A hotel promotion can surface in a conversation, a telephone call and a room search. Paraty Tech has placed the information shared by those interactions at the centre of its artificial intelligence proposition: Paraty AI Knowledge. Its dedicated June 23 presentation explains how a structured knowledge base for the property supplies five tools in its suite, from webchat to the booking engine and pricing intelligence.
The Spanish company addresses a specific commercial problem: keeping the offer consistent when a traveller asks about it at different points in the direct channel. The hotel initially supplies and supervises the information, supported by Paraty's team. The provider then describes a shared source that distributes knowledge and is updated through interactions passing through the connected tools.
The proposition builds on components that the company had explained during March and May. June's presentation develops the relationship among them and the role of their common source. For hotel marketing, the focus is on organising the information that helps someone choose, the conversation that resolves a question and the process that presents a bookable option as connected parts of the purchase journey.
The hotel has a role in what its assistants know
Paraty uses two everyday examples to explain the arrangement: changing breakfast hours and launching a promotion. In its description, the update enters the common knowledge base and the tools work with that context. The property therefore participates in the information accompanying a guest considering a stay, covering both hotel services and the commercial offer.
The starting point is structured, centralised information. The company links separate repositories for chat, voice and booking with different answers about the same hotel. AI Knowledge brings that information together in a node connecting the tools. This concerns how commercial content is managed: a policy, service or promotion can inform several interfaces serving the same purchase intention.
Initial validation belongs to the hotelier, with support from Paraty's team. Property knowledge includes decisions particular to that business: how a room is described, the conditions attached to a promotion and which services form part of the experience. The provider proposes making this content a common reference for systems using it during the relationship with the traveller.
The presentation also describes information travelling in the other direction. Conversations, queries and searches contribute to updating the source, within what Paraty calls a dynamic, bidirectional structure. What a customer receives and what the customer reveals by asking belong to the same product design. A question expresses something the traveller wants to resolve; the answer connects that need with the hotel's available knowledge.
From an enquiry to a search containing dates
Webchat illustrates the connection between information and selling. Its May 25 presentation describes assistance available around the clock in seven languages, covering rooms, prices, services, policies and promotions. It also describes recommendations adapted to the traveller's profile and recognition of returning guests. Appearance can follow the property's colours, wording, tone and other brand elements.
The commercial function becomes visible when a conversation needs to move from an informational answer to a particular search. Paraty's published example concerns a double room for specified dates. The assistant checks availability in real time, identifies options and updated rates, then supplies a link with the search already loaded. The traveller can continue from the question already expressed, with part of the work carried into the next screen.
The provider identifies MCP, Model Context Protocol, as the protocol connecting webchat with the engine. That integration gives the conversation a defined commercial destination: identify an option at the property and prepare access to booking. Knowledge and availability perform complementary jobs. The shared source supplies context about the establishment, while the engine connection provides bookable options for the request being handled.
The journey includes human intervention. Paraty describes chat as a triage system handling informational conversations and escalating qualified purchase intention or a situation requiring a person. Ring2Travel , its contact centre, provides specialist booking agents, with continuous support in five languages according to the presentation. The design distributes the conversation between automated assistance and the commercial team.
An initial account of 130 conversations
In its May communication, Paraty reported an early balance: of the first 130 real conversations handled, 97% were resolved within chat and 3% required human intervention. The company also reported 540 minutes saved. The percentages describe how that initial set of conversations was handled; the time figure is the operational saving the provider attributes to the experience.
How the first 130 webchat conversations were handled
Distribution reported by Paraty Tech in its May 25, 2026 presentation of Paraty AI Webchat.
% of 130 conversations
Tap or hover for details. All values are also available in the data table.
Collapse notes
Preliminary provider account communicated in May; these are handling outcomes, not booking percentages. The publication date does not establish the conversations' exact measurement period.
Explore the data table
| Category | Share of the initial set |
|---|---|
| Resolved within chat | 97 % of 130 conversations |
| Required human intervention | 3 % of 130 conversations |
The distribution shows which part of the interaction remained in chat and which passed to the team. They are two outcomes within the same base of 130 conversations. Their relationship with selling lies in the tool's intended journey: answering questions, recognising when an availability search is needed and helping the traveller move to the engine or an agent. A resolved conversation and a confirmed booking are different measures within that journey.
