AI features in short-term-rental software Contents

Hostify

AI features in short-term-rental software

What the competitors ship in November 2026, where Hostify stands, and which features would be new for the market.

November 2026

Laid out for A4 as well: Print, "Save as PDF", paper A4, background graphics on.

1 Summary

This report answers three questions. What do the competitors ship in AI today? Where does Hostify stand against that? Which features would be new, for Hostify and for the market?

It starts from the four pages the CTO named (RentalReady Maia Insights, Guesty AI, Hospitable AI, Boom AI review) and widens to the rest of the field: about twenty-five further PMS vendors, about forty point tools, the booking platforms, and eight neighbouring industries. It builds on the analysis made for the mobile app on 2026-10-09 and records what changed in the five weeks since.

1.1 What the market looks like

  1. Guest messaging no longer sets vendors apart. Every serious vendor has drafts and an autopilot with a confidence gate, a refusal list and a stated reason when it stays silent. The competition has moved to price: Hostaway includes it, Guesty charges $39 to $59 per listing per month for its AI package, Hospitable keeps autopilot on its top plan.
  2. The new front is the back office. Guesty documents 57 agents that audit taxes, fees, owner ledgers, calendars and team access, and a few that move money on their own. Boom runs work orders under a spend threshold. RentalReady applies bulk price changes from a sentence, after approval.
  3. An MCP server became standard within one quarter. Hospitable, OwnerRez, Uplisting, Hostex, Zeevou, RentalReady and every revenue tool now let the customer's own assistant read the account, and most let it write. Hostify has a server in beta that it has not announced.
  4. Voice arrived as a product line, in point tools and on the platforms, not yet inside any PMS.
  5. The platforms now answer the guest before the host can. Airbnb, Booking.com and Expedia answer pre-booking questions from the structured listing data, and the host does not see the question.

1.2 Where Hostify stands

Hostify has more AI than it shows. The autopilot is live and answers about half of the conversations it sees. A knowledge base is mined from past conversations, which no competitor in this research documents. A reporting assistant and an MCP server with about 38 tools exist but are not released. ID documents are read at check-in. Three staff agents work in support.

The public picture is thinner: two help articles from April, an autopilot still labelled beta, and a home page that does not mention AI. Outside messaging, Hostify has no released feature in reviews, tasks, listings, finance or owner communication.

1.3 What we propose

Three groups, detailed in the last chapters:

  • Release what is built (the MCP server, the reporting assistant) and close four parity gaps that customers now expect: review replies, tasks from messages, urgency, and the AI disclosure line that EU law requires on automated replies.
  • Fast-follow the back office, starting with the checks only a channel manager can run across fourteen channels.
  • Build what nobody has. The research found no PMS with: a simulation gate that tests the guest agent before each change; a quality score on every conversation; one knowledge base that feeds and audits the structured data on every channel; an "explain this" agent over the audit logs; a migration agent that reads a competitor account; an expense agent that reads the owner contract as policy; a damage file from photographs; a receipt reader with a VAT check.

1.4 How to read the evidence

Every number about a competitor is that vendor's own claim unless the text says otherwise. No vendor publishes accuracy. Facts taken from a search excerpt and not from the page itself are marked in the research files. "Not found" means this research did not find it, on 2026-11-16, in public sources; it is not proof of absence.

2 How the market moved

Five shifts since the review of 2026-10-09. Each is followed by what it means for Hostify.

2.1 Messaging became a commodity with a shared design

The design is now the same everywhere. A draft tier first. An autopilot gated by a confidence threshold, a topic list or a written rule. Scope per listing, booking stage and hours. A reply delay so a person can answer first. A refusal list in words: money, date changes, access details before confirmation. A reason recorded whenever the AI stays silent.

What differs is the control around it:

ControlWho has itWhat it does
SandboxHostaway, OwnerRez, Akia (voice)Test a reply against a listing and a real reservation before switching on
Context panelHostawayThe user sees every fact the AI was given
Planned-reply bannerGuesty (pilot)Countdown before the automatic send, with edit, send now and delete
Second safety passHospitableA separate model reviews every automatic reply before it leaves
Policy tiers with a logBoomMay answer, must ask, never; each send is logged against the policy that allowed it
State gatingBoom, HospitableDoor codes only after the clean is signed off, or after the guest finished verification and deposit
Stand-downRentalReadyThe AI does not reply while a person is drafting or has taken the thread
Disclosure lineHostaway, Hospitable, GuestyHostaway: "Some messages may be sent using AI", forced on for EU listings. Hospitable: an optional attribution text. Guesty: "This message was written by AI" appended to the reply

For Hostify: the autopilot has the core and lacks most of this surrounding control. The disclosure line is a legal matter. Hostaway cites Article 50 of the EU AI Act and Smily cites a duty from 2026-08-01; our legal reading should be confirmed before the autopilot leaves beta in Europe.

2.2 The back office is the new front

Guesty is the reference. Its help centre documents 57 agents, the first dated 2026-06-25 and 13 added since October. Of these, 28 only report, 12 act after an approval card that expires, and 17 act alone.

What the agents do falls into four kinds:

  • Configuration audits: tax set-up compared across properties in one city, fee drift between a multi-unit parent and its sub-units, listings with no owner, team permissions that look unusual.
  • Money checks: unpaid reservations, owner ledgers that will come out negative, journal entries outside the stay, deposits left unapplied, a month-end workspace that counts the days to close.
  • Operational watch: cleanings that clash with stays, calendar blocks nobody can explain, promises staff made in chat and did not keep, double bookings with a recommendation.
  • Actions: charging the stored card for a checkout balance, re-creating a card hold, publishing replies to reviews rated 4.5 and above, blocking the calendar for a repair.

The marketing page says "hundreds of agents". The documentation supports 57, and about 16 names on the marketing page have no article. Guesty's terms put the liability for a wrong automatic reply on the customer.

For Hostify: this is where the vendors now add features fastest, and it is cheaper to build than messaging was. Most of these agents are a query, a rule and a report. The ones only a channel manager can do well are named in the chapter on new features.

