Reservations AI Toolkit

Who this helps: Hotel Apartment reservations managers, booking agents and front-office coordinators.
What you can do: Use the prompts to review booking quality, modifications, cancellations and allocation exceptions before confirmation.

Track C · Hotel Apartment & Serviced Apartment AI Operations Playbook · Reservations chapter

A reservation is more than a name and an arrival date. It is a compact operating contract connecting inventory, price, room type, occupancy, cancellation conditions, payment or billing arrangements, channel rules, special requests and the promises the property will later be expected to deliver.

That makes Reservations an excellent use case for AI—and a dangerous place to use AI casually. A model can summarize a booking in seconds, but it can also turn a request into an entitlement, interpret the wrong cancellation rule, assume two similar bookings are duplicates, or suggest a room-type solution without understanding live inventory. The damage may not appear until check-in, when Front Office inherits a promise that was never operationally valid.

The control principle for this chapter is:

AI may organize reservation evidence and prepare a decision. The approved reservation system, commercial terms and authorized human still determine what is confirmed.

The ten prompts follow the reservation-control cycle:

Validate → Review change → Explain terms → Review no-show → Support allocation → Structure requests → Detect duplicates → Handover → Validate corporate terms → Escalate

Why reservation quality is an operations issue

Reservation errors travel downstream. An incorrect room type becomes a Front Office conflict. A missed child occupancy detail can affect bedding and capacity. An unverified airport-transfer note becomes an arrival failure. A corporate inclusion copied from an old booking can become a billing dispute. A vague cancellation explanation can become a complaint.

For that reason, the objective of Reservations AI should not be “answer faster.” It should be reduce ambiguity before the booking reaches the next operating stage.

A useful AI workflow can compare supplied records, extract requests, expose contradictions, prepare a structured handover and identify which cases need human intervention. It should not silently resolve a conflict by choosing whichever field looks most likely.

HOTEL framework for reservations

The common HOTEL framework applies directly:

• H — Hospitality role & human owner: name the reservations role and the person accountable for the final action.

• O — Operational objective & context: define the booking stage, stay dates, channel, segment and decision deadline.

• T — Trusted inputs & sources of truth: anchor analysis to PMS/CRS, approved channel records, rate plans, contracts, policies and controlled communication logs.

• E — Expected output, exceptions & evidence: require facts, conflicts, unknowns, next actions and closure evidence.

• L — Limits, privacy & leadership approval: prohibit unauthorized changes, commercial promises and unnecessary exposure of guest data.

Reservation source-of-truth matrix

Reservation source-of-truth matrix
Question	Primary approved source	AI role
Is the booking confirmed?	PMS/CRS or approved reservation system	Summarize; flag conflict
What rate/room type was booked?	Confirmed reservation + approved rate plan	Compare; do not invent
What cancellation terms apply?	Confirmed rate/booking terms	Explain; do not rewrite
Is a corporate inclusion valid?	Current approved contract/rate sheet	Validate against booking
Is payment complete?	Approved payment-status workflow	Report supplied status only
Is a request guaranteed?	Confirmed booking/approved entitlement	Separate entitlement from preference
Is inventory available for a change?	Authorized live inventory source	Analyze supplied snapshot; human confirms
Can a fee be waived?	Authority matrix + approved policy	Escalate; do not authorize

Scroll horizontally to read every column. Keyboard: focus the table, then use the arrow keys.

What to learn: Read each row across to connect Question with Primary approved source, AI role. Compare the rows to identify the relevant question for your situation; use the accompanying section to interpret the result.

The word current matters. A screenshot, copied email or last week's inventory can be useful evidence, but it should not automatically override the live controlled system.