The example places information work within a commercial operation. Hours, services, conditions and room types can occupy part of a discussion before someone is ready to book. The proposition combines an answer available at the time of the enquiry with a route towards specific options. May's reported balance offers early context for this webchat function inside the suite described together in June.
A telephone call provides another commercial entry point
Paraty AI Voice's March 12 presentation documents the same division of work on the telephone. The system analyses the caller's needs, answers informational questions and identifies purchase opportunities. The provider describes storing qualified leads in the Leads Panel or the hotel's CRM and escalating the conversation to an agent when there is conversion potential. The call can continue into the property's commercial work through that record.
The shared source also appears in how the assistant handles a question it cannot resolve. Paraty describes transferring to a person and using that conversation to inform the knowledge supporting future calls. Assistance captures information about needs expressed by a guest, while human intervention retains a role in resolving the situation and continuing work on the opportunity.
The Ring2Travel connection places that process within an operation including incoming calls, pre-stay contact and recovery of lost demand. The company explains that automated handling of informational enquiries gives agents room for follow-up and outbound calls to registered opportunities. This provides another specific destination within the journey: an enquiry produces an answer or a referral and, where commercial intention exists, can leave a contact for the team to work with later.
Booking becomes a configurable journey
The relationship between AI Knowledge and purchasing also appears in Paraty AI Bookings, presented on March 24. The engine includes an Intelligent Planner: a button opens a window containing a text field and optional filters for describing a trip in natural language. The system interprets the request, consults the knowledge base and returns a hotel proposal in an editable shopping cart, according to the company.
The editable cart is a specific part of that mechanism. The initial conversation expresses intention; subsequent selection allows the traveller to review the proposal before proceeding. Knowledge about the property enters a stage concerned with choosing products and conditions. The presentation connects how people ask with how they buy, inside the engine the hotel uses for direct sales.
Bookings also introduces what Paraty calls Intelligent Funnels, booking journeys a property can create and activate while retaining a default process. Configuration considers traffic source, origin market, device, guest segment, price range, search dates and loyalty membership. These criteria allow the journey to reflect the context in which a visitor arrives and the type of stay being considered.
The hotel has discretion over adopting those journeys. The presentation keeps the baseline process operating whether additional funnels are created or not; the property can introduce the new options when it is ready. It can also activate or deactivate processes and compare them in A/B mode. The shared knowledge source sits alongside a commercial configuration retaining a starting point and allowing variants to be tried, rather than requiring every booking journey to change at once.
The company provides two examples: a visitor from a metasearch engine might encounter a more direct process, while a returning customer might receive a journey with more personalisation and opportunities to purchase improvements. Distinguishing these contexts connects acquisition with conversion. The channel bringing the guest and the guest's previous relationship with the hotel become inputs to the purchase journey's design.
The presentation includes A/B comparisons between processes and analysis of funnel steps. According to the provider, AI can monitor their performance and optimise distribution of the journeys. The hotel also has independent desktop and mobile versions, collapsible cards, guided navigation and a visible booking summary. These are defined changes to the interface and process management accompanying conversational search.
Another component is selling by attributes, described as micro-upgrades: views, position, floor or orientation can appear as progressive improvements during booking. In the March text, Paraty reports a 12% increase in acceptance of these improvements and a contribution to raising ADR by between 4% and 8%. These are the figures it attributes to that selling approach; the ADR range and improvement acceptance concern different measures.
AI Knowledge's role becomes clearer through this variety of decisions. Property information helps interpret the traveller's request; the engine organises options and the journey used to examine them. June's presentation brings those dimensions under a common source, alongside the other tools, so the explanation of the offer can accompany a person moving between interfaces.
The planning and purchase stages also have different controls. A traveller can describe preferences in the open text field, refine them through optional filters and then edit the proposal. The hotel can manage the journeys offered to particular contexts and compare their performance. Paraty's descriptions place those choices on both sides of the transaction: the property shapes the process and the traveller reviews the resulting offer before continuing.