2.3 MCP became standard

VendorStatusWrite accessPrice
HospitableLive since April 2026, in the Claude directory since 2026-11-13Yes, about 30 write tools including smart locksAll paid plans from $29
OwnerRezAll accounts since October 2026Yes, defaultFree
UplistingDocumented August 2026Yes, 15 tools with a scope per permissionAll plans
HostexJuly 2026YesNot stated
ZeevouJune 2026Yes£2 per unit
RentalReadyJune 2026Yes, with approval pointsShared credits
LodgifyBeta, September 2026Read-only, 12 toolsAll plans
GuestyExperimental betaRead-onlyNot stated
HostawayNone officialn/an/a
PriceLabs, Beyond, Wheelhouse, AirDNA2026Yes for pricing, with floors, ceilings and approve-before-pushMostly included
HostifyBeta, about 38 tools, not announcedInbox reply and note, calendarn/a

Two designs stand out. Uplisting publishes a scope per permission (read properties, create bookings, write calendar, send messages). Hospitable documents no such scoping while exposing lock control. The first is the model to copy.

For Hostify: the server exists. Releasing it with scoped permissions and the dedicated agent key already planned is the cheapest move in this report.

2.4 Voice became a product line

Akia's phone agent books a room and takes payment during the call by texting a payment link. HelloShift quotes live availability by voice. ProhostAI sells an AI phone add-on from $30 a month. Conduit includes voice in its $18 per unit plan. Airbnb and Vrbo run AI voice support. None of the PMS vendors in this research ships a working voice agent of its own; Boom claimed one in 2024 and no 2026 page shows it.

For Hostify: the voice concierge proposal (P6) is still ahead of every PMS. Point tools already sell it, so the lead will not last long.

2.5 The platforms moved into the host's territory

Airbnb now answers pre-booking questions on the listing page from the structured data, chooses which cover photo each searcher sees, writes its own summary of the listing in search results, summarises the whole review history into themes, and since October 2026 offers its own dynamic pricing. About 40% of its support issues were resolved without a person in the third quarter of 2026, by Airbnb's own figure. Booking.com and Expedia have similar pre-booking answers.

Two rules bind every vendor. Airbnb accepts no AI-generated or AI-modified evidence in a damage claim. Hosts may use AI messaging only through approved software partners.

Agentic booking exists as infrastructure, and the demand for it is still small. Google AI Mode completes hotel bookings with ten partners in the United States and does not include vacation rentals. OpenAI withdrew in-chat checkout in March 2026. Booking Holdings reports that referrals from language models are under 1% of room nights.

For Hostify: the completeness of the structured listing data on each channel is now a distribution requirement, because the platform's AI answers from it and the host cannot correct the answer. A channel manager is the one place that sees all channels at once.

3 The vendors, one by one

The four vendors the CTO named come first and in depth. The rest follow in short.

3.1 Guesty

What is shipped. Four layers: inbox AI (ReplyAI drafts, the AutoReply agent, task suggestions from messages), the AI hub with 57 agents, Copilot (a read-only analytics chat that keeps no conversation history), and revenue AI (PriceOptimizer, a revenue centre with recommendation tiles). An MCP server is in an experimental read-only beta.

How the hub works. A catalogue of cards by category. Each agent is switched on by itself and runs on a schedule Guesty fixed, or on demand. Results land in one activity feed with a status and a severity. Agents that act show an approval card with Approve, Edit and Decline; the card expires after a window that differs per agent, from 55 minutes to 7 days, and an expired card changes nothing. Only admins see the hub unless a role is granted access.

What is unusual.

  • Agents that act on money alone: charge the stored card after checkout, re-create a deposit hold, clear a channel deposit.
  • Public replies to reviews rated 4.5 and above are published without approval; mid ratings get a draft; low ratings get a task and no reply.
  • A tracker that reads staff messages for promises made to guests and follows whether they were kept.
  • A challenger that checks one to three star reviews against the channel's policy and files a removal request after approval.
  • Vision on listing photos: amenities and rooms seen in the photos and missing from the listing.
  • A month-end workspace and a reservations workspace that shows "revenue at risk" and which reservations no active agent covers.

Design rules worth copying. A failed or partial scan is never reported as "all clear". State is re-read just before acting and read back after. Every run has a hard cap on items. Structured property fields win over the knowledge base when the two disagree.

Price. The Guesty AI add-on (hub, AutoReply, PriceOptimizer) costs $59 per listing per month up to 50 listings, falling to $39 above 500. Drafts remain free.

Weak points. The agents cannot be scoped per listing or rescheduled. There is no builder and no marketplace, only a "Request an agent" button. Five agent articles contradict themselves. Adoption figures and user complaints were not found.

3.2 Hospitable

The landing page the CTO named renders only in a browser and returned nothing to the fetcher on three attempts; the facts come from the help centre and the public changelog.

What is shipped. Draft and improve in the inbox (10 a week on the free plan). AI Auto Reply on the top plan (Mogul, $99 plus $30 per property). A Knowledge Hub generated in under 30 seconds from the listing, past messages and PDFs. A guest summary built from reviews by and about the guest. Review reply drafts, and host-to-guest reviews with optional automatic publishing. Copilot, a portfolio chat with charts, read-only. Its own dynamic pricing. An Opportunity Center with revenue prompts.

The MCP server is the widest write surface found: create and cancel reservations, change calendar prices, send guest messages at once, post reviews, edit the knowledge base, publish owner statements, lock and unlock doors. The vendor's own text calls unlocking "like handing someone your keys". No per-tool permission is documented.

Auto Reply in detail. The longest explicit refusal list of any vendor. Check-in details are withheld until the guest finished verification, agreement and deposit. Since January 2026 it offers the host's upsells on its own and can read deposit, insurance and the financial breakdown of a reservation. Every draft passes a second AI safety review, which is why replies take one to two minutes. The reason for every skip is written into the conversation notes.

Weak points. The Knowledge Hub reads the listing and the message history once and cannot refresh them. Users report the AI promising actions the host cannot deliver and sending outdated third-party contacts. Trustpilot complaints concern the support bot, not the inbox AI.

3.3 Boom

Positioning. "The AI that runs the whole operation": one agentic layer (BAM) with six named agents for customer service, sales, reviews, operations, owners and distribution.

What is unusual.

  • Four audiences, four data scopes. Guests see their reservation. Prospects get live availability and price. Owners ask about their own units and statements. Contractors ask about their assigned jobs on WhatsApp.
  • Policies in three tiers. May answer, must ask, never. Refunds are drafted with an amount for a person to approve. The sales agent "can hold and quote but cannot discount past your floor".
  • Explainability. The reasoning, the policy or memory used and the confidence score are visible per reply. Rejecting a draft requires a reason.
  • Spend authority. A work order is created, assigned to a vendor and approved automatically under a threshold the operator sets; the example shows a $140 repair under a $200 limit.
  • Alerts with the action attached. "Miami pace is +18% against the comp set. I've queued weekend rate lifts on 6 units. Approve?"
  • Reviews. One inbox across Airbnb, Booking.com, Vrbo and Google; replies drafted in the property's voice; complaints become tasks; issues are caught mid-stay before they reach a public review.