The reservations AI control loop

The reservations AI control loop
Stage	Control question	Prompt
Booking quality	Is the file complete and internally consistent?	031
Modification	What changes and approvals does the request trigger?	032
Cancellation	How can confirmed terms be explained accurately?	033
No-show	What evidence exists and what still requires a decision?	034
Inventory pressure	What room-type options can management review?	035
Requests	Which notes are entitlements, preferences or operational needs?	036
Duplication	Which bookings look similar enough for manual review?	037
Shift change	What open reservation work must transfer ownership?	038
Corporate	Does the booking match approved account terms?	039
Exception	What does the authorized decision-maker need to decide?	040

Scroll horizontally to read every column. Keyboard: focus the table, then use the arrow keys.

What to learn: Read each row across to connect Stage with Control question, Prompt. Compare the rows to identify the relevant stage for your situation; use the accompanying section to interpret the result.

Fictional UAE case: one booking, five possible failures

Illustrative scenario

Operational teaching scenario, not a documented incident. Uncited figures and outcomes are illustrative assumptions.

Situation

Use the fictional 60-unit UAE serviced-apartment portfolio from the previous chapters. A corporate booker creates a 21-night reservation for a one-bedroom apartment. The booking notes mention airport pickup, early arrival, breakfast and company billing. A second reservation with a similar guest name appears through an OTA for overlapping dates.

Evidence

The team now has several questions. Is early arrival guaranteed or merely requested? Is breakfast included in the current corporate agreement? Does company billing cover accommodation only or incidentals as well? Is the OTA reservation a duplicate or an intentional booking for another traveler? Is airport pickup actually confirmed with Transport?

A weak AI response might “clean up” the file and conclude that the corporate booking includes all four services and that the OTA record should be cancelled as a duplicate. That would be fast—and operationally reckless.

A controlled workflow produces a verification queue instead:

Fictional reservation case: verification queue
Item	Current evidence	Classification	Next action
21-night 1BR stay	PMS confirmed	CONFIRMED FACT	none
Early arrival	note only	REQUEST / NOT GUARANTEED	check approved availability/process
Breakfast	booking note conflicts with current contract	CONFLICT	Sales/Reservations verify contract
Company billing	account indicator present; scope unclear	UNVERIFIED	Finance/contract owner confirms scope
OTA overlap	similar name/dates	POSSIBLE DUPLICATE	manually compare IDs/source/booker
Airport pickup	request recorded; supplier confirmation absent	UNVERIFIED	verify transport confirmation

Scroll horizontally to read every column. Keyboard: focus the table, then use the arrow keys.

What to learn: Read each row across to connect Item with Current evidence, Classification, Next action. Compare the rows to identify the relevant item for your situation; use the accompanying section to interpret the result.

The difference is important: AI has made the case easier to manage without pretending to possess authority it does not have.

Check the booking IDs, PMS confirmation, current corporate contract, billing scope, early-arrival terms and supplier confirmation for requested airport pickup.

Decision

Decide which items are confirmed, conflicting, unverified or possible duplicates; do not cancel a booking or promise an inclusion without authorised verification.

Action

Route contract conflicts to Reservations and Sales, billing scope to Finance and pickup confirmation to Transport; record the owner and deadline for each pending item.

Verification

Reconcile the final booking record with approved terms and confirmations, retain the duplicate-review evidence and verify the reservation handover before guest communication.

Risk classification in Reservations

Many reservation tasks move from MEDIUM to HIGH risk because they affect money, inventory, contractual terms or a guest's confirmed stay.

MEDIUM examples include extracting requests, preparing a handover or checking file completeness. The output still needs review, but the AI is mainly organizing information.

HIGH examples include modifications, cancellation/no-show interpretation, room-type allocation, duplicate handling and corporate billing terms. A mistake can create a charge dispute, revenue leakage, inventory displacement or a broken guest promise.

The practical rule is that AI can prepare the case, but the authorized employee executes the reservation action in the approved system.

Modification control: compare before you change

A professional modification review should show before, requested, impact, unknown, and approval required. This is more reliable than asking an AI model, “Can we change this booking?”

