By: Milka Holmberg, AI Supervisor, boost.ai
How a virtual agent named Botzilla is helping attendees of Europe’s biggest startup conference stay informed.
Managing customer support and service at a large-scale conference or event can be an enormous and challenging task. It requires that every attendee question is answered in a timely manner and to the same standard of quality to ensure an optimal experience. Something that can be almost impossible (or at least prohibitively expensive) to achieve when relying solely on human support staff. This is where strategic automation via conversational AI can really help.
Research shows that consumers increasingly want (and expect) to use online chat as a way to get instant answers to their questions - whether it’s for online banking, in a retail setting or, yes, at an event. Nobody has time to wait on hold for 20 minutes or hurridly search through a website when they’re trying to make their way to a panel discussion or client meeting.
At next month’s Slush conference, 25,000 attendees from startups and venture capital firms from around the world will descend upon Helsinki, each with their own questions about everything from programme scheduling to sauna timings (yes, that’s a thing!). Via the official event website, they have instant access to a vast wealth of conference knowledge at their fingertips thanks to Botzilla, a purpose-built virtual agent designed to answer any and all questions about the event. Botzilla is developed using boost.ai’s conversational AI technology and - when combined with AI-assisted live chat - gives Slush organizers the ability to provide a consistent customer experience to all attendees and employees.
Built on a foundation of natural language understanding (NLU) and deep learning, the conversational AI that powers Botzilla allows it to truly understand each and every interaction without relying on the user to trigger keywords or phrases to get the answers that they need. This makes for a much more flowing and conversational interaction not dissimilar to chatting with a friend or colleague via other popular chat platforms.
Technology is, however, only as good as the humans that implement it, which is why there are always a few key points to consider when rolling out a virtual agent to enhance the event experience. It is important to accurately understand your attendees’ needs in order to meet them in a satisfactory manner. While this may sound simple enough, it is crucial to do the correct amount of legwork so you have enough data (and enough good data) when you start building your virtual agent.
The two most important questions to understand are: what information do attendees need, and when do they require it? In order to accurately answer these questions, you should consider the entire customer journey, from researching and buying tickets, right up until the moment they leave the venue at the end of the final day. In the case of Botzilla, this process took a while but in the end, it provided a strong foundation for what the virtual agent should be able to assist with and how to reach it.
Once we had determined what Botzilla needed to answer (and when), we drafted a project plan with four key sprint criteria:
Establishing these criteria early in the process provided a great basis for building, training and testing the virtual agent before going live. It is equally important to take a moment to analyze the user data and learn how a virtual agent can be developed further. Once launched, we began to learn even more about what attendees wanted to know about via their daily interactions with Botzilla and used this data to further improve the model.
Before Botzilla, visitors would send Slush an email with any questions they might have about the conference and would typically need to wait 1-2 days for a response. Today, Botzilla can instantly answer many of these questions, helping save valuable time for both attendees and Slush organizers.
It’s not enough to only be thinking about how a virtual agent will help your event today, but more so how can it improve the experience for visitors in the next five or 10 years. Implementation of conversational AI isn’t a one-time, short-term project, but rather a long-term endeavor that should allow scope for pushing the boundaries of the technology to its full potential.