Price. Boom Host, for up to 19 listings, is $30 per listing per month with three autonomy modes. The full product is by quote. The terms disclaim liability for AI actions and reserve customer data for model training.

Weak points. Most figures on the product pages are illustrative screenshots. Trustpilot shows 36 reviews with start-up troubles and channel connection faults. What is AI in the accounting module is not explained.

3.4 RentalReady

Maia Insights is a chat over the portfolio with two things the others lack. It benchmarks against the local market with AirDNA data. And it applies bulk changes from a sentence or a CSV: prices, promotions, payout adjustments with the right VAT rate, calendar blocks, missions, listing rewrites. It lists every change and waits for approval; bulk rights are granted per user. Reports can be scheduled by email. It runs in the web product, in the staff mobile app and through MCP.

Maia Messaging is a paid add-on. It extends a stay by checking live availability and sending a payment link, follows up an enquiry until payment, creates tickets from messages and reviews with an audit page, and marks each conversation Urgent, Maintenance, Finance or Escalate.

In the standard plans: guest mood detection, a property audit that groups repeated complaints from reviews and private feedback, review reply generation, listing texts, risk scoring.

Price. Insights, the agents and the MCP connection draw on one pool of credits with a free monthly allowance. Neither the allowance nor the credit price is published.

Weak points. The vendor's own blog gives two automation rates that contradict each other. "AI agents" are named as a credit consumer and described nowhere.

3.5 The others, in short

VendorWhat matters
HostawayEverything included at no add-on fee. Rules in plain language, each set to draft or auto-reply. A sandbox and a context panel. A knowledge base that learns from every host reply. CoHost, a portfolio chat that since September 2026 also prepares messages, tasks, review replies, calendar blocks and direct bookings behind a confirmation card. The EU disclosure line. No MCP
LodgifyWeakest delivery against its marketing: the Co-Host is still a limited beta that drafts replies and answers questions. Read-only MCP on every plan. Direct booking sites prepared for AI search, and a ChatGPT matching pilot
OwnerRezRezzy at $9 per property: three modes, opens and also closes tasks, logs why it did not reply, validators against timing and outcome promises. A free MCP server with write access by default
UplistingConfidence-gated sending, a knowledge centre extracted from manuals, and the cleanest MCP permission model
StreamlineLeo: a listing agent, an insights agent pitched for owner retention, and a service that optimises how the direct site appears in AI search
Apaleo, Mews, Cloudbeds (hotels)Copilots trained on the operator's own procedures that act on the PMS; automations built from a plain-language description; a migration from another PMS done through MCP "in a few hours" in one customer case
Avantio, Rentals United, Beds24, Kross Booking, SmoobuNo native AI beyond drafts. The European incumbents are behind

3.6 The point tools

  • Conduit ($18 to $48 per unit): pre-built agents for guest services, sales, housekeeping and owner communication, with voice included.
  • Besty moved from a tool to a managed service paid as a share of revenue, with a minimum of $500,000 in annual guest revenue.
  • Properly reviews turnover photos with AI at $5 per inspection; a person verifies every flag.
  • PriceLabs Athena is an analyst that runs on a schedule without a prompt and emails recommendations to accept.
  • Autohost screens guests with rules written in plain English and returns a reasoned recommendation.
  • Minut runs models on the sensor: cigarette or cannabis smoke told apart from cooking, crowd detection.
  • Topkey and Clearing read receipts and invoices; neither checks that net plus tax equals gross.

4 Feature by vendor

The table compares Hostify with the eight PMS vendors that matter most. Legend: Yes shipped and documented; Beta pilot, beta or limited accounts; Built exists in Hostify's code and is not released; Claim on a marketing page with no documentation found; a dash means not found.

FeatureHostifyGuestyHostawayHospitableBoomRentalReadyLodgifyOwnerRezUplisting
Reply draftsYesYes, freeYesYesYesYesYesYesYes
AutopilotBetaYes, paidYes, per ruleYes, top planYesYes, add-on-Yes, $9Beta
Knowledge base built automaticallyYes, from conversationsBeta, from filesYes, from replies and PDFsYes, from listing, messages, PDFsYes, from corrections---Yes, from manuals
Sandbox before switching on--YesTest questionsDebug mode--Yes-
Reason shown when the AI stays silentYes, escalation noteBeta, bannerYes, missing factYes, skip reasonYes, policy and score--Yes, logScore
Sentiment or urgency-YesYes-YesYes-YesYes
AI disclosure line to the guest-Yes, appended textYes, forced on in the EUYes, optional-Signature---
Task from a guest message-Yes, approveYes, approve-Yes, aloneYes, alone-Yes, alone-
Upsell offered by the AI-Yes-YesYesYes---
Reservation changed by the guest agentYes, arrival times and statusApprove-RefusedMust askExtension by link---
Review reply-Yes, alone above 4.5Yes, approveDraftYesYes---
Reviews turned into tasks or themesBuilt, read-only briefingYesThemesThemesYesYesDashboard--
Portfolio chatBuiltYes, read-onlyYesYes, top planYesYesBetaThrough MCPThrough MCP
Chat that acts after approval-Approval cardsYes-YesYes, bulk---
Catalogue of back-office agents-Yes, 57--Claim----
Finance and accounting checks-Yes--Claim----
Listing text-Yes, with photo visionYes-ClaimYesGenerator--
Own dynamic pricingPartnerYesYes, paidYesYes-Partner-Partner
Owner-facing assistant----Claim----
Official MCP serverBuilt, betaBeta, read-only-Yes, read and write-YesBeta, read-onlyYes, read and writeYes, read and write
Voice agent---------
ID document read at check-inYes--Name matching-----
Damage from a photograph---------
Expense from a receipt---------
Price of the AIIn the plan$39 to $59 per listingIncluded$99 base plus $30 per property for autopilot$30 per listing (small accounts)Add-on and creditsFree in beta$9 per property100 actions, then £49 a month

4.1 What the table says

Hostify is level on three rows and ahead on three. Level: drafts, a reason when silent, autopilot (in beta). Ahead: the knowledge base mined from up to 5,000 past conversations, the guest agent that writes arrival times and check-in status into the reservation, and ID documents read at check-in.

