## TL;DR
A bot booking meetings in the past is a timezone bug: the bot offered times in one zone and booked in another, or it validated availability against a stale or zoneless clock. Pin the visitor's timezone explicitly, validate every proposed slot is in the future in that zone, and confirm the booking in the visitor's local time. Never let a booking proceed for a time that has already passed.

## The query

```text
drift bot booked meeting in the past: timezone fix
```

## Use this when

- Meetings get booked in the past
- Offered times do not match the visitor's zone
- You are debugging scheduling bot behavior
- Auditing booking accuracy

## Not for

- Designing calendar interfaces
- Writing meeting agendas
- Sales outreach playbooks
- Choosing a scheduling tool

## Steps

### 1. Pin the visitor's timezone at conversation start

Detect or ask the visitor's timezone before offering any times. Every slot computation flows from the pin. Bots that assume the company's zone misbook every visitor elsewhere.

Expected output: an explicit visitor timezone on every scheduling conversation.

### 2. Validate each slot is future in the visitor's zone

Before offering or booking, check the slot against now in the pinned zone. Reject anything in the past with a clear message and offer the next valid slot. The check is cheap; the misbook is expensive.

Expected output: no past slots ever offered or booked.

### 3. Confirm bookings in the visitor's local time

The confirmation message states the time in the visitor's zone with the zone named: "Tuesday 2:00 PM Pacific." Confirmations in the company's zone confuse visitors and hide bugs. Local confirmation is both UX and a correctness check.

Expected output: local-time confirmations with named zones.

### 4. Make all scheduling timestamps timezone-aware

Store and compare in UTC, convert only for display. Naive datetimes are the root cause of most scheduling bugs: a zoneless "2 PM" means different moments to different code paths. Audit for naive timestamps and eliminate them.

Expected output: no naive timestamps in the scheduling path.

### 5. Reconcile bookings against the calendar regularly

Sample recent bookings and verify they landed at sensible future times. Reconciliation catches systematic drift (a zone misconfiguration) before it compounds into a week of misbooks.

Expected output: periodic booking reconciliation.

## Template: the checklist

```text
Scheduling timezone correctness:
[ ] Visitor timezone pinned before offering times
[ ] Every slot validated future in the visitor's zone
[ ] Confirmations in visitor local time with zone named
[ ] All timestamps timezone-aware, UTC in storage
[ ] Bookings reconciled against the calendar
```

## Variant phrasings

### bot scheduled meeting in the past

Steps 1 and 2. Pin the zone, validate the slot.

### scheduling bot wrong timezone

Steps 1 and 4. Pin plus timezone-aware timestamps.

### meeting booked at wrong time chatbot

Steps 3 and 5. Local confirmation, reconciliation.

## Why it happens

Scheduling code is written and tested in one timezone, then serves global visitors. The implicit "our timezone" assumption survives because tests use local fixtures and the developers never see the misbook: the visitor does. Timezone bugs are invisible to the team that ships them and obvious to everyone else, which is why visitor-side validation matters more than unit tests here.

## Edge cases

- Daylight-saving transitions: a slot that exists twice or never on transition days. Handle the ambiguity explicitly.
- Visitors who travel: the pinned zone governs the booking; confirm if the visitor mentions travel.
- Calendar sync delays: a slot free at offer time may book before confirmation. Re-validate at booking time.
- Rescheduling: the same pinning and validation apply; reschedules are where stale zones linger.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_Mikb7f09KsfOQ54-FDbRaw
