How we verify, score and update — and who owns this site

Every fact in this index comes from the vendor’s own surface: its pricing page, product documentation, support articles, changelog or release notes. Third-party aggregators, review sites and resellers are not sources here, however confident they sound. Where an aggregator publishes a figure we could not corroborate on the vendor’s own site, that figure is deliberately not recorded — Zeevou is the current example.

Each capability cell carries an evidence grade, and the grade is about proof, not quality:

• Documented — the vendor documents the capability, or documents its absence, in product documentation, a support article or a dated release note.
• Vendor claim only — the capability appears in marketing copy with no documentation, no plan tier and no ship date behind it.
• Announced, GA unconfirmed / Announced, not shipped — the vendor has said it is coming. Announcements are recorded as announcements.
• No evidence found — we could not establish the capability either way on the pages we checked. This is not a claim that the product lacks it, and the matrix cell says Unclear rather than Not shipped.

That last distinction matters enough to state twice. An unevidenced negative claim about a named competitor is not something this site publishes. Where a vendor’s changelog carries no AI entries, we record exactly that — “no AI entries on their changelog as at 29 July 2026” — and go looking at the marketing pages, documentation and support articles before drawing any conclusion about the product.

Six capabilities are tracked for every platform. Nothing on a roadmap counts. Nothing in closed beta or behind a waitlist counts. The test the grades are applied against is set out in full on what is an AI PMS.

• AI guest messaging
• AI review replies
• AI dynamic pricing
• AI guest vetting
• AI upsells
• Agentic / MCP surface

Statuses are Shipped, Partial (it ships, but narrower than the label implies — a risk flag is not guest screening), Not shipped (documented as absent, marked coming soon, or announced only) and Unclear.

A price is published or it is not. There is no third state and no estimate. Every vendor gets one of three transparency grades:

• Published — the vendor publishes a price we read from its own page.
• Partly published — some tiers carry a number and others are quote-only.
• Not published — no number at any tier. Rendered as Quote only in the comparison table.

Prices are recorded in the currency the vendor bills in and are never converted. A conversion rate is a number we would have to invent, and it would be wrong by the time you read it. That is why the true cost calculator is alphabetical rather than ranked: totals in three currencies cannot be honestly sorted.

The calculator computes a total only where the published structure supports one. Of the 12 platforms in this index, 8 qualify. A vendor that publishes a price at one property but not the scale above it does not qualify, and is listed as not computable rather than extrapolated. Excluded from every total, and named in the assumptions each time: VAT and US sales tax, percentage booking fees, promotional rates and negotiable onboarding fees.

Where a vendor publishes a “from” rate rather than a rate, the total is labelled a floor. Where a vendor publishes a minimum commitment, the minimum is applied to the whole bill and the row says so — that single rule changes the answer more than any headline per-unit price in this category.

The Finder asks seven questions and returns three names. There is no model in it. It is a deterministic rules engine over this same dataset, so the same answers always produce the same shortlist and every figure traces to a published page. The rules are printed in full on the tool itself; this is the summary.

• Nothing here is a model. The shortlist is produced by the rules below, applied in this order, to the same dataset that renders the capability matrix and the true cost calculator. The same answers always produce the same shortlist, and every figure can be traced to the vendor's own published page.
• Rule 1 — a platform you already run is not a recommendation. If you name your current PMS, it is removed from the shortlist and listed in the ruled-out section instead.
• Rule 2 — a documented absence is disqualifying for that job. If your biggest time sink maps to a tracked capability and a platform's own material documents that it does not ship that capability, or marks it coming soon, it is ruled out for you. A capability we could not establish either way is not treated as an absence: it scores nothing, and the platform stays in.
• Rule 3 — a budget cap is applied to published prices only. If you set a budget, we compute each platform's monthly total at your property count from its own published structure and divide by your property count. Anything above your cap is ruled out with the figure shown. Prices are never converted between currencies, so a €20 platform and a £20 platform are both treated as 20 — the currency is printed beside every figure so you can see which is which.
• Rule 4 — quote-only pricing cannot clear a cap it will not publish. If you set a budget cap and a platform publishes no price we can total at your property count, it is ruled out and told so plainly. It stays in if you answer that you have no fixed limit, because then there is nothing for it to fail.
• Rule 5 — what earns points. Shipping the capability behind your biggest time sink (+3, or +1 if it ships in a narrower form than the label implies); that capability being documented rather than merely claimed (+1 documented, −1 claim only); publishing pricing (+2 in full, +1 partly); publishing enough to total a bill at your property count (+2); each further capability graded shipped (+0.5, up to +3); keeping a monitorable public release record (+1); billing in the currency of the region you gave (+1). A solo operator gains +1 for published pricing and loses 1 where you must talk to sales; a team of six or more gains +1 where three or more tiers are published.
• Rule 6 — ties break alphabetically, never randomly. Two platforms on the same score appear in alphabetical order, the same order every table on this site uses.
• Rule 7 — the operator's own product is never placed first. Zugrow is made by the company that runs this site. It is scored by exactly the same rules as everything else, but if those rules would place it in position one, positions one and two are swapped and this note is what tells you it happened. If it is the only platform that qualifies, it is shown alone and labelled as ours. The share of results that do not recommend it is published quarterly.
• What is deliberately not used. Channel coverage is recorded on your answer but does not move the shortlist, because this index does not verify which channels each platform connects to — inventing that would be worse than admitting it. Cleaning and turnover coordination is offered as a time sink and no platform scores for it, because no platform in this index is recorded as shipping AI for turnovers.

