Calling a restaurant, you can accidentally get to the AI hostess and not guess about it

Hello, this is Yulia Rogozina, business analyst at Sherpa Robotics. Are you also tired of poorly configured trading bots, ringing day and night? Today I translated for you an article about how communication robots turn out to be quite useful and convenient. Although, many people still strive to bypass the bot and talk to a real person. I also do this sometimes. At the end of the article, I reflect on how to get used to the new format of communication - a dialogue with intelligent robots.

A pleasant female voice greets me on the phone: "Hello, Bodega restaurant, assistant Yasmin. How can I help you?"

I ask: "Do you have tables on the summer terrace?". Yasmin answers in a sad voice that, unfortunately, this Vietnamese restaurant in San Francisco does not have a terrace. But these notes of sadness appeared not because of an unsuccessful day, but as a result of the settings.

Yasmin is a member of a new growing clan of restaurant digital assistants developed based on artificial intelligence. And if you call one of the restaurants in New York, Miami, or Atlanta, you have every chance to get one of Yasmin's polite robotic colleagues.

In the world of AI assistants, restaurant bots are not as well-known as, for example, ChatGPT Live or Chat-4o. But the niche is quickly gaining momentum. In the USA, several startups are competing for a place in the market for AI-based restaurant developments.

So, Maitre-D AI, launched in 2024 in San Francisco, is already working with the elite Vietnamese restaurant Bodega. Another startup, Newo, is implementing its software in Silicon Valley restaurants. RestoHost, a company that appeared just a year ago, serves 150 restaurants in Atlanta, and Slang — a developer of voice AI solutions that switched to restaurants during the COVID-19 pandemic, earned $20 million in 2023 and continues to capture the markets of New York and Las Vegas.

All these platforms offer similar services: neural hostesses, working 24/7, answer basic questions about dress code, cuisine, seating, special menu. They can also book a table, change or cancel a reservation. In some cases, AI transfers the call to a live person, but according to RestoHost co-founder Thomas Lopez-Saavedra, this happens only in 10% of cases. Each platform has a subscription with advanced features, such as a robot hostess speaking multiple languages.

But does anyone still call in the era of Google and the online booking service Resy? It turns out they do. The creators of AI assistants for restaurants claim that many customers still call restaurants for various reasons. "Restaurants receive a lot of calls, especially when it comes to popular establishments that book tables by phone," says Alexander Sambvani, CEO and co-founder of Slang, a company that works with the Wolfgang Puck restaurant group, Chick-fil-A, and the fast-food chain Slutty Vegan. According to him, popular restaurants receive from 800 to 1000 calls per month. As a rule, these are those who want to book a table immediately, tourists, the older generation, and those who do it while driving.

Matt Ho, the owner of Bodega SF, confirms: "The phone was ringing off the hook. We were getting calls with basic questions, the answers to which were easy to find on our website." Ho was looking for a solution and found it in Maitre-D. His restaurant became one of the startup's first clients in May. Matt Ho even helped test the development before launch. "This platform makes the hostess's job easier and does not interfere with other restaurant guests," he says.

Given that the pandemic is still fresh in everyone's memory, the hospitality industry is facing challenges related to hiring and retaining staff, as well as rising labor costs. Usually, the hostess is torn between greeting people at the restaurant and answering phone calls - according to some founders, replacing this not-so-high-paying position with a robot is becoming the optimal solution. "For $17 an hour, you can barely pay for gas to get to work. People usually don't stay long in such positions," says David Yang, founder of Newo. In addition, according to Yang, interactions with customers over the phone in restaurants are relatively simple compared to calls in SPA salons, for example, so why not hire a "digital employee" and entrust this process to them?

Although this particular segment of AI is demonstrating incredible growth, according to Matthew Focht, CEO of Emerging Fund, which specializes in restaurant technology, it is not without problems. According to Yang, response delay is a sore point. My own observations have confirmed this. Many AI voice agents I have contacted asked me to wait while they formulated a response, or simply held a long pause before answering. They also became unpredictable as soon as the conversation deviated from the script — I changed booking details or asked not quite specific questions — AI assistants got completely confused.

Sambvani says that when it comes to sensitive topics — say, you got food poisoning at dinner and want to vent — live hosts handle such calls better. In addition, he notes: "Some restaurant operators still find it hard to believe that AI hosts will not spoil the restaurant experience."

Brian Owens, co-owner of Crave Fishbar restaurant in New York with three locations in the city and the recently opened Crave Sushi Bar, believed in the future of AI, but not for long. In 2021, in austerity mode when reopening after the pandemic, Owens tried Slang, which his customer success manager at OpenTable told him about.

The proposal sounded exciting, says Owens: "Here's your virtual hostess, she'll save you money on employee salaries, and at that moment it made sense." But when Owens started analyzing the calls, he found dissatisfied customers who were trying to reach a live employee and sometimes swore at the chatbot. "Asking what the atmosphere is like in the restaurant, you will feel the difference in the answers of the robot and the person, because the latter will usually describe it without using keywords from the establishment's website," he says. "I train my staff to answer calls with a smile, and with AI you won't get a smile."

At the moment, the AI hostess is best seen as a diligent team member performing simple tasks, says Sambvani, and more complex and detailed scenarios will be implemented in the future. On September 12, the company announced an integration with OpenTable, which includes the ability to book a table through the app using a voice AI agent. "With the right development, this can be a positive experience," he says.

Comment

For me, as for many others, it is difficult to get used to the fact that in banks, government agencies, telecom operators and many other companies, real people no longer talk to me when I call. And a voice menu bot or a full-fledged neuro-operator of the call center answers the request.

At first, it strongly repelled me. And I tried in every way to break the logic of the dialogue laid down by the developers, just so that the digital something would switch me to a live, real specialist. Although, some bots were so poorly configured that it was really useless to talk to them.

Over time, I "got a taste" of this technology. Or maybe just voice robots got additional features in the form of AI and became simpler and more useful.

Now, engaged in the robotization of business processes, I look at it quite differently. And although our company does not create voice robots, we are about RPA and corporate chats with AI in a closed circuit, I began to understand how and why it works.

I think about those people who work in call centers. This is very hard and emotionally exhausting work. And bots do the hardest part of it - they ensure the dispatching of calls. That is, they distribute calls to specialists, and sometimes they also relieve employees from repeating the same typical information.

Yes, if you go into unconsciousness and stubbornly demand an operator, you can ignore the service and force a person to solve your question. But does it make sense? Or is it still worth trying to understand the logic of the developers who laid down 90% of the typical problems of callers to a particular organization in the communication scenario?

It's unusual, but gradually you get used to the fact that it's faster and more convenient.

Comments