests arrive as ClickUp tickets, the rotation and escalation policies are in Google Drive, and the shifts themselves are Google Calendar events engineers rely on at 3 AM. The manager's draft is a spreadsheet. A leave notice usually sets off a full rebuild, and the rebuild moves shifts people had already planned their weeks around.

Who works in the room

People

  • Engineering manager

    Owns the rotation and approves the plan, every reassignment and the closeout records

  • On-call engineers

    Post blackouts and swap requests, and confirm any shift that changes

  • SRE lead

    Owns the escalation policy and explains how each amendment applies

  • People partner

    Confirms leave dates and when a new hire becomes eligible for on-call

Agents

  • FinRotation planner

    Builds the block from Airtable shift records, ClickUp swap and blackout requests and the policy, and creates calendar events once the plan is approved

    AirtableClickUpGoogle CalendarGoogle Drive
  • NovaDraft reviewer

    Checks the manager's draft and every reassignment against caps, blackouts and the pairing rule

    Google DriveAirtable
  • MycoNotes and closeout

    Drafts shift change notes for a person to send, posts the final block in Slack and writes one closeout record per engineer

    GmailSlackAirtable

On-call planning lives in one Kylon room. Fin is the default there and owns the plan; under the room guidelines, Nova reviews drafts and reassignments, and Myco drafts the notes and writes the closeout records. When Fin finishes a reassignment, it mentions Nova with the changed dates listed, and Nova checks only those dates before the manager sees them. Fin creates events through a lasting grant the manager approved for the team's shared on-call calendar.

Building the block, then keeping it fair

The block runs from build to closeout in this order. Airtable records and calendar events change only after the manager approves.

  • Agent works
  • Person decides
  • Arrives or gets recorded
Scroll sideways to see the whole flow.
StageWhat happensWho actsWho decides
BuildEach date in the block gets a primary and a secondary within caps, blackouts, the escalation policy and approved swapsFin, from Airtable shifts, ClickUp requests and the policy in Google DriveThe engineering manager approves the plan
Events and draft reviewCalendar events are created for each approved shift, and the manager's draft is checked against the planFin for the events, Nova for the reviewThe manager approves the events and corrects the draft
LeaveExactly the dates the leave covers are reassigned, within caps and blackoutsFin, then Nova to checkThe manager approves and the new engineers confirm
New hireThe new engineer joins the roster from their eligible date and is offered only open or over-cap datesFinThe people partner confirms the date and the manager approves
Escalation amendmentDates that break the amended pairing rule are reassigned, and every other date staysFin, then Nova to checkThe SRE lead confirms the reading and the manager approves
Cap correctionsA lowered cap moves only the shifts over it; a raised cap moves nothingFinThe manager approves
CloseoutOne record per engineer with their shifts in the block, including engineers with noneMycoThe manager signs off the records
Stages of an on-call block. Engineers, dates and caps in this example are illustrative.

Leave, a new hire and a corrected cap

The narrowest reassignment is the safest one. When an engineer's leave covers five dates, Fin reassigns those five and touches no others, even if a wider shuffle would look tidier. Each replacement is checked against caps, blackouts and the pairing rule, and Nova checks the five again before the manager approves. Myco drafts a short note to each engineer whose shifts changed, and Fin moves their calendar events after approval.

A new hire or a cap correction looks like a reason to rebuild and is not one. A new engineer is offered open dates and dates where someone is over cap, nothing more. A cap correction for one engineer changes only the date it affects, and a raised cap never moves anyone, because nobody was over it. Two shift records sit just outside the block in Airtable; they stay unassigned whatever changes inside it.

What the checks catch

  • Shift records outside the block, which stay unassigned even when they look like gaps
  • A reassignment that strays beyond the dates the leave covers
  • A replacement that puts an engineer over their cap or on a blackout date
  • A primary and secondary from the same team after the escalation amendment
  • Closeout records missing for engineers who had no shifts in the block

What stays with people

  • Approving the plan and every reassignment before Airtable or the calendar changes
  • Granting a swap or a blackout the policy does not cover
  • Confirming how an amended escalation policy applies
  • Sending each note to an engineer, from a Gmail draft Myco prepares

Integrations in this workflow

Kylon is not a paging or scheduling system and it does not replace one. The shift records, the calendar and the request tracker stay where they are; the room is where the planning between them happens.

  • Airtable
  • ClickUp
  • Gmail
  • Google Calendar
  • Google Drive
  • Slack
  • Airtable holds every shift record and the closeout record for each engineer
  • ClickUp carries swap and blackout requests from engineers
  • Google Calendar carries an event for every approved shift
  • Google Drive keeps the rotation policy, the escalation policy and the manager's draft
  • Gmail sends shift change notes once a person sends Myco's draft
  • Slack is where engineers ask for last-minute swaps and where the final block is posted after approval

Where to start

The first block is the place to begin: one team, four weeks. Connect the Airtable base that holds shifts and the team's on-call calendar, and put this rule in the room guidelines: a reassignment touches only the dates it was asked to, and the manager approves every change before a calendar event moves.

“Leave used to mean a new rotation. Now it means five dates, five new names and nobody else's week disturbed.”

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.

Set-up prompt

Your first company harness. Where humans and agents run your business together.

Get started

More Use Cases

View all

API Deprecation Migration: Cutoff Waves That Stay Right as Integrations Move

Retiring an API version runs through one Kylon room. Agents set a disposition for each customer integration in Attio, schedule cutoff waves from the deprecation policy and track blocking bugs in GitHub, while the API product lead approves every disposition and notice.

Software & SaaSEngineering operationsEmail operations

Backup Restore Verification: Backups Proven Restorable Within Their Class Interval

A backup counts only once someone has proved it restores. Agents in the team's Kylon room read each backup-job issue in GitHub against restore reports in Box and live targets in Vercel and propose a verdict by class. The reliability lead approves before any issue changes.

Software & SaaSIT & security operations