Starting in 2027, the Appointment Is Law: What California's Right to a Human Means for Every Company Using AI in Service

On September 28, 2026, California Governor Gavin Newsom signed a law into effect that is part of no AI strategy yet will affect every single one: the Right to Human Customer Service Act, Assembly Bill 1609. Starting January 1, 2027, every customer in California will have an enforceable right to opt out of a conversation with an AI bot — and reach a human within 15 minutes.

And if that fails? Then the law mandates what this blog has stated since its very first article: a scheduled appointment with a human, within one business day.

I know of no other law that establishes a confirmed appointment as a minimum standard for customer relationships. That is why it is worth taking a closer look — even from Luzern (Schweiz), Zurich, or Frankfurt.

What the Law Requires

AB 1609 applies to companies with more than 500 million US dollars in annual revenue serving customers in California. Exemptions include regulated utilities, credit reporting agencies, and hospitals for health-related communications, among others. For everyone else, starting January 2027, the following applies:

Inquiries sent via email, web form, or voicemail do not fall under the 15-minute rule. That is a loophole open to debate. However, what the legislature explicitly permits is far more decisive than any loophole.

What the Law Does Not Prohibit

AI in customer service remains permitted. The bot can still handle initial contact, clarify issues, and deal with routine tasks. The law does not ban automation. It bans dead ends.

"Technology should make life easier, not prevent people from getting help." — Rick Chavez Zbur, author of the law, Assemblymember from Los Angeles

As his motivation, Zbur shares a story everyone knows: During the pandemic, he tried to get medication for his mother and was put on hold for over four hours — when he finally arrived at the pharmacy in person, the call was still connected. Business associations, led by the California Chamber of Commerce, fought earlier drafts as overly rigid. The result is a law with flexibility: "commercially reasonable efforts" satisfy the good-faith standard.

It is precisely this flexibility that makes the appointment clause so interesting. The legislature acknowledged that not every company can connect every customer to a human within 15 minutes — and defined an equivalent alternative that many customers would prefer over an immediate conversation anyway: a binding time slot.

Why a Law Was Needed

Because while the market punishes dead ends, it does so too slowly. In December 2023, Gartner surveyed over 5,700 customers: 64 percent would be happy if companies didn't use AI in customer service. 53 percent would consider switching to a competitor if they learned that a company uses AI in customer service. The main concern isn't the technology itself—it's that it blocks the path to a human. Keith McIntosh from Gartner summarizes it like this: Many customers fear that generative AI will simply become another obstacle between them and a human representative.

These figures are three years old. Since then, chatbots have doubled—in Switzerland from 20 to 40 percent of companies according to the Swiss CEX Study—and Forrester has measured declining customer experiences for the fourth consecutive year, citing "disappointing AI implementations" as one of the causes. California hasn't legislated against the technology, but against its most common misapplication: the bot as a bouncer.

The appointment as a legal fallback—and why it's often the better way

Read the appointment clause again: If a human is not reachable within 15 minutes, a specific appointment must be offered within one business day. The legislator treats the appointment as a fallback. In many cases, I consider it the better way—for three reasons that have to do with revenue, not compliance:

First: The customer chooses the time. A 15-minute callback catches the customer wherever they happen to be—on the train, in a meeting, cooking dinner. An appointment at 6:30 PM catches them where they can actually engage in the conversation. For a service request, that doesn't matter. For a consultation regarding a mortgage, an insurance policy, or a kitchen, it's the difference between closing the deal and "let me think about it."

Second: The human is prepared. Anyone connected in 15 minutes takes over cold. Someone taking a scheduled appointment knows the issue from the bot dialogue, the history, the urgency. The first five minutes belong to the solution, not the repetition.

Third: Capacity is planned instead of suffered. A service center that must handle every bot exit immediately builds buffers for peaks—expensive and still never enough. A service center that schedules appointments distributes the exact same demand across the capacity it actually has. That is the difference between a queue and a calendar.

What this means for Switzerland and Europe

Legally speaking, very little at first. A right to human customer service exists neither in Swiss nor in EU law. Since August 2, 2026, the EU AI Act includes a transparency requirement: Anyone interacting with an AI system must be informed about it unless it is obvious. This covers the disclosure requirement from AB 1609—but not the opt-out, the 15 minutes, or the appointment. Swiss data protection law is silent on both.

In practice, it will arrive anyway—and faster than through legislation. Major contact center and chatbot platform vendors serve customers in California. They will build the opt-out, the 15-minute logic, and the appointment fallback into their products—and once built, a feature becomes a standard, not a regional option. What is mandatory in California becomes an expectation in Zurich. Customers here read the same news.

And they are already judging companies by it today. A customer stuck in a Swiss bank's chatbot who can't find a way to reach an advisor doesn't know that the law doesn't grant them that right. They only know that the bank doesn't want them.

Designing the path to a human

Anyone planning automation should build the path to a human right along with it — not as an emergency exit, but as part of the journey. Five elements prescribed by AB 1609 that every company can implement:

None of this requires a new system. It requires connecting the bot, calendar, and advisor — and the decision to treat the transition as part of the product, not as an exception.

What I find remarkable about this law

Legislators usually regulate what has gone wrong. AB 1609 regulates what needs to be done right — and in doing so, describes quite accurately the mechanics of turning customer interaction into revenue: The bot sorts, the human solves, and in between lies a transition that either works or invalidates everything before it.

The fact that a legislature defines scheduled appointments as a statutory minimum standard is, for me, the strongest argument in recent years for treating it as a leadership priority. Not because California demands it, but because California has apparently seen what happens when you leave it to chance.

Takeaway

Starting January 1, 2027, large companies in California must allow every customer to exit an AI bot, connect them to a human within 15 minutes or offer a scheduled appointment within one business day — and disclose when a bot is speaking. AI remains allowed; dead ends are prohibited. While the law does not apply to Swiss and European companies, the standard already does: Platforms are building it in, customers expect it. Anyone planning automation builds the path to a human along with it — exit, intent recognition, choice between immediate and scheduled, context handoff, measurement. This isn't compliance. This is the journey where an inquiry turns into a deal.