Hostify is absent on twelve rows where at least three competitors are present: sandbox, sentiment, the disclosure line, tasks from messages, upsell, review replies, review themes, portfolio chat, chat that acts, listing text, own pricing, MCP. On three of these (review themes, portfolio chat, MCP) the work is built and not released.

Four rows are empty for everyone: voice inside the PMS, an owner-facing assistant with documentation behind it, damage from a photograph, an expense from a receipt. Two of these four were proposed for the mobile app on 2026-10-09; five weeks later they are still open ground.

5 Hostify today

Two pictures: what a prospect sees, and what exists in the code. The gap between them is the first finding.

5.1 What a prospect sees

  • Two help articles, both dated 2026-04-19: an overview and "what's new".
  • Prompt Mode: suggest, refine, ask.
  • Auto Pilot, labelled beta, with a per-listing switch, active hours, custom instructions and an escalation message that carries a type, a priority and an assignee.
  • A knowledge base that can be generated from conversations.
  • A home page that does not mention AI.
  • An "AI Revenue Optimizer" banner that is a partner's product.

Nothing in AI has been announced since April. In the same seven months Guesty documented 57 agents, Hostaway shipped a chat that acts, and six vendors released an MCP server.

5.2 What exists in the code

CapabilityStateNotes
Auto PilotLive, betaThe agent answers, escalates or stays silent, with tools for the thread, the calendar, flexible check-in, arrival times and local places. In July 2026 about half of its runs ended in a reply to the guest and about one in thirteen in an escalation
Suggest, refine, askLiveThe largest user base of the AI features. A release on 2026-06-27 brought quality complaints (too many "no reply needed", answers to an old message, reasoning text in the draft) and weekly use fell by more than half
Knowledge from conversationsLiveBuilds questions, answers and behaviour rules from up to 5,000 threads over 1 to 12 months, with review, compact and undo
TranslationLivePer thread and per message, in the legacy core and in the mobile app
Hostify AssistantBuilt, not releasedA chat over the reporting module with skills for KPIs, a daily review briefing and a listing set-up audit. The channel-health skill is a stub
MCP serverBuilt, beta, not announcedAbout 38 tools over listings, calendar, reservations, guests, inbox, reviews and reporting
ID document readingLive in the guest appPassport, EU identity card and US licence fields extracted at check-in; a second model checks that the photo is a document
Staff agentsLive, internalDavid takes Intercom conversations, Dexter investigates across the six clusters, Vicky answers from the knowledge base
Listing descriptionDead codeWritten and never called

Sources: wiki/services/ai-agents-and-mcp.md and the documentation folder of the guest agent repository. Usage figures are from the July 2026 overview; nothing more recent is on record.

5.3 What this means

Strengths to build on.

  1. A working guest agent with tools that write to the reservation. Hospitable refuses this outright and Guesty does it only behind approval.
  2. The conversation-mined knowledge base. Nobody in this research documents an equivalent. It is also the raw material for two of the new features proposed later: the simulation gate and the listing truth check.
  3. Fourteen distribution channels in one place, with the content, the rates and the reviews of each.
  4. A guest app that knows whether the guest is verified, has paid the deposit and has signed, and a lock integration. This is exactly the state that Boom and Hospitable use to gate what the AI may say.
  5. Dexter. An agent that already explains what happened in the system from the logs, for staff. No competitor shows this for customers.
  6. Activity logs of the autopilot in a searchable index.

Weaknesses the market will notice.

  1. One quality regression halved the use of the most used feature. There is no test gate that would have caught it.
  2. Nothing released outside messaging.
  3. The autopilot has been in beta for seven months while competitors moved it to a paid tier.
  4. No sandbox, no context panel, no disclosure line.
  5. Built work that is not released earns nothing and is invisible in every comparison a prospect makes.

5.4 Ideas already on record

So that nothing below is presented as new when it is not:

IdeaWhere it is recordedState
Message type classifier, urgency by the stage of the stay, automatic answer settings, schedule change, message to taskHostifyMobile, wiki/ideas/ai-in-the-app.md (2026-10-09)Ideas; the market has since made the first, third and fifth standard
Damage report from a photograph, expense from a receiptSame pageIdeas; still nobody ships them
Voice concierge (P6)HostifyMobile, wiki/proposals/voice-concierge.mdProposal; still ahead of every PMS
Hostify Invest (P5): forecasting, risk agents, data productsHostifyMobile, wiki/proposals/hostify-invest.mdProposal
Support agent pipelinewiki/company/support-agent-pipeline-plan.mdDraft, not built
A dedicated, scoped key for AI agentswiki/raw/surveys/2026-10-24-feature-ai-agent-dedicated-key.mdPlanned
An AI tab for review responses; switching Suggest off per userCustomer requests in the guest agent's documentationRequested

6 What the screens show

Marketing copy says what a feature is for. The screens say how much control the user is given. Forty screenshots and thirty-six demo videos were collected; fourteen screenshots are shown here. The full index, with the address of every image and video, is in the sources.

Two cautions. Screenshots from a help centre show the shipped screen. Images from a marketing page are renders of the intended design, and the captions say which is which. The videos were listed from their titles, descriptions and chapter lists; none was watched.

6.1 Autonomy is granted in slices

No vendor offers one switch. Guesty combines a confidence band, a list of message categories and an assignment of properties. Hostaway sets delivery per rule. Hospitable has named rule sets with their own scope and timing.

Guesty: the set-up page of the AutoReply agent, with the confidence band, message categories and property assignment (help centre)
Guesty: the set-up page of the AutoReply agent, with the confidence band, message categories and property assignment (help centre)
Hostaway: the rules list. Each rule has its own switch, a use counter and a chip when it may send without review (help centre)
Hostaway: the rules list. Each rule has its own switch, a use counter and a chip when it may send without review (help centre)

6.2 The reply delay is a veto window

The planned reply sits in the thread with a countdown. A person can edit it, send it at once or delete it.

Guesty: a queued automatic reply with a five-minute countdown and Edit, Send now, Delete (help centre)
Guesty: a queued automatic reply with a five-minute countdown and Edit, Send now, Delete (help centre)

6.3 A refusal is explained where it happens

Guesty names the category the agent may not answer and offers "Adjust rules" on the spot; a draft is still prepared below. Hospitable leaves an internal note in the thread with the reason in a sentence.

