Built for the night out you'd send your sister on.
Most marketplaces sell trust as a marketing line. Greety built it as the product. Six layers, every vertical, every city.
Trust theatre, layered.
Each layer alone catches some bad actors. Together they catch effectively all of them. We don't pretend the system is perfect — we publish what fails and what we change in response.
Government-ID verification
Every host clears a government photo-ID + selfie liveness check before they go live. We re-verify annually. Pseudonyms not accepted.
Powered by industry-standard KYC (we partner with Onfido and Veriff regionally). Records retained per local data-protection law and deleted on account closure.
Two reference checks
Two people who know the host vouch for them — colleagues, mentors, or community contacts. We talk to both, on a recorded call.
References are reviewed by a Greety curator, not crowdsourced. The curator can decline a host even with two passing references — judgement matters.
30-minute video interview
Every host sits through a 30-minute structured video interview before going live. Roughly half the applicants are turned away.
Same call across all four verticals. Au pair candidates have a longer 60-minute interview given the stakes of long-term placements.
On-platform chat only
Conversations stay inside Greety. Hosts who repeatedly push contact off-platform are warned, then removed. The audit trail is part of the safety net.
We don't read message content; we look for off-platform-contact signals (phone numbers, links to other apps). Privacy-respectful, safety-first.
Points held in escrow
Points lock at confirmation and release only when both parties mark the engagement complete. No money changes hands ahead of time.
Disputed releases go to a 48-hour adjudication queue. We side with the guest by default and review patterns weekly.
24/7 human safety line
A real human responds within ten minutes, every hour, every city. One tap from inside the app. Translation included.
Staffed across three timezones (Bangkok, Berlin, Brooklyn). Average actual response time over the last 90 days: 6 minutes 12 seconds.
A real team, three time zones, six minutes average.
Trust isn't a moderation queue. It's a live human at the other end of a button — in your language, in your time zone, inside the app.
One tap from inside the app. Average response time over the last 90 days: 6m 12s. Staffed Bangkok / Berlin / Brooklyn.
Disputed point-releases handled within 48 hours. Default-to-guest stance. Patterns reviewed weekly by safety lead.
Every December: total bans, top failure categories, response-time distribution, what we changed in response. Published on /trust/transparency.