For example, changing a stay from three nights to five may affect more than dates. It can change availability, rate conditions, package eligibility, corporate terms, tax/fee calculations, room continuity and cancellation conditions. The AI should surface those dependencies rather than answer yes or no from incomplete information.

Cancellation and no-show: explain policy without becoming the policy

Guest-facing clarity is useful, particularly when rate terms are technical. But rewriting a cancellation rule must not change its meaning. The safest workflow supplies the exact confirmed booking/rate terms, relevant timestamps and approved policy, then asks AI to create a plain-language explanation while preserving the original conditions.

No-show handling needs even stronger controls. An AI summary should build a chronology of confirmed facts and evidence gaps. It should not authorize a charge, cancellation or reinstatement merely because a narrative says the guest did not arrive.

Duplicate detection: probability is not proof

Duplicate detection is a strong AI use case because models can compare patterns across names, dates, room types, sources and booking timestamps. But false positives are expensive. A spouse, colleague or family member may have a similar name; a traveler may intentionally hold two rooms; a corporate booker may create related reservations.

Therefore Prompt 037 produces possible duplicate clusters for manual review, never an instruction to cancel.

Special requests: protect the difference between request and entitlement

Reservation notes often contain phrases such as “high floor,” “early check-in,” “airport pickup,” “baby cot,” “quiet room,” “connecting rooms” or “late checkout.” Some may be confirmed paid services, some operational requirements, and some preferences subject to availability.

The AI should classify them rather than flatten them into a checklist of promises. This is also where Reservations connects directly to Front Office, Housekeeping, Engineering, Transport and Guest Experience.

Corporate reservations: contract before memory

Corporate accounts are especially vulnerable to institutional memory: “We normally give this company breakfast,” “they usually have late checkout,” or “Finance knows their billing.” Those statements may be operationally familiar but are not substitutes for the current approved contract or rate sheet.

Prompt 039 forces a contract-to-booking comparison and routes discrepancies to the appropriate owner. This supports Sales and Finance without allowing AI to invent negotiated terms.

Cross-website knowledge connections

Reservations should not live as an isolated knowledge pillar. On the Ahmed Quality Ops website, this chapter should connect contextually to the broader operating system:

• The Tourism AI Operations Playbook — for the wider responsible-AI and tourism prompting architecture: The Tourism AI Operations Playbook

• SOP Exception & Escalation Management — when a reservation moves outside normal policy or frontline authority: SOP Writing Part 4: Exception and Escalation Management

• SOP Roles, Authority & RACI — when the problem is not missing information but unclear decision ownership: SOP Writing Part 3: Roles, Authority and RACI

• Professional SOP Procedure Design — when recurring reservation failures need a controlled process rather than repeated case-by-case fixes: How to Write a Professional SOP: Part 2 — The Procedure

• MICE operations content — for group arrivals, rooming-list dependencies and movement coordination where accommodation connects to event operations.

• Transport/dispatch knowledge — when airport transfers or movement services are attached to a reservation; the reservation record should reference a confirmed transport workflow rather than assume fulfillment.

Future hotel apartment chapters should link back here whenever they rely on reservation truth: Front Office for arrival promises, Revenue for rate/inventory decisions, OTA/Distribution for channel records, Corporate Sales for negotiated terms, Finance for billing controls and Guest Experience for service recovery.

This creates a knowledge network: reservation → arrival → stay → service → exception → evidence → improvement, rather than a set of disconnected articles.

Management KPIs for an AI-supported reservations workflow

AI should be evaluated against operational quality, not the number of generated messages. Useful measures include:

Measures for a controlled reservations workflow
KPI	What it reveals
Reservation files with unresolved critical fields before arrival	upstream data quality
Modification cases returned for missing evidence	process discipline / input quality
Confirmed-vs-requested special-request mismatches	promise-control quality
Possible duplicates reviewed before action	protection against false cancellation
Corporate booking variances detected pre-arrival	contract-control effectiveness
Reservation exceptions with named owner/deadline	accountability
Handover items closed within agreed SLA	continuity between shifts
Repeat reservation-error categories	improvement opportunities

Scroll horizontally to read every column. Keyboard: focus the table, then use the arrow keys.

