Executive Assistant Rescheduling Workflow: Repair a Priority Swap

Repair a priority swap in eight steps

Use one controlled scenario: an urgent executive-customer decision displaces a partner review, while the partner review’s only reasonable replacement overlaps an internal operating meeting with one external adviser. The EA decides the priority. The workflow must repair only the meetings that decision actually breaks.

1. 1

   Record the executive’s priority decision

   Name the new meeting, reason class, owner, required and optional people, duration, deadline, protected window, confidentiality level, organizer calendar, approval mode, and the exact event it may displace. Keep private reasons in an EA-only field. The system needs the scheduling consequence, not the whole executive context.
   • The EA or another authorized owner makes the priority decision.
   • Urgency does not silently waive protected time, travel, preparation, or attendee rules.
   • The new brief defines completion as one verified invitation or a bounded exception.
2. 2

   Build the impact graph before moving anything

   List the displaced event, its attendees, companies, providers, required and optional status, dependencies, preparation blocks, travel, holds, deadlines, and organizer. Then add any event that would be affected by a candidate replacement slot. Mark unaffected meetings as frozen so a broad optimization pass cannot churn the whole week.
   • Every affected person and calendar has an explicit reason for inclusion.
   • Optional attendees do not become blockers by accident.
   • Existing events stay unchanged until the repair plan is approved.
3. 3

   Classify each meeting as protect, move, shorten, delegate, or stop

   Apply the executive calendar policy and relationship context. Protect meetings whose value or commitment outweighs the new request. Move only when authority exists. Shorten, delegate, substitute, or cancel only when those choices are already approved or returned to the EA as one decision. Software should expose the tradeoff, not make the political call.
   • Required roles and approved substitutes are distinct from optional observers.
   • Cancellation and rescheduling use separate authority and wording.
   • A sensitive relationship can remain fully EA-owned.
4. 4

   Set the disclosure and outreach boundary

   For every affected participant, record what reason may be shared, approved channel, local-time send window, response deadline, reminder cap, and opt-out or human-contact preference. A client may hear ‘an executive conflict arose’ without receiving the name, topic, or attendee list of the meeting that displaced them.
   • Possessing contact details does not grant permission to send a rescheduling message.
   • Private event titles and unrelated attendee names stay out of the repair thread.
   • The EA approves relationship-sensitive wording before it leaves the office.
5. 5

   Reuse fresh evidence and collect only what is missing

   Read approved Google and Microsoft free/busy for connected required people. Reuse still-current bounded availability from the original coordination record. Ask only affected or stale participants for new windows, with calendar connection optional. Normalize source, time zone, freshness, consent, and response state before searching.
   • An unchanged participant is not chased again without a reason.
   • Missing access or an unanswered request is not treated as open time.
   • A correction invalidates proposals derived from the old window.
6. 6

   Generate a repair plan for the smallest meeting set

   Find overlap for every required person in each affected meeting, preserve buffers and working-hour policy, and rank plans by number of moved meetings, optional attendance, participant burden, deadline, and executive preference. Return the exact sequence: new urgent meeting, replacement partner review, and any operating-meeting change that sequence requires.
   • A high score cannot drop a required person or exceed rescheduling authority.
   • The plan names every attendee who will receive an update.
   • No-overlap returns the smallest duration, date, substitute, or priority decision the EA can make.
7. 7

   Approve and update one canonical event at a time

   Bind approval to the repair revision, ordered plan, event identifiers, exact attendee sets, new times, organizer calendars, and approved messages. Re-check fresh required free/busy before each update. Use idempotent commands, reconcile uncertain provider responses before retry, and stop the remaining sequence if an earlier update fails in a way that changes the plan.
   • Repeated approval or worker delivery does not create duplicate events or notices.
   • The workflow distinguishes an update from a cancel-and-recreate operation.
   • The EA can pause, revise, reclaim, or end the repair between events.
8. 8

   Verify invitations and close the exception ledger

   Read back every changed event and verify time, zone, organizer, required people, optional-attendee decision, wording, and update delivery. Record unresolved declines, expired approvals, stale evidence, duplicate-message suppression, and human reclaims. Return one concise result to the EA instead of forwarding the whole thread.
   • Every changed participant receives the correct current invitation or update.
   • Unchanged events and private fields remain untouched.
   • The final record explains what moved, why it moved, and what still needs judgment.

Compare rescheduling by the repair work it finishes

The useful workflow protects executive judgment while carrying the mechanical tail from one priority decision to a verified set of current invitations.