Manifesto · 2026

Hosts deserve reasoning,
not another black box.

Short-term rental pricing is a guess dressed up as a number. We're building the operating system a smart host would build if they had the budget to — and these are the seven principles that decide what we ship.

A system that reads the market like an analyst.
Decisions that come with receipts.
A tone that treats you like the operator you are.

Every number names its reason.

Not “the model thinks…”

Not “AI recommends…”

Not “algorithm-driven.”

Every recommendation cites the signals that produced it — comp rates, market pacing, the listing audit. Pulled live, timestamped, inspectable.

Receipt: every queue card opens to the comps, events, and pacing it cited.

No write happens without your tap.

Not auto-tuned.

Not “fully managed.”

Not “set it and forget it.”

Approval is the only state of every recommendation — there is no auto-write mode. We do not silently change your listing.

Receipt: human-in-the-loop is on by default on every paid tier; auto-write is a toggle you flip yourself.

The data is live. Not paste-able.

Not screenshots.

Not last week’s bookings.

Not a quarterly report.

Your PMS sync layer, a 365-day forward-rate index across your comp set, and the market's forward booking pace — joined into one view. The screen you check is the screen the agent saw.

Receipt: dashboard timestamps show last refresh; every scan writes a row you can audit.

One subscription. No usage surprises.

Not metered tokens.

Not per-recommendation fees.

Not “$400 because the AI got chatty.”

Flat tier pricing by property count. Every cap is numbered, and at the cap the feature gracefully degrades — never an overage invoice.

Receipt: see the pricing page. Three tiers, one price each, 7-day trial.

Your data is not training data.

Not ours.

Not our vendors’.

Not for benchmarking or “product improvement.”

Service-tier contracts prohibit training on your inputs. Per-tenant isolation. Self-serve export and delete on every plan.

Receipt: see /privacy § 7.6 — “No AI training on your data,” written without hedges.

The trial is the product.

Not a demo sandbox.

Not sample data.

Not “20% of the value behind a paywall.”

Your 7-day trial runs the same engine every paid plan runs — pricing posture, comp deltas, listing health, and recommended moves, computed on your real listing data from your first hour.

Receipt: connect your PMS and your first audit fires within minutes. Same data, same agents, same math as every paid plan.

If we’re wrong, you hear it from us.

Not from a status page after the fact.

Not from a Reddit thread.

Not from your accountant.

A public changelog. Honest tier labels — if a feature ships in Q3 we label it Q3. A “Coming” badge on anything aspirational. We surface our misses.

Receipt: /changelog is dated, public, and updated the same day shipping changes land.

That every recommendation is inspectable. That every write is reversible. That the trial is the full product, not a crippled teaser. That paid tiers pay for themselves on the first rate lift they catch.

And that if we're wrong about any of this, you'll hear it from us first.

— The restay teamrestay · 2026