Guesty: the banner shown when the agent declines, naming the category (help centre)
Guesty: the banner shown when the agent declines, naming the category (help centre)
Hospitable: the note left in the thread when Auto Reply did not answer, with the reason (help centre)
Hospitable: the note left in the thread when Auto Reply did not answer, with the reason (help centre)

6.4 The host writes the escalation in words

Hospitable takes the escalation policy as free text and fixes the exact message the guest receives. The same screen has the fallback message and the AI attribution text.

Hospitable: the settings of an auto-reply rule, with the escalation policy, the fixed escalation message, the fallback and the attribution toggle (help centre)
Hospitable: the settings of an auto-reply rule, with the escalation policy, the fixed escalation message, the fallback and the attribution toggle (help centre)

6.5 Disclosure and a sandbox, side by side

Hostaway's general settings put three things on one screen: the reply delay, the AI disclosure that is mandatory for EU listings, and a test panel where a reply can be tried against a listing and a real reservation.

Hostaway: reply delay, the EU disclosure toggle with its sign-off, and the Test AI panel with four presets (help centre)
Hostaway: reply delay, the EU disclosure toggle with its sign-off, and the Test AI panel with four presets (help centre)

6.6 Actions are proposed, then approved, then recorded

Guesty shows the guest's message, the proposed task with priority and type, and two buttons. Lodgify shows why the recommendation matters, one Approve button, and afterwards a stamp with the date and the exact message that was sent.

Guesty: a task proposed from a guest message, with Dismiss suggestion and Create task (help centre)
Guesty: a task proposed from a guest message, with Dismiss suggestion and Create task (help centre)
Lodgify: a recommendation opened, with the reason and one Approve button (help centre)
Lodgify: a recommendation opened, with the reason and one Approve button (help centre)
Lodgify: the same recommendation after approval, stamped with the date (help centre)
Lodgify: the same recommendation after approval, stamped with the date (help centre)

6.7 An escalation reuses an object the team already works with

OwnerRez files an escalation as a row in the normal task list, written "by Rezzy", with the reason spelled out. RentalReady lists what the assistant did to the reservation as a log with links.

OwnerRez: escalations as "Message Review" rows in the task list, with the reason (help centre)
OwnerRez: escalations as "Message Review" rows in the task list, with the reason (help centre)
RentalReady: the conversation overview with the list of actions the assistant took (marketing render)
RentalReady: the conversation overview with the list of actions the assistant took (marketing render)

6.8 The agent catalogue looks like an app store

Category tabs with counts, a badge for active or inactive, a search box that takes a sentence, and a "Request an agent" button.

Guesty: the agent catalogue in the AI hub (help centre)
Guesty: the agent catalogue in the AI hub (help centre)

6.9 The retrieval is shown

Enso prints the steps the autopilot took and which guidebook entries it read, under the reply itself, with thumbs up and down.

Enso Connect: an autopilot reply with its "Thinking Steps", and the side panel that edits the configuration from chat (marketing render)
Enso Connect: an autopilot reply with its "Thinking Steps", and the side panel that edits the configuration from chat (marketing render)

6.10 Patterns to keep

  1. The AI lives in three places: a button in the composer, a drawer for questions about the account, and a settings area in the inbox section.
  2. Confidence is shown as a band ("High, above 90%"), never as a number on each message.
  3. Every proposal carries its reason, and every action leaves an audit line.
  4. AI authorship is marked three ways at once: a visual mark on the bubble, an author line, and a text the guest can read.
  5. When a person takes over, the agent pauses and says so in the thread.
  6. Several vendors show their copilot inside the customer's own AI client through MCP and build no chat of their own.

6.11 Videos worth the time

VideoPublisherLengthWhy
Automate your hospitality operations with AI agentsGuesty, 2026-07-1243 minThe longest walk through the agent hub, the revenue agents and the activity feed
Hostaway AI CoHost DemoHostaway, 2026-06-2413 minA portfolio chat on real account data, with voice and a daily action plan
Hostaway AI Replies: set up, rules and auto-replyHostaway, 2026-07-108 minThe screens shown above, in motion
Hospitable AI MasterclassHospitable, 2026-02-1648 minAuto Reply configuration, Knowledge Hub training, Copilot
RentalReady connects to Claude via MCPRentalReady, 2026-06-222 minOne sentence in an external client returns a formatted cleaning report
EnsoAI AgentEnso Connect, 2026-11-043 minAn agent that edits the account configuration from chat
AI Agents for Hospitality (Conduit V3)Conduit, 2026-06-042 minGuest-facing and internal agents, a template marketplace
I Plugged My Pricing Tool Directly Into ClaudePriceLabs, 2026-07-1716 minWhat customers do with a pricing MCP: ask, analyse, build

7 Catching up

Before anything new, three kinds of work bring Hostify level with what a prospect now compares. They are ordered by how little each costs for what it returns.

7.1 Release what is built

ItemState todayWhat release needs
MCP serverAbout 38 tools at a beta address, deployed from a branch that is not mergedA permission scope per tool in the manner of Uplisting; the dedicated agent key already planned; a help article with the tool list; a listing in the client directories
Hostify AssistantA chat over reporting with KPI, review and listing-audit skills; deployment reported failing since August, not verifiedA working deployment; the access control per listing enforced on the server; the channel-health skill finished
Review briefingA read-only skill inside the AssistantThe same data drives the review reply and the review-to-task features below

Releasing the MCP server also gives small customers a portfolio chat and an owner report generator through their own assistant, at no further cost to us. That is how OwnerRez, Uplisting and Zeevou cover those two rows of the table.

7.2 Put the controls around the autopilot

The autopilot has the core. These six controls are what the market now treats as the minimum for trusting it, and each exists at two or more competitors.

  1. A disclosure line on automatic replies, editable, on by default for EU listings. Confirm the legal duty first.
  2. A sandbox: test a reply against a listing and a real reservation, with the sources shown.
  3. A context panel: every fact the agent was given for this reply.
  4. A planned-reply banner with a countdown and Edit, Send now, Delete.
  5. State gating: access details only when the guest app shows verification, agreement and deposit complete, and the housekeeping status shows the unit clean. Hostify holds all four states already.
  6. Stand-down while a person is typing or has taken the thread.

Then take the autopilot out of beta. Seven months in beta reads as a product nobody owns.

