Can AI Replace Your Missing Receptionist?


We seem to be entering the ‘golden age’ of Artificial Intelligence, where tech giants like Google and Apple are introducing AI into everything, where voice-based AI devices (like Amazon Echo or Google Home) are both mainstream and affordable, and where people are generally accepting smarter user experiences. Now is simply the best time to explore ways to exploit Artificial Intelligence to our benefit.

Here at Myplanet, we’re working with leading AI technology partners like IBM Watson to introduce new and novel ways of using AI in the workplace.

We see Artificial Intelligence as having the potential to (1) help offload simple tasks at scale, (2) simplify multi-step tasks, and (3) help people re-think the role of technology in our lives.


The Non-Present Receptionist Challenge

In between building products for customers, and for ourselves, we set aside time for our teams to take on fun and interesting challenges— the kind we don’t have a chance to tackle in our usual day-to-day.

One of the challenges we face in our Toronto headquarters is that we don’t have a dedicated human sitting at a desk at our front door, greeting visitors, signing for deliveries, etc. We don’t even have a front desk. Like a lot of companies with multiple office locations, people just walk right into our office with no idea what to do next. Do they bug the first person they see? Should they continue standing there waiting for someone, anyone, to come and offer to help? Or, more likely, do they wait patiently, feeling increasingly uncomfortable as the seconds tick by, wondering if they’re in the right place or not? 

Empty front entrance @ Myplanet headquarters

This is definitely not the kind of first impression we want to make. We are a user-focused software studio. We try hard to demonstrate high emotional intelligence, we practice mindfulness, and we would never design a ‘first user experience’ like the one visitors face when entering our HQ for our customers. I cringe just thinking about the physical first-time experience visitors face when they enter our historic building. We’re better than that!

As part of our decision to resolve our non-existing receptionist experience, we decided to see if/how AI could be part of a possible solution.


Design Considerations

In a twisted way, we love working on challenges that have tight constraints. Forcing our teams to think creatively and critically about a problem/opportunity and adopt an experimental mindset to what constitutes an appropriate outcome is the kind of challenge we relish taking on. This one fit the bill perfectly.

Here were our initial constraints and considerations for our empty reception challenge:

  • Greeter — It needed to greet people for various reasons including (but not limited to) visiting the building, delivering packages, interviewing for a job, and joining a meeting.
  • Answer Questions — It had to be able to also answer questions about the company, about our historic building, and even chit-chat with visitors like a human receptionist would.
  • No Training Required — It had to require zero user training. Visitors should immediately know what to do.
  • Simple Setup — As much as possible, it should be an extremely easy setup by someone in IT.
  • Low Maintenance — It should do its best to require little-to-no maintenance. IT teams are already busy enough!
  • Corporate Connections — It should be integrated into internal corporate systems (Calendar, Address Book, Email, etc.).
  • Managed Investment — It should be super lean to build, requiring the least amount of people, time and money as possible.
  • Fun — If possible, it should be a playful demonstration as to what’s possible for customers and friends

These considerations were just a starting point. Based on our other intelligent work (like our Siri-Powered Videoconferencing tool), and our own AI-Powered Marketing Assistant product affectionately called “Q” (entering beta as we speak), we knew there would be other constraints and considerations that would surface at some point.


Hello Myplanet Virtual Receptionist

After research and some initial iterations, and with these constraints and considerations in mind, we ended up building ourselves a fun and lean Virtual Receptionist agent for Amazon Echo. 

Myplanet Virtual Receptionist, product concept video

Here is what the Myplanet Virtual Receptionist can do:

  • Greeting — Welcome guests and visitors to the Myplanet office in a friendly way, using just their voice, with the help of Natural Language Processing (NLP) and a flexible grammar model that can be trained.
  • Triage — Work out why the visitor is here, who they are here to see, confirm which meeting room they are in, and even send a quick email to the right team member telling them they have a visitor, all through an integration into our internal corporate systems.
  • Chit-chat — Answer questions visitors may have about the company, about our historic building in downtown Toronto, and even talk about the weather, just as you could with a real receptionist.
  • Corporate Ready — Easy to setup, govern, and administer using a Single Sign-on, with a light admin portal that’s as easy to use as a Facebook.
  • Amazon Echo Based — Works on Amazon Echo devices, available at various price points, with the option to make available on other devices like Google Home or Apple HomePod in the future.
  • Cloud Based — Stores all associated backend technology in the cloud for easy deployment on AWS, Google Cloud, Azure Cloud, Acquia Cloud, or other technologies.
  • Fun — Offers visitors with a new and playful alternative to a human receptionist experience, that’s easy to use and surprisingly memorable.


Our Learnings

We learned that yes, it is possible to leverage inexpensive and readily available Voice-based AI devices to replace a non-existent receptionist in a way that could be scaled to multiple locations.

Here are a few other nuggets we learned, or were reminded of, while crafting this experience:

  • People don’t speak ‘flowchart’. Humans are complex animals. Our conversations are nuanced, massively dependent on context, and almost never happen in a linear fashion. In practice, we need to design conversational agents with flexibility in mind, leveraging Natural Language Processing and Machine Learning wherever possible, and constantly adapting to how people communicate.
  • People want to be understood. “Can you say that again?” and similar re-prompt phrases plague most conversational systems. One of the last things someone wants to do in an unfamiliar/uncomfortable scenario is have to repeat themselves. We need these intelligent systems to leverage the largest possible data set for spoken languages available, understand the context, and ultimately adapt to how humans communicate.
  • People secretly fear ‘Rise of the Machines’. The Terminator movies, The Matrix franchise, and even recent public statements from Tesla CEO Elon Musk, have taught people to have a healthy fear of machines taking over the world. When designing AI-powered products, we need to be empathetic, and clearly not overstep the unspoken boundary that exists between helpful systems, and a sentient machine overlord.
  • People do think AI can be creepy. Phrases like “always-on” and “always available” are considered end-user benefits for all kinds of products. But, when people know you’re “always listening” or accessing your personal information without consent, your product turns from helpful to super creepy. We need to aggressively avoid the creepy factor and not create products that leech ambient data without human consent.
  • Voice-First/Voice-Only is … different. When designing visual user experiences, we have a breadth of design patterns and heuristics that inherently populate even the earliest seeds of design approaches. Tapping, typing, swiping. Buttons, forms, progress bars. It all just rolls out of the design process (and brain) naturally. Designing for non-visual experiences requires we exercise different creative muscles— and practice helps!
  • People Like It! Visitor reactions to our Amazon Echo based virtual receptionist have been extremely positive, with some visitors asking when they can try this product out in their own offices. This is great validation of the concept of using Voice-based AI solutions in this context.


What’s Next?

Based on feedback and requests from visitors, we’re going to make the Myplanet Virtual Receptionist available for other companies to try and buy.

If you’re interested, check out receptionist.myplanet.com to pre-order.


PS — We’re looking for designers, developers and product managers who also want to craft intelligent software experiences. Visit www.myplanet.com/careers for more.

 

Written by

Greg Fields

Greg Fields

Sign up for our newsletter