Loading ParaVida…
Demonstration dataA demonstration portfolio. Every property and every reading was authored to exercise the product. No hotel supplied any of it and no company is described.
A pilot exists to answer five questions and no others: can the inventory be profiled reliably, do guests understand compatibility, do they trust the explanation, does the partner see differentiation, and does any of it produce evidence worth expanding on. It is a test, not a sale, and the product refuses to present it as ready while prerequisites fail.
Every gate is checked against the live workspace, not against a checklist someone filled in.
No partner has authorised this engagement and nobody has verified the property data. Everything here is an unvalidated external assessment.
10 of 12 properties currently return a score for all 5 guest profiles. The minimum a pilot can learn from is 10; the maximum it can run is 40.
8 readings across the portfolio are AI proposals no person has confirmed. They are excluded from scores today, and they must be resolved rather than tolerated before a pilot measures anything.
The compatibility endpoint is unauthenticated and serves synthetic inventory. A pilot on real inventory needs API key issuance, scoping and rate limiting.
There is no tenant model. A partner's portfolio could not be stored, isolated or access-controlled today.
Criteria are hashed as fnv1a-854d2ac5 before launch, so an edit after the result is known is detectable rather than deniable.
Produced by the pilot engine itself, not written here.
1 blocker stop this pilot being scoped
While any blocker stands, the pilot proposes no properties at all. Reducing scope is not a workaround: a pilot on inventory nobody has verified would measure whether ParaVida can render a page, which is not in question.
Deterministic selection, with the basis carried alongside it.
No properties are proposed while blockers stand.
Properties returning a score for all 5 guest archetypes, in portfolio order, capped at 40. No property was chosen for its score.
One mark on one surface. A pilot that changed five things would tell you nothing about any of them.
Pre-registered and hashed. A threshold ParaVida has no basis for is shown as unset rather than filled in.
| Measure | Instrument | Threshold |
|---|---|---|
| Share of pilot properties with a complete verified projection | Assessment verification queue, recounted at pilot start and end | 100% of pilot properties clear the coverage floor for every archetype |
| Share of readings at partner_confirmed or public_verified grade | Evidence grade distribution across the pilot inventory | No reading in the pilot set carries an unverified AI proposal |
| Distribution of per-property confidence across the pilot set | Assessment output, reported as a distribution rather than a mean | Cannot be set without you |
| Whether the pilot set differentiates between guest archetypes | Best-minus-worst spread per archetype over pilot properties | Spread exceeds the undifferentiated threshold for every archetype |
| Share of guests who open the compatibility explanation | Partner-side event on the property detail page | Cannot be set without you |
| Guest-reported confidence in their booking decision | Single post-interaction question, partner-fielded | Cannot be set without you |
| Whether the partner's leadership team found the assessment decision-useful | Structured post-pilot review against the leadership questions raised | Documented decision taken or explicitly declined, with reasons |
ParaVida has never run a pilot, so it has no basis for what share of guests open an explanation, how confident they feel about a booking, or what distribution of confidence counts as good. A plausible-looking number in those rows would be a goalpost that has not moved yet. They are set with the partner before launch and hashed with the rest, and the hash is what makes a later edit detectable.
Whether guests engage with an explanation, not whether they book more.
ParaVida sees no bookings and receives no guest identity, so it cannot measure conversion and will not claim to. The partner measures commercial outcomes on their own side; ParaVida measures whether compatibility was understood, inspected and trusted. Both belong in the review, and they answer different questions.
The most likely failure is data, not product.
Produced underscoring 1 · weights 4 · ontology 1 · climate model 1 · engagement 1 · min coverage 0.5