Two of those rules exist only to constrain us. The operator’s own product is scored by the identical rules but is never placed first: where the scoring would put Zugrow in position one, positions one and two are swapped and the tool states on the page that the swap happened. And every platform that is not shortlisted is named with the rule that removed it, because a recommendation with no rejections is an advertisement.

The integrity number for this tool is the share of shortlists that do not include Zugrow. It is measured and published quarterly. If it is ever low, the rules above are the thing to argue with — they are all here, and they are all the tool uses.

Every capability cell and every price cell carries the date it was last read. The commitment is that no fact stays on this site for more than 90 days without being re-read at its source. When a cell is re-checked and has not changed, the date moves and nothing else does. When it has changed, the change is published in updates and the affected pages have their modification date bumped — a date that never moves without an actual content change.

The whole dataset was last verified 29 July 2026. If that date is more than 90 days before the day you are reading this, the site has broken its own rule and you should treat every figure with suspicion. We would rather you could tell.

Corrections are dated and logged in public. If a figure here is wrong, tell us and it gets fixed, with the fix recorded in updates naming what was wrong, what it now says, and when it changed. Corrections are not quietly edited in.

This applies to vendors as much as to readers. If you work for one of the platforms in this index and a cell misrepresents your product, point at the page on your own site that proves it and the cell will change. We will not take a claim over a URL, from anyone.

The corrections that shaped the current dataset are already on the record. The research behind this launch caught four figures that would have been published wrong: an OwnerRez base price that was actually a five-property total, an apparent iGMS price conflict that turned out to be two portfolio sizes on the same page, a Lodgify pricing page that returns an error to automated fetchers but renders fine in a browser, and an Uplisting changelog that had been recorded as missing when it had simply moved. Each resolution is printed on the relevant profile.

• Ratings and scores out of ten.There is no star rating, no aggregateRating markup and no review schema anywhere on this site, including on the operator’s own product. A numeric score would imply a precision the underlying evidence does not support.
• Rankings.No “best AI PMS” list, no editor’s choice, no pinned row. Every table on this site is alphabetical, and the verdict blocks give a segment-by-segment read instead.
• Estimated prices.A missing price renders as “Not published” or “Quote only”. It never renders as a guess.
• Unevidenced negative claims about named companies. See what counts as verified.
• Vendor leads. No email address collected here is ever sold or passed to a platform in this index. There is no affiliate link on this site and no revenue-sharing arrangement with any vendor named on it.

Links to a vendor’s marketing surface — its homepage or product pages — carry rel="nofollow". Links used as citations — changelogs, release notes, support articles, documentation and published pricing pages — are followed for third-party vendors because they are references. Links to the operator’s product remain nofollowed on those editorial surfaces.

There are exactly two exceptions: the branded corporate-family link on the ownership page and the matching link in the ownership section below are followed so the corporate relationship is explicit. The operator’s product is not linked from the navigation, footer or a sidebar, and all Zugrow links outside those two ownership disclosures remain nofollowed. No other site the operator runs is linked.

aipmssoftware.com is operated by the company behind Zugrow, an AI PMS for short-term-rental portfolios. Zugrow is one of the 12 platforms recorded in this index.

Zugrow is held to the identical rubric: the same six capabilities, the same evidence grades, the same transparency test, the same alphabetical position and the same mandatory “who should not buy this” section. The current record against Zugrow includes two of its six agents marked coming soon, two usage-based prices it does not publish, and no public changelog at all — the same failing this index records against Boom and Jurny. If those sentences ever disappear without the underlying facts changing, this site has stopped being worth reading.

Sister properties operated by the same company, named here in plain text: zugrow.com, app.zugrow.com, guides.zugrow.com. They are not linked from this site.

The full statement, including what the disclosure is for and what it does not fix, is on the ownership page.

Everything here is published by The AI PMS Index editorial team as a corporate byline. There is no invented individual author and no fabricated expert persona attached to any page on this site. Where you want to know who stands behind a claim, the answer is the company named in ownership, and the evidence is the source URL on the cell.