7.3 Close four parity gaps

GapThe market standardHostify's starting point
Review repliesA draft per review in the property's voice; automatic publishing only above a rating the manager sets; low ratings become a task and get no automatic replyReviews per channel are in the reporting data; customers have asked for an AI review tab
Task from a messageA proposed task in the thread with type, urgency and the guest's words; one tap to createThe escalation already carries a type and a priority; the task module exists
Urgency and sentimentA label on the conversation and a filterProposed for the mobile app on 2026-10-09 as urgency by the stage of the stay, which only Duve and Hospitable approached
Upsell in the conversationEarly check-in and late check-out offered when the calendar and the cleaning allow, bought through a linkThe guest app sells upsells; the agent already checks flexible check-in

7.4 What not to chase

  • A catalogue of 57 agents. Guesty's count is a marketing position. Ten checks that are right are worth more than sixty that contradict their own documentation.
  • An own dynamic pricing engine. Airbnb now gives one away, AirDNA entered at $20 per listing, and every revenue tool has an MCP connector. Integrate and explain; do not rebuild.
  • In-chat booking through ChatGPT or Google. Under 1% of room nights by Booking's own figure, hotels only at Google, withdrawn at OpenAI. Prepare the data; do not build the checkout.

8 Features that would be new

Seventeen features in five groups. Each entry says what it is, how new it is, what Hostify already has to build it on, and the guardrail it needs.

"How new" uses three grades, all as of 2026-11-16 and within the limits of this research:

  • Open: not found in any short-term-rental PMS or point tool.
  • Point tools only: exists in a specialised tool, in no PMS.
  • One claim: one PMS claims it, with thin or no documentation.

Sizes are the author's estimates for a first usable slice: S a few weeks, M about a quarter, L longer.

8.1 Trust in the agent

The documented failures of every vendor are the same: a wrong fact stated with confidence. Hostify had its own in June, when one release halved the use of Suggest. The three features below are what the customer-support industry built in answer, and no vendor in our market has them.

1. A simulation gate

  • What. Before a change to the prompt, the model, a knowledge entry or a house rule goes live, a library of synthetic guests is replayed against the agent: the late arrival who is locked out, the refund seeker, the party booker, the guest who asks for the door code before paying. The result is pass or fail per scenario. Two levels: ours, on every release of the agent; the customer's, when they edit their knowledge or instructions.
  • How new. Open. Sierra, Intercom and Decagon ship it for support agents and gate releases on it like unit tests. Hostaway and OwnerRez have a sandbox for one reply at a time; nobody replays a library.
  • Hostify has. The conversation miner that already reads up to 5,000 real threads per customer, the activity logs of the autopilot, and the escalation types as labels.
  • Guardrail. Scenarios are built from real threads with personal data removed. A failed scenario blocks the release and names the reply.
  • Size. M for our own gate, M more for the customer's.

2. A quality score on every conversation

  • What. Every thread, answered by a person or by the agent, is scored against the manager's own rubric: response time, facts correct against the knowledge base, policy followed, tone, upsell offered. A weekly view ranks the listings, topics and team members that produce bad conversations, before the bad review arrives. A second view lists the gaps: "17 guests asked about luggage storage and there is no rule", and how often staff overrule the agent, per topic.
  • How new. Open. Vendors show sentiment per conversation. None scores the answer.
  • Hostify has. All messages, the knowledge base, the escalation record, and the classification tool at low cost per decision.
  • Guardrail. Scores on people are shown to the account owner only and never used automatically.
  • Size. M.

3. Knowledge that knows when it is stale

  • What. Every fact in the knowledge base carries its source, the date it was last confirmed, and the events that invalidate it. A lock code rotated in the lock integration retires the code in the knowledge base at once. A review that says "there was no hairdryer" opens a question against the amenity. A cleaner's checklist confirms an item. A fact nobody has confirmed for a season is flagged before the season starts. The agent refuses to state a retired fact and says why.
  • How new. Open as a mechanism. Enso flags conflicts inside the knowledge base. Guesty lets structured fields win. Hospitable's hub cannot refresh what it read at generation. Nobody ties facts to operational events.
  • Hostify has. The lock integrations, the task checklists, the reviews, and a knowledge generator with review and undo.
  • Guardrail. Retirement is automatic; a new fact always needs a person.
  • Size. M.

8.2 What only a channel manager can do

4. Listing truth across channels

  • What. Airbnb, Booking.com and Expedia now answer the guest before booking from the structured listing data, and the host sees neither the question nor the answer. This feature compares, per listing, what each channel's structured data says, what the knowledge base says, and what guests actually ask in conversations. It reports the contradictions ("Booking.com says free parking, the knowledge base says 15 euro a day") and the holes ("23 guests asked about a cot; no channel has the field set"), and proposes the correction on every channel at once.
  • How new. Open. Guesty detects amenities in photos and fixes rejected Airbnb content. PriceLabs and Beyond score listing content for ranking. Nobody reconciles one knowledge base with the structured data of every channel.
  • Hostify has. Fourteen channels with their content, the conversation-mined knowledge base, and the listing set-up audit already written for the Assistant.
  • Guardrail. Corrections are proposed and approved; nothing about what the property has is invented from text.
  • Size. M. If only one new feature is built, the author recommends this one.

5. A distribution health agent

  • What. A daily check across all channels of one account: connections that stopped syncing, room types that are not mapped, rates that differ between channels beyond the mark-up, content that drifted, a listing that lost its quality status, a calendar block nobody can explain. Each finding has a severity, the revenue at risk and a prepared fix.
  • How new. One claim. SiteMinder announced the same idea for hotels in September 2026 and it is not live. Guesty has single-channel detectors for Booking.com and Airbnb.
  • Hostify has. The channel connections, the Booking.com health tab released this autumn, and a channel-health skill in the Assistant that is a stub.
  • Guardrail. A partial scan is reported as partial, never as "all clear".
  • Size. S to M.

6. "Explain this"

  • What. A button on any object that asks the system why. Why did this reservation not reach Booking.com? Why is the price 140 on Saturday? Why was the check-in message not sent? Why is this night blocked? The answer is built from the audit logs and the rules that fired, in plain language, with the next step.
  • How new. Open for operations. AirDNA explains a rate decision and Hostaway explains the formula behind a number. Nobody explains a sync or an automation. The support bots of competitors are the most complained-about part of their products.
  • Hostify has. Dexter, the staff agent that already investigates across the six clusters and the logs, and two rule-based reasoning engines for overbooking and support.
  • Guardrail. Read-only, scoped to the customer's own account by the server, never by the prompt.
  • Size. M. It would also take questions off the support team.