What to learn: Read each row across to connect KPI with What it reveals. Compare the rows to identify the relevant kpi for your situation; use the accompanying section to interpret the result.

Do not invent benchmark percentages. Establish the property's baseline, define the measurement method, then improve from verified internal data.

Lean Six Sigma application

Reservations is well suited to a DMAIC-style improvement project because many defects are measurable.

Define: choose a defect such as incomplete special-request classification, corporate-rate mismatch or unresolved modification at arrival.

Measure: establish a clean operational definition and collect verified cases.

Analyze: stratify by channel, shift, rate plan, booking source, process step or defect category. Do not assume the AI's thematic summary is the root cause.

Improve: redesign the input fields, verification checkpoint, authority rule, handover or SOP.

Control: monitor the defect and maintain a clear exception path.

AI can accelerate coding, summarization and hypothesis generation, but the improvement team validates the data and causal conclusion.

30-day implementation plan

Days 1–7 — Control the inputs. Define approved sources for booking, rate, cancellation, corporate terms, payment-status indicators and special requests. Remove unnecessary personal data from AI workflows.

Days 8–14 — Pilot low-to-medium risk. Start with Prompt 031, 036 and 038: file quality, request extraction and handover. Compare AI output with supervisor review.

Days 15–21 — Introduce high-risk decision support. Pilot modification, cancellation, duplicate, allocation and corporate validation with mandatory supervisor approval and a documented authority matrix.

Days 22–30 — Measure and standardize. Review recurring errors, update prompt instructions, define KPIs, connect the workflow to SOPs and train the team on what AI is explicitly not authorized to do.

Before → Action → After → value protected

Before: a corporate reservation contains unclear billing, a possible duplicate and an early-arrival note. The next shift reads the notes differently and Front Office discovers the gaps at arrival.

Action: the Reservations team uses the controlled prompts to separate confirmed facts from requests, compare contract terms, flag the possible duplicate, verify the transfer and hand over unresolved decisions with owners.

After: Front Office receives a cleaner file, the commercial questions reach authorized owners before arrival and no booking is cancelled or modified based only on AI inference.

AED saved/protected: calculate only from verified avoided refunds, rate leakage, compensation, chargebacks, room displacement or labor rework. Do not publish an invented savings figure.

Reservations manager checklist

Before operationalizing these prompts, confirm that:

• PMS/CRS remains the booking system of record;

• current rate, policy and contract sources are defined;

• AI cannot execute reservation changes;

• payment-card and identity data are excluded from unapproved tools;

• request versus entitlement labels are standardized;

• duplicate cases require manual verification;

• cancellation/no-show actions follow the authority matrix;

• corporate terms come from current controlled documents;

• handovers preserve open ownership and deadlines;

• final actions are recorded back in the approved system.

Copy-ready working library

#Ten professional prompt templates

Each prompt contains editable placeholders and a built-in stop rule for missing information. Copy it, replace the placeholders with approved facts and keep the verification and human-approval sections intact. The templates are intentionally detailed because the omitted detail is often where a hospitality failure begins.

Prompt 01

#Reservation File Quality Check

#When to use it

Before arrival review, audit a reservation file for missing, conflicting or unverified operational fields. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - PMS/CRS reservation record - channel or direct booking confirmation - approved rate/package terms - payment-status indicator without card data - recorded guest requests

Copy-and-Paste Template

#Complete prompt

H — HOSPITALITY ROLE & HUMAN OWNER
Act as a reservations operations support analyst for a hotel apartment or serviced-apartment property.
Supported role: [Reservations Agent / Supervisor / Manager]
Accountable human owner: Reservations Supervisor

O — OPERATIONAL OBJECTIVE & CONTEXT
Before arrival review, audit a reservation file for missing, conflicting or unverified operational fields.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- PMS/CRS reservation record
- channel or direct booking confirmation
- approved rate/package terms
- payment-status indicator without card data
- recorded guest requests

If records conflict, show the conflict explicitly. If a critical field is missing, stale or unverified, label it UNVERIFIED and state which system/owner must confirm it. Never infer a rate, entitlement, availability, payment status, room status or contractual term from memory.
Minimize personal data. Do not reproduce passport/ID numbers, payment-card data, credentials or unrelated sensitive notes.

