Words thedesk and theagents share.

AI that can propose multi-step hotel work — with policy and confirm.

In NISKA, Agentic AI means specialist agents that reason over the property model, draft actions, and help staff execute. It is not a chatbot that invents answers outside hotel truth.

NISKA’s hotel management system built around agents and one property record.

The product line: PMS, CRS, channel, revenue, guest, and ops surfaces sharing one hospitality database, with confirm gates on high-impact writes.

Also: HMS

Availability, Rates, and Restrictions — the sellable inventory signal.

Canonical ARI is what distribution and channel partners consume. Nightly and hourly stay modes stay separate planes; channels do not invent a second book.

Also: Availability · Rates · Restrictions

A durable log of who proposed, confirmed, or declined an action.

Every agent recommendation and operator confirm should leave a trail. Useful for night audit, disputes, and training — not just compliance theatre.

The hotel’s direct booking path on the website or app.

Direct demand that must share the same inventory and rates as OTAs — otherwise the desk fights two truths.

The control plane that pushes ARI and pulls bookings across OTAs.

In NISKA, distribution owns mapping and sync. Agents may watch connectivity; they do not invent inventory outside the property model.

Also: CM

Stay start and end events that change occupancy and folio state.

High-visibility desk moments. Agents can prepare, identity connectors can capture, but stay status still writes through PMS rules.

A standard envelope for cross-module business events.

When a reservation is created or ARI changes, other services consume events — they do not write into another domain’s database.

Guest-facing Agentic AI on the booking and stay path.

NISKA Concierge helps guests with questions and requests while staying on property truth. High-impact changes still wait for confirm when policy requires it.

Human approval before a high-impact write goes live.

Agents may draft rate changes, guest replies, or folio posts. Money, inventory, and guest-impacting writes stay behind an operator confirm until policy says otherwise.

Also: Confirm-gated

Central reservation system — group and multi-property booking control.

Inventory, rate plans, and availability upstream of channels. In NISKA it sits on the same property model as PMS — not a parallel book.

Also: Central reservation system

Sales and marketing agent for visibility, schema, and AEO briefs.

Works from property truth — not invented metrics — so public claims stay honest.

India’s Digital Personal Data Protection Act — guest data handling lens.

NISKA designs India-first data handling with DPDP awareness: consent, purpose, and property-scoped access — not a locale sticker after the fact.

The guest’s bill of charges and payments for a stay.

Finance owns posting. Connectors and agents may propose charges; settlement should not create a shadow ledger outside the folio.

Inventory held for a group or event against the hotel’s book.

Sales and events tools must respect PMS-owned inventory. A parallel banquet book is how overbooking starts.

Tax treatment designed for India GST, not a generic locale overlay.

Billing, invoices, and exports should understand GST rules at the desk — especially for India GTM properties.

FAQ deflection for guests — not confirm-gated action on the hotel book.

A hotel chatbot answers common questions. It is not the same as Agentic AI Concierge, which can propose stay actions under confirm gates on one property model.

Also: Chatbot · FAQ bot

Operators remain decision owners for risky or brand-sensitive actions.

Confirm gates are the product expression of human-in-the-loop. Autonomy grows only after policy and eval say it is safe.

The service boundary for non-CM partners, gateways, locks, and webhooks.

Adapters land here. Partners do not share write databases with PMS or finance. See Marketplace for connect lanes.

Sellable room capacity by date, type, and stay mode.

One book across OTAs and direct. Channel projections may lag; they must not invent rooms the PMS does not own.

The end-of-day close that rolls the hotel’s financial and occupancy day.

Agent writes that affect folios, rates, or occupancy must respect audit boundaries — confirm gates exist partly so night audit stays coherent.

Online travel agency — a distribution channel for hotel inventory.

Bookings and ARI sync through distribution. Guest messaging and reputation still resolve to the same reservation record.

Also: Online travel agency

Transactional pattern that publishes events after a successful write.

Keeps domain writes and event fan-out honest — partners consume what actually committed, not a hopeful side queue.

Property management system — reservations, stay, assignment, and desk ops.

In NISKA, PMS Core stays the source of truth for reservations and stay state. Agents advise on that record; they do not fork it.

Also: Property management system

The single hospitality record agents and modules share.

Reservations, rates, inventory, guest context, and folio live as one truth. Product modules may differ in UI; they should not invent a second database of record.

Also: Property truth · One book

A priced offering tied to rooms, rules, and distribution.

Sell rates and restrictions attach to plans. Revenue agents may recommend changes; live sell rates move only after confirm when gated.

Competitive rate intelligence against a defined competitor set.

Feeds recommendations into revenue workflows. Shopping data informs; it does not silently overwrite ARI.

Review ingest, sentiment, and response workflows tied to stays.

Agents may draft replies from stay truth. Publishing a response stays confirm-gated so the brand voice remains human-owned.

Sell controls such as CTA, CTD, min/max stay, and closed to arrival.

Part of ARI. Restrictions protect inventory strategy; channel sync must preserve them, not drop them silently.

Revenue management system — pricing, forecast, and yield recommendations.

NISKA revenue intelligence recommends against canonical ARI. Confirm gates keep pricing autonomy with the operator.

Also: Revenue management system

Matching a reservation to a physical room.

Front-office and ops agents may suggest assignment. The PMS stay record remains authoritative for who is in which room.

The owning service for a domain fact — others consume, not overwrite.

PMS owns reservations; revenue owns canonical ARI publish; finance owns folio posting. Marketplace connects project onto those owners.

Also: SoT

An agent scoped to a hotel desk — revenue, channel, ops, reputation, and more.

Each agent has a job and tools. They share the property model and confirm policy; they are not a single generic chatbot with hotel stickers.

Access and data bounded by organisation and property.

RBAC and ABAC keep agents, staff, and connectors inside the hotels they are allowed to see. No cross-tenant leakage by convenience.

Also: RBAC · ABAC

See Outbox — commit the business write, then publish the event.

Preferred pattern for ReservationCreated, AriChanged, and similar hotel events so consumers never race an incomplete write.

Guest messaging on the channel Indian hotels already run.

Templates and conversations sit on the same reservation and profile record. Messaging partners attach through Integration Hub — not a side CRM.

Optimising rate and availability mix for revenue and occupancy goals.

Revenue agents support yield decisions with recommendations and guardrails. Final live rates remain confirm-gated when policy requires.