8.3 Money and owners

Finance is the emptiest part of the market. Guesty has auditors, Boom makes claims, the rest have nothing.

7. The owner contract as a policy

  • What. Each owner contract is read once into rules the system can apply: "repairs up to 150 euro without asking", "linen replaced at the owner's cost", "commission 20% of the net". An expense within the rules is approved and billed to the owner statement. An expense outside them goes to the owner with the invoice and the photographs attached.
  • How new. Open. Ramp does this for company spend policies. Boom has a single spend threshold for work orders.
  • Hostify has. Owner accounting, expenses, the owner portal.
  • Guardrail. The rules extracted from a contract are confirmed by a person before they apply. Every automatic approval cites its rule.
  • Size. M.

8. Payout reconciliation and a month-end close

  • What. One Airbnb or Booking.com payout hides many reservations, fees, refunds and taxes. The agent matches each payout line to its reservation and to the owner ledger, auto-confirms only the matches it is sure of, and queues the rest with the reason. At month end it drafts every owner statement with a short note explaining what changed against the month before, and waits for approval.
  • How new. One claim for the reconciliation (Boom; VRPlatform as a point tool). Guesty's month-end workspace reports and does not close. The explained statement is open.
  • Hostify has. Payments through six processors, the accounting integrations, the owner statements.
  • Guardrail. Money is never moved by the agent. Amounts are integers of minor units; a sum that does not add up is marked, never corrected.
  • Size. L.

9. An assistant for the owner

  • What. In the owner portal, the owner asks about their own property: why March was lower, what the plumber's invoice was for, when the next payout comes. Answers come from the statement, the reservations and the tasks of that owner only.
  • How new. One claim. Boom lists an owner agent with one example and no documentation. No point tool has a chat that explains a statement.
  • Hostify has. The owner portal and the owner accounting data.
  • Guardrail. The owner's scope is enforced on the server. The manager's margin is never shown unless the manager allows it.
  • Size. S to M once feature 8 produces explained statements.

10. An expense from a receipt, with the VAT check

  • What. A photograph of a receipt or an invoice fills the expense form: amount, currency, date, vendor, net, tax, gross. Net plus tax must equal gross before the form is shown; a mismatch marks the field. Already proposed for the mobile app on 2026-10-09.
  • How new. Open in a PMS. Topkey and Clearing read receipts as point tools. No vendor anywhere documents the VAT sum check.
  • Hostify has. Expenses, the document reading service used for identity documents.
  • Guardrail. The person confirms every field.
  • Size. S.

8.4 The property and the people in it

11. A damage file from photographs

  • What. Three parts. A sixty-second guided photo check by the guest at self check-in, which becomes the dated baseline. A screen of the cleaner's checkout photos for leaks, breakage and missing items, with type, severity and description filled in for the cleaner to confirm. A claim pack: baseline, checkout photos, the timeline and the invoice, assembled in the order the channel asks for, inside its deadline.
  • How new. Point tools only for the photo screen (Properly at $5 per inspection, with a person verifying every flag; RapidEye). The guest baseline and the claim pack are open in our market; RentCheck does the baseline for long-term rentals.
  • Hostify has. The guest app at check-in, the task app with photographs, the damage report form.
  • Guardrail. Airbnb rejects any claim that contains AI-generated or AI-modified material. The agent sorts, describes and assembles; every photograph is untouched and every text sent to a channel is written or confirmed by a person.
  • Size. M.

12. Triage before dispatch

  • What. A guest reports no hot water at eleven at night. The agent walks them through the reset for the boiler in that unit, with the property's own photographs. Only a real emergency wakes the technician on call.
  • How new. Open as a product in our market. EliseAI reports, for long-term rentals, 45% of after-hours emergency calls resolved without dispatch (the vendor's figure, one customer).
  • Hostify has. The autopilot, the per-property knowledge, the escalation with priority.
  • Guardrail. A written list of what is always an emergency: gas, water flooding, fire, a guest locked out, anything medical.
  • Size. S on top of the autopilot.

13. An assistant in the cleaner's and the technician's hand

  • What. In the task app: "where is the water shut-off in this flat", "which filter does this unit take", answered from the property's knowledge. Messages between a cleaner and a guest or a manager translated both ways. A forecast of cleaning hours per day from the calendar, with a warning a week ahead of a shortage.
  • How new. Open. Boom lets contractors ask about their jobs on WhatsApp. ServiceTitan and Actabl do the rest in other industries.
  • Hostify has. The task app, the new mobile app, translation.
  • Size. S for questions and translation, M for the forecast.

8.5 Growth

14. A migration agent

  • What. A prospect connects their current system and an agent rebuilds the account in Hostify: listings, rates, owner splits, message templates, the knowledge base. Most competitors now expose exactly the read access this needs through their own MCP server or API. The prospect sees the differences and approves.
  • How new. Open in our market. One hotel customer of Apaleo reports migrations "in a few hours" this way.
  • Hostify has. The import paths of the channel connections, the knowledge generator, and onboarding as a known cost.
  • Guardrail. Runs only with the prospect's own credentials and consent, read-only on the other system. The other vendor's terms of use must be read before this is offered.
  • Size. M per source system. The switching cost is the largest barrier in this market, and this removes most of it.

15. Pricing by outcome

  • What. The agent is charged per conversation resolved without a person, per upsell sold, per review reply published. Not per listing.
  • How new. Open. Everyone in our market charges per listing or per month. Intercom charges $0.99 per resolved conversation and Sierra per agreed outcome.
  • Why it suits Hostify. It is the answer to Guesty's $39 to $59 and to Hostaway's "included": the customer pays only for what worked, and Hostify has to measure it honestly. Feature 2 is the measurement.
  • Size. Mostly a commercial decision; the engineering part is the measurement in feature 2.

16. AI channels in the channel manager

  • What. A row next to Airbnb and Booking.com that publishes the direct-booking inventory, with live availability, price and policies, in the form assistants and AI search read. Booking completes on the manager's own site.
  • How new. One claim. Lodgify prepares its direct sites for AI search and runs a ChatGPT pilot.
  • Caution. Demand is small today. Build the data feed, which is cheap, and wait with anything more.
  • Size. S.

17. Benchmarks across the customer base

  • What. What a turnover clean costs in this city, what commission managers take, how fast comparable accounts answer. Anonymised and aggregated.
  • How new. Open in our market; Ramp does it for vendor prices.
  • Hostify has. The data, and the thesis already written for Hostify Invest that the data is worth more than the platform.
  • Guardrail. Customer consent, a minimum group size, no figure that could identify an account.
  • Size. M, after a legal reading.