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- reservation quality table
- missing/conflicting fields
- verification queue
- owner and deadline

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not mark a field correct merely because it is populated; test consistency across supplied records.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: Medium.
Final operational action requires verification/approval by: Reservations Supervisor.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - reservation quality table - missing/conflicting fields - verification queue - owner and deadline

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 02

#Booking Modification Review

#When to use it

Compare an existing confirmed booking with a requested modification and identify operational, rate, inventory and policy dependencies without executing the change. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - current PMS/CRS booking - original confirmation - requested modification - approved rate/cancellation/modification terms - current authorized inventory/rate information

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Compare an existing confirmed booking with a requested modification and identify operational, rate, inventory and policy dependencies without executing the change.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- current PMS/CRS booking
- original confirmation
- requested modification
- approved rate/cancellation/modification terms
- current authorized inventory/rate information

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- before-versus-requested comparison
- impact/dependency list
- items requiring repricing or approval
- safe response draft

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not execute or imply that the modification is accepted. Flag repricing, availability, penalty, occupancy and entitlement dependencies.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Reservations Supervisor / Revenue or authorized manager.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - before-versus-requested comparison - impact/dependency list - items requiring repricing or approval - safe response draft

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 03

#Cancellation Policy Explanation Draft

#When to use it

Translate the confirmed cancellation terms into clear guest-facing language without changing the contractual meaning or deciding whether a fee should be waived. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - confirmed reservation terms - approved cancellation policy/rate-plan terms - booking channel record - relevant timestamps

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Translate the confirmed cancellation terms into clear guest-facing language without changing the contractual meaning or deciding whether a fee should be waived.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- confirmed reservation terms
- approved cancellation policy/rate-plan terms
- booking channel record
- relevant timestamps

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- plain-language explanation
- confirmed deadline/condition table
- unknowns requiring verification
- neutral response draft

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not calculate or waive a fee unless the supplied approved rule and authorized calculation inputs make the amount deterministic; otherwise route for review.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Reservations Supervisor / authorized commercial role.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - plain-language explanation - confirmed deadline/condition table - unknowns requiring verification - neutral response draft

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 04

#No-Show Review

#When to use it

Structure a no-show case review using confirmed arrival, contact, channel and policy evidence while keeping charge or reinstatement decisions with authorized staff. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - PMS reservation/status - approved no-show policy - channel terms - approved contact log - arrival/transport evidence if relevant

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Structure a no-show case review using confirmed arrival, contact, channel and policy evidence while keeping charge or reinstatement decisions with authorized staff.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- PMS reservation/status
- approved no-show policy
- channel terms
- approved contact log
- arrival/transport evidence if relevant

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- fact chronology
- evidence gaps
- policy checkpoints
- decision items for authorized reviewer

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not classify a guest as a no-show or authorize a charge solely from absence in a narrative. Require the approved operational and policy evidence.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Duty Manager / Reservations Supervisor / Finance as policy requires.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - fact chronology - evidence gaps - policy checkpoints - decision items for authorized reviewer

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 05

#Room-Type Allocation Support

#When to use it

Analyze room-type demand and constraints and propose allocation options without assigning a specific unit, changing inventory or promising an upgrade. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - PMS/CRS room-type bookings - approved inventory snapshot - out-of-order/out-of-service status - confirmed accessibility/operational requirements - approved upgrade/room-move rules

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Analyze room-type demand and constraints and propose allocation options without assigning a specific unit, changing inventory or promising an upgrade.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- PMS/CRS room-type bookings
- approved inventory snapshot
- out-of-order/out-of-service status
- confirmed accessibility/operational requirements
- approved upgrade/room-move rules