Commercial information also returns to the team
The suite provides another destination for data: the hotel's own decisions. Paraty Business AI, described on May 8, is integrated into Paraty Manager and uses the usual credentials. The tool supports natural-language questions, automated summaries on the dashboard and reports delivered through WhatsApp or email. Those summaries update as filters change, according to the provider's explanation.
Personalisation includes filters remembered between sessions, hotel groups configured once and individual views within Mi Panel. That arrangement supports people reviewing several properties or returning to the same combination of criteria. Information can be examined within the platform or exported to PDF and Excel for further work outside it.
The analysis catalogue is specific to hotels: bookings already on the books, weekly pickup, sales pace, rate parity and markets by country of origin. The company distinguishes booking windows from stay windows, two ways of organising information over time. It also includes digital marketing monitoring and modules covering quotations, leads, funnels, acquisition, devices, attribution, loyalty and the purchase of improvements.
These distinctions help explain the commercial journey. Origin identifies how a customer arrived; a funnel shows the steps followed; a quotation or lead records another contact point. The tool brings these views to the team managing direct sales. The shared-source proposition therefore covers answers provided to travellers and information helping the property organise its response.
Business AI also describes availability problems that have a specific meaning during purchasing: searches not satisfied because of capacity, minimum-stay restrictions or an incorrect promotional code. Each cause requires a different interpretation by the hotel. Distinguishing them helps locate where a query encounters a condition preventing the requested option from appearing within the engine.
The company describes modules for cancellations, demand and dynamic packages, alongside predictive features and an anonymous mode hiding hotel names when figures are shared. The attention moves from having many indicators towards helping the team consult them in context. Within AI Suite, this analytical role sits alongside conversational assistance and the selection of a stay.
The difference between booking and stay windows is particularly relevant to that work. A sale enters the system at one moment and concerns accommodation on other dates. The presentation makes both views available for origin markets, alongside sales pace and weekly pickup. This lets a team examine how demand is entering the direct channel while retaining the timing of the accommodation it concerns, rather than reducing all activity to the date a booking was made.
Five tools around one reference
June's presentation names five components: Paraty AI Webchat, Paraty AI Voice, Paraty AI Bookings, Paraty Business AI and Price Seeker AInsights. Voice handles triage in the telephone channel; Price Seeker is identified as a rate shopper and pricing-intelligence platform. Knowledge's position is to supply a common reference to distinct functions, each with its own interface and moment of use.
Paraty arranges the journey into three moments: inspiration and search; reinforcement through voice and chat; and decision and booking. This describes the journey the suite is designed to serve. A conversation can start as a question and become a search; a booking generates information also relevant to people reviewing demand and prices. Property knowledge provides a connection between these activities.
The company links this design with travellers using assistants and search engines based on language models. Its declared goal is for the property to have centralised, consistent information that such interfaces can interpret. June's presentation concentrates on the role of that common source, while the March and May communications show its place in booking, analytics and assistance.
Who's who and how to read the figures
Paraty Tech is the brand of Paraty Hoteles S.L.U., identified in its legal notice with tax number B93165090 and registration in Málaga's Mercantile Registry. Paraty AI Suite brings together the named tools; Paraty AI Knowledge is their shared knowledge base. Ring2Travel supplies the human contact centre. MCP is the protocol the provider identifies for connecting webchat and engine; LLM means a large language model.
June 23 is the date of Knowledge's dedicated presentation. Bookings and Business AI details come from March and May, and the balance of 130 conversations was communicated on May 25. The 97% and 3% distribute the handling of that initial set; 540 minutes is the reported operational saving. ADR means average daily rate: its 4%–8% range and the 12% improvement-acceptance increase belong to the micro-upgrade explanation, with different bases from the chat conversations.
In this article
Marian Vasile
Marian Vasile is a MarTech and growth professional. His work spans product marketing, SEO, paid advertising and deep online tracking, including behavioral analytics and attribution.