9 Recommendations

In order. The costs are the author's estimates for a first usable slice, in calendar time for a small team with agents, not in money. None has been priced by the people who would build it.

No.WhatWhyCostPriorityWho decides
1Confirm the legal duty to disclose AI on automatic guest replies in the EU, and add the disclosure lineHostaway forces it on for EU listings and cites Article 50 of the EU AI Act; our autopilot has noneA legal reading, then daysHighCTO, with legal advice
2Release the MCP server with a scope per tool and the dedicated agent keySeven PMS vendors have one; ours is built and unannounced; it also covers portfolio chat and owner reports for small accountsWeeksHighCTO
3Build the simulation gate for our own agent releasesOne release halved the use of Suggest; nothing would catch the next one; no competitor has thisA quarterHighCTO
4Put the six controls around the autopilot and take it out of betaSandbox, context panel, planned-reply banner, state gating, stand-down, disclosure: the market's minimum for trustA quarterHighCEO for the release; CTO for the build
5Close the four parity gaps: review replies, task from a message, urgency, upsell in the conversationThree or more competitors have each; customers have asked for the firstA quarter, in parallel with 4HighCEO
6Build "listing truth across channels" as the first feature nobody hasThe platforms answer guests from structured data the host cannot see; only a channel manager sees every channelA quarterHighCEO
7Finish the channel-health skill into a distribution health agent, and release the Hostify Assistant with itA stub exists; SiteMinder announced the idea and has not shipped itWeeks to a quarterMediumCTO
8Turn Dexter into "Explain this" for customersSupport load falls and no competitor explains a sync or an automationA quarterMediumCEO and CTO
9Choose the first finance feature: the receipt reader with the VAT check now, the owner contract as policy nextFinance is the emptiest part of the market; the receipt reader is small and already proposedWeeks for the firstMediumCEO
10Prototype the migration agent against one competitor, after reading that vendor's termsSwitching cost is the largest barrier to winning accountsA quarter per sourceMediumCEO; legal reading first
11Decide whether the agent is priced by outcomeIt answers both Guesty's price and Hostaway's "included", and needs the quality score to be honestA decision, then the quality score (a quarter)MediumCEO
12Keep the voice concierge proposal (P6) on the plan and set its dateNo PMS ships voice; point tools and platforms doAlready proposedMediumCEO
13Do not build: an own pricing engine, a catalogue of agents for its count, in-chat bookingThe first is being given away, the second is marketing, the third has no demand yetNoneLowCEO
14Repeat this review each quarterGuesty added 13 agents in six weeks and four revenue tools shipped MCP in five monthsHalf a day with agentsLowCTO's office

9.1 Open questions for the CTO

  1. What this report shows of our own. It quotes usage of our own features and names their weak points, and it reproduces fourteen screenshots of competitors' products for analysis. The CTO ordered its publication on 2026-11-16 knowing both.
  2. The numbers rule. The report tool cannot yet fill figures from a measurement at the root (DEBT-009), so the figures in this report were copied from the research files and the guest agent's July overview. Each is traceable to its source file; none was measured for this report.

10 Sources and method

10.1 How the research was done

Seven agents worked in parallel on 2026-11-16, each on one slice, and each wrote its findings to a file. The files are kept unchanged in the knowledge base. The report was written from them.

SliceFile under wiki/raw/research/Scope
A2026-11-16-ai-features-hospitable-boom-rentalready.mdThree of the four named vendors, every AI feature
B2026-11-16-ai-features-guesty-hostaway-lodgify.mdThe three large PMS vendors, from their help centres
B, annex2026-11-16-ai-features-guesty-agents-extract.mdOne record for each of Guesty's 57 agents
C2026-11-16-ai-features-other-pms.mdTwenty further PMS and channel managers, and Hostify's public pages
D2026-11-16-ai-features-point-tools-and-platforms.mdAbout forty point tools; Airbnb, Booking.com, Vrbo, Google
E2026-11-16-ai-features-adjacent-industries.mdThirty-four capabilities from hotels, long-term rental, field service, accounting, customer support, e-commerce, insurance
F2026-11-16-ai-features-screenshots-and-videos.mdForty screenshots and thirty-six videos, with the address of each
GNo file; reported in the sessionHostify's own AI in the code and the knowledge base

The baseline is the review made for the mobile app: raw/research/2026-10-09-ai-in-str-pms-products.md, 2026-10-09-ai-guest-messaging-tools.md and 2026-10-09-ai-in-str-operations.md in the HostifyMobile repository, compiled in its wiki/ideas/ai-in-the-app.md.

10.2 The four pages the CTO named

  • RentalReady, Maia Insights: https://www.rentalready.com/features/maia-insights/ (read)
  • Guesty, AI for short-term rentals: https://www.guesty.com/features/ai-for-short-term-rentals/ (read)
  • Hospitable, AI landing page: https://hospitable.com/airbnb-ai-lp (did not render to the fetcher on three attempts; the facts come from help.hospitable.com and the community changelog)
  • Boom, AI review: https://www.boomnow.com/ai-review (read)

10.3 Limits

  1. Vendor numbers. Every percentage, time saving and customer count about a competitor is the vendor's own. No vendor publishes accuracy. No independent accuracy data was found in any category.
  2. Snippets. Where a page blocked the fetcher, the fact comes from a search excerpt and is marked "(snippet)" in the research file. Slice E is mostly snippets: six pages were read in full.
  3. Pages that could not be read. Hospitable's feature pages; G2 and Capterra; the iGMS press release; the Track one-pager; the OwnerRez change log index; several press pages behind a 403.
  4. Marketing against documentation. Where the two differ, the report follows the documentation and says so (Guesty's "hundreds of agents", Lodgify's Co-Host, Hostaway's CoHost article that still says read-only).
  5. "Not found" and "open". These mean this research did not find the feature in public sources on this date. A vendor may have it in a pilot, behind a login or in a market we did not read.
  6. Videos were not watched. Their entries come from titles, descriptions and chapter lists.
  7. Screenshots. Twenty-five of the forty are from help centres and show shipped screens. Fifteen are marketing renders. All belong to their vendors and are reproduced here for internal analysis.
  8. Hostify's own figures come from the guest agent's usage overview of July 2026. Nothing more recent is on record, and no audited totals of listings, customers or messages exist in the knowledge base.
  9. Sizes and costs in the last two chapters are the author's estimates.