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- demand-versus-available view
- constraint list
- allocation options
- escalation triggers

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not assign unit numbers, move inventory, oversell, downgrade, upgrade or displace another booking. Treat output as planning options only.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Rooms Division / Reservations / Revenue authorized owner.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - demand-versus-available view - constraint list - allocation options - escalation triggers

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 06

#Special Request Extraction

#When to use it

Extract guest requests from approved reservation text and classify each as confirmed inclusion, preference/request, operational requirement or unverified item. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - approved reservation notes - booking confirmation - package inclusions - approved guest-request categories - relevant operational policy

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Extract guest requests from approved reservation text and classify each as confirmed inclusion, preference/request, operational requirement or unverified item.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- approved reservation notes
- booking confirmation
- package inclusions
- approved guest-request categories
- relevant operational policy

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- structured request register
- entitlement versus preference labels
- department routing
- verification needs

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Never convert a preference such as high floor, adjacency or early arrival into a guaranteed entitlement unless the approved record says it is guaranteed.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: Medium.
Final operational action requires verification/approval by: Reservations Supervisor / Front Office Supervisor.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - structured request register - entitlement versus preference labels - department routing - verification needs

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 07

#Duplicate Booking Detector

#When to use it

Identify probable duplicate reservations using supplied booking attributes and produce a review queue without cancelling, merging or altering any booking. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - reservation IDs - stay dates - room types - guest/booker identifiers minimized where possible - channel/source - booking timestamps

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Identify probable duplicate reservations using supplied booking attributes and produce a review queue without cancelling, merging or altering any booking.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- reservation IDs
- stay dates
- room types
- guest/booker identifiers minimized where possible
- channel/source
- booking timestamps

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- possible-duplicate clusters
- matching evidence
- confidence rationale
- manual review actions

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Similarity is not proof. Never cancel, merge or contact the guest as a duplicate without human review of the underlying reservations.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Reservations Supervisor.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - possible-duplicate clusters - matching evidence - confidence rationale - manual review actions

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 08

#Reservation Handover Summary

#When to use it

Convert open reservation work into a concise handover showing facts, pending actions, owners, deadlines and escalation conditions. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - open reservation queue - modification/cancellation cases - unresolved channel messages - corporate/group follow-ups - approved action log

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Convert open reservation work into a concise handover showing facts, pending actions, owners, deadlines and escalation conditions.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- open reservation queue
- modification/cancellation cases
- unresolved channel messages
- corporate/group follow-ups
- approved action log

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- priority handover table
- pending decisions
- owner/deadline list
- verification reminders

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not drop unresolved items because they appear low priority. Preserve ownership, next action and evidence required for closure.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: Medium.
Final operational action requires verification/approval by: Outgoing and incoming Reservations Supervisors.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - priority handover table - pending decisions - owner/deadline list - verification reminders

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 09

#Corporate Booking Validation

#When to use it

Validate a corporate reservation against supplied approved account terms without inventing negotiated rates, inclusions, credit status or billing authority. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - PMS/CRS booking - approved corporate contract/rate sheet - authorized billing/credit indicator - approved inclusions - stay details

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Validate a corporate reservation against supplied approved account terms without inventing negotiated rates, inclusions, credit status or billing authority.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- PMS/CRS booking
- approved corporate contract/rate sheet
- authorized billing/credit indicator
- approved inclusions
- stay details

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- contract-to-booking comparison
- variance list
- billing/inclusion checkpoints
- approval queue

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not infer negotiated terms from prior stays or guest statements. Use only the current approved account/contract evidence supplied.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Reservations Supervisor / Sales / Finance authorized owner.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - contract-to-booking comparison - variance list - billing/inclusion checkpoints - approval queue

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Prompt 10

#Reservation Exception Escalation

#When to use it

Turn a reservation exception into a decision-ready escalation brief with evidence, guest impact, options and authority boundary. Property: [hotel apartment / serviced apartment, location] Stay/booking context: [dates, channel, segment, room type, relevant operational context] Decision deadline: [time/date if applicable]

Hotel Apartment control: Use approved property information and a named accountable owner.

