Google has quietly rolled out haircut appointment booking for Google Duplex, its AI chat agent, after demoing haircut scheduling on stage in 2018
Kyle Wiggers / VentureBeat :
Context & Ripple Effects
Duplex's path from spectacle to product has been unusually slow and deliberately quiet. The 2018 on-stage demo drew scrutiny when Google declined to name the salon and restaurant involved or say whether the calls were edited, and the company responded by restricting access to an opt-in pool of trusted testers and businesses.
Since then the rollouts have been narrow and regional: a New Zealand pilot confirming holiday store hours in late 2019, then an expansion to the UK, Australia, and Canada in April. Quietly shipping haircut booking — the exact task demoed on stage — closes that loop, turning a demo artifact into a live consumer feature.
First-order effects
- Salons that opted in to receive Duplex calls now take bookings from a machine rather than a person, making them the first merchants handling routine transactions through Google's voice agent at scale.
- Google converts its most scrutinized AI demo into a shipped Assistant capability, giving the company a concrete answer to two years of questions about whether Duplex was anything more than a stage act.
Second-order effects
- The opt-in model Google built after the demo backlash becomes the distribution constraint: adoption depends on how many local businesses consent to receive automated calls, not on the technology itself.
- Other businesses in appointment-heavy categories — restaurants above all, given the original demo pairing — become the obvious next candidates, pressuring Google to widen the merchant opt-in funnel before rivals' booking tools do.
Third-order effects
- If the demo-to-deployment lag holds as the pattern, voice agents will reach consumers through narrow, consent-gated tasks first, with transparency norms like opt-in call receiving hardening into standard practice for human-sounding automation.
- Appointment booking is a wedge into local commerce: whoever routes these calls sits between consumers and merchants, positioning Google's assistant layer as a transaction surface rather than a search box.
The trend: Voice AI agents are grinding from stage demos to production through narrow, opt-in transactional tasks, with consent mechanics — not model quality — setting the pace.