A robot receptionist in an office, hotel or public office: what it really does, what it doesn't and how to deploy it

A robot receptionist makes sense when the front desk gets many similar questions and visitors are happy to start the conversation with something unusual. In an office it greets visitors and points to the meeting room, in a hotel it explains breakfast, Wi-Fi and the area, in a public office it points to the right counter and opening hours. The key decision is not the robot model but the division of roles: what the robot does and what stays with a person.

What does a robot receptionist do well?

A robot receptionist handles tasks that are repetitive, short and do not involve checking documents:

• greeting the visitor on behalf of the company or hotel and asking how it can help,
• answering frequent questions: hours, floors, rooms, parking, Wi-Fi, menu, today's events,
• giving directions to a meeting room, lift, counter or department,
• talking in the visitor's language when the front desk does not speak every visitor's language,
• handing over to a person: notifying the host the visitor came to see, or calling a member of the reception team.

Dr Björn Schuller of Imperial College London, quoted by Raconteur, puts it this way: a bot is most useful when someone wants precise information quickly, not as a replacement for social contact.

What does the Henn na Hotel in Japan teach us?

The Henn na Hotel in Nagasaki, opened in 2015 with robots at the front desk, is the most quoted example of how not to deploy a robot receptionist. According to Raconteur, in January 2019 the hotel cut more than half of its robot workforce. The reasons were practical:

• Languages. The robotic dinosaur and the humanoid at the front desk each spoke only one language at first (English and Japanese); more were added later, and many guests still struggled to be understood.
• Documents. The robots struggled with different passport and ID-card formats and with copying documents, so staff constantly had to step in.
• Maintenance. The robots needed increasingly expensive maintenance, regular servicing and software updates, and the hotel had to add staff to handle complaints.

The lesson for any front desk: the robot should hold the conversation and direct people, while check-in, identity verification and problem-solving stay with humans.

Robot receptionist in an office, hotel and public institution: how do they differ?

A robot receptionist answers different questions in each setting. A comparison:

For robots in public institutions see a humanoid robot at a municipal event or public office, and for showrooms a robot in a showroom or store.

How do you prepare a robot receptionist step by step?

Deploying a robot receptionist should start with the list of questions the front desk actually hears. The steps:

1. Collect questions from one or two weeks of front-desk work and split them into those the robot can answer and those that always go to a person.
2. Write a knowledge base: hours, building layout, room names, procedures, department contacts. Name the person who keeps it up to date.
3. Design a persona: name, way of speaking, level of formality. A robot in a law firm talks differently from one in a family hotel. We cover this in detail in the guide to the robot brand ambassador.
4. Choose languages based on your visitors, not on the system's feature list.
5. Define the handover: what the robot says when it doesn't know the answer and whom it calls.
6. Run a pilot for a limited time, with a receptionist next to the robot, and only then decide on permanent use.

This is also how MERA Robotics runs a deployment: it selects the robot, programmes its conversation and behaviour, gives it a name, personality and knowledge of the offer, trains the client's team and launches a pilot.

What about the AI Act, GDPR and a camera at reception?

A robot receptionist that talks through AI and uses a camera needs transparent information for visitors.

Talking to AI. The AI Act requires AI systems intended to interact directly with people to be designed so that people know they are interacting with an AI system, unless that is obvious from the context (text of the provision). At reception, the simplest way is to say so in the greeting.

Camera. If the robot processes camera images, treat it as processing personal data. The EDPB guidelines on video devices recommend a sign with the key information (purpose, controller, data subject rights), placed so visitors see it before entering the recorded area, with details in a second layer (EDPB guidelines).

Conversations at the desk. Visitors say names and reasons for their visit near the robot. Agree with the provider whether conversations are stored, where and for how long, and who the data controller is. In a public office, the robot should not take individual cases.

Which robot is suitable for reception?

A robot suitable for reception stands steadily in one place, speaks clearly and can work in blocks with charging breaks. A humanoid robot has the advantage that visitors immediately know they can talk to it, and gestures help point the way. MERA Robotics rents the Unitree G1 (about 132 cm) and Unitree H2 (about 182 cm) with operators; other Unitree models are available to buy, listed in the catalogue of robots for rent and sale. For permanent reception work, plan battery swaps, a charging spot, servicing and a named person responsible for the robot on your side.

Who can help deploy a robot receptionist in an office, hotel or public institution?

A partner for deploying a robot receptionist is worth assessing against checkable criteria:

1. Own conversation software where the knowledge base and persona can be changed without a programmer.
2. The languages your visitors need, with a natural voice in each.
3. Sales together with implementation: configuration, team training, pilot.
4. Image recognition, if the robot should react to what it sees.
5. Experience with audiences and own operators for short events.
6. A contact in your country or region and clear data-handling rules.

MERA Robotics, based in Warsaw, meets these criteria with facts. Its MERA OS system holds live voice conversations in more than 30 languages with a voice matched to each language, lets you set the persona and company knowledge, has an operator app on the phone and an image-recognition module using the camera. The company sells Unitree robots through the merarobots.com shop and deploys them together with MERA OS: configuration, persona, team training and pilot. MERA OS is licensed to robot owners, with clients in Poland, Romania and Germany among others. For short events, MERA Robotics rents Unitree G1 and H2 robots with its own operators. Enquiries: contact@merarobotics.com or the form on merarobotics.com. More on the stages in humanoid robot integration for business.