#Required inputs

Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log. Required inputs: - reservation record - exception description - approved policy/SOP - contact/action log - inventory/rate/contract evidence relevant to case

Copy-and-Paste Template

#Complete prompt

O — OPERATIONAL OBJECTIVE & CONTEXT
Turn a reservation exception into a decision-ready escalation brief with evidence, guest impact, options and authority boundary.
Property: [hotel apartment / serviced apartment, location]
Stay/booking context: [dates, channel, segment, room type, relevant operational context]
Decision deadline: [time/date if applicable]

T — TRUSTED INPUTS & SOURCES OF TRUTH
Use only the supplied approved records. Primary sources may include PMS/CRS, original booking confirmation, authorized channel/extranet record, approved rate plan, contract/rate sheet, inventory snapshot, payment-status indicator, controlled SOP/policy and documented communication log.
Required inputs:
- reservation record
- exception description
- approved policy/SOP
- contact/action log
- inventory/rate/contract evidence relevant to case

E — EXPECTED OUTPUT, EXCEPTIONS & EVIDENCE
Produce:
- escalation brief
- facts versus unknowns
- guest/operational impact
- options with approval owner

Separate the output into CONFIRMED FACT, REQUEST/CLAIM, UNVERIFIED/CONFLICT, and RECOMMENDED ACTION where relevant. For each open item show owner, next action, deadline/checkpoint, evidence needed for closure and escalation trigger.
Do not choose a commercial remedy outside the stated authority matrix. Present options and route the decision to the authorized role.

L — LIMITS, PRIVACY & LEADERSHIP APPROVAL
Do not create, modify, cancel, reinstate, merge, allocate, charge, refund, waive, upgrade or confirm a reservation. Do not promise availability, price, benefit, room type, early check-in, late checkout, compensation or exception unless an approved source and authorized human decision explicitly support it.
Risk classification: High.
Final operational action requires verification/approval by: Reservations Manager / Duty Manager / authorized commercial owner.
AI output is decision support and a draft—not the PMS/CRS record, contractual confirmation or authorization.

#Expected output

Produce: - escalation brief - facts versus unknowns - guest/operational impact - options with approval owner

Human verification

Professional hotel apartment trainer’s note

Replace every placeholder and keep unresolved items visible until an authorised hotel apartment professional reviews the output.

Final human-verification checklist

Before using any AI-generated reservation output operationally, verify:

1. Is the reservation identity/reference correct?

2. Is the PMS/CRS record current?

3. Are the rate, room type, stay dates and occupancy confirmed?

4. Are requests clearly separated from guaranteed inclusions?

5. Are cancellation/modification/no-show terms taken from the correct approved source?

6. Are corporate terms current and applicable to this booking?

7. Are payment or billing statements limited to approved status information?

8. Are duplicate indicators manually verified?

9. Does the employee have authority for the proposed action?

10. Has the final approved action been recorded in the proper system?

Key takeaway

Reservations AI is most valuable before a problem reaches the guest. Its job is to make the booking clearer, more complete, more traceable and easier to govern. It should not become a shadow reservation system.

Use AI to find the question that still needs an answer—not to manufacture the answer that the system or authorized person has not yet confirmed.

Connect this guide to your next step

Continue with related methods and the wider series.

Put this article into practice

Use the prompts to review booking quality, modifications, cancellations and allocation exceptions before confirmation.
Verify the supplied facts and record any missing evidence before acting.
Name the accountable owner and confirm the decision or approval limits.
Test the result in the operating context and retain the verification record.

Your next step

Supply the required inputs, copy the complete prompt and verify the output before using it in the operation.

Apply Reservation File Quality Check