“Communication data itself is extremely important for hotels which was not being captured until now.” #AIBoss [Interview]
We, at NextBigWhat, are attempting to drive forward the conversation around Artificial Intelligence in India beyond headlines that scream for attention but don’t offer much, buzzwords that make your head buzz after a while, and countless utopian tales that vie for your eyes only to leave you with a scratching head and a lingering itch to actually get to the meat of the matter.
In this spirit, the #AIBoss series intends to educate, inform and elevate our readers understanding of the subject as well as the space. If you’re part of the leadership team of an AI based product, from enterprise or startup, willing to share your AI learnings – then reach out to us and become part of this initiative (you can apply using this form).
Here is the first of two parts of our interview with Parag Arora, who is a technology entrepreneur with deep insight into messaging solutions, having sold his first startup Plustxt to Paytm in 2013 and currently holding the position of Chief Technology Officer at Glowing.io – a YCombinator selected startup that provides an AI based messaging solution to the hospitality industry.
NextBigWhat caught up with him for a rather extensive tête-à-tête on the journey of Glowing.io from conception to product, AI & NLP in the hospitality space, and whether or not he considers voice based solutions like Alexa a threat in the coming future.
For those who’re unaware, how would you describe Glowing.io?
Glowing enables digital communication between hotels and guests with the main focus being on enhancing guest experience while at the same time increasing the operational efficiency of hotels.
How did you arrive at the pain point and then eventually the product?
For me, it was a mix of personal experience and excitement when I saw immense opportunities for implementing technologies I love and have built at previous companies.
On the personal side, I always felt a gap when visiting even luxury hotels where I was not able to enjoy basic experiences like ordering a beer without physically having to move when chilling near the swimming pool and so on.
I personally see it as a big trend now that people prefer buying international internet packs than voice packs which creates a problem when you have to connect with your hotel. Also, in extremely busy schedules – people always prefer texting more than having to wait to talk to the hotel over voice calls.
On the technology side the opportunity is huge, as guests are going to connect to their hotels pre-stay, during stay and post stay and the fact that the problem of broken communication can be solved to the extent that conversations can be 100% automated eventually is pretty exciting.
This communication data itself is extremely important for hotels which was not being captured until now and a lot of interesting things can be done over this data.
Product conception was done mainly to solve the problem of digital communication at hotels as a first step with the realisation that we need to fix other problems which this means affords us to solve, such as communication barriers owing to factors like language, reluctance to talk, downloading a new application from guest app etc.
Also, the ability for hotels to have centralized data of messages where conversations can be assigned and escalated.
What was the process like in terms of evangelising hotels to adopt the service?
There was definitely a gap between hotels’ understanding and guest needs initially where hotels used to believe conventional face-to-face is the mechanism of communication guest prefer.
Two major things which have changed that perspective is the rise of the millennial market for luxury travel and international travel with a lot of communication barriers.
We believe the ability for guests to communicate in their own language over the messaging platform they are already used to, in their interaction with their friends and families, is going to redefine the relationship between hotels and guests.
Most of the evangelism process was involved in educating hotels on the benefits of messaging by showcasing how seamlessly they can provide guest services once they integrate all their systems to the communication platform itself (Imagine guest orders food over WeChat in Chinese in Switzerland and ticket is automatically created in Food Ordering System).
What are the frictions in the hospitality industry that Glowing.io went in to solve and what were newer ones you were able to learn from your customers?
There were mainly two kinds of frictions we have faced from the industry, one was the reluctance of hotels as they started believing they will need to add more resources for communication and other was the reluctance of staff in trying a new product.
Providing demo and onboarding to staff became a winning factor to solve the above problems where we designed the system to minimize decision making and in using the product. The most important thing was we had started selling and talking to hotels before even building the product and we were well aware of these problems which also formed as a core design philosophy behind while building Glowing.io.
On onboarding, we worked with a general manager of one of the most luxurious properties of the world who helped us create the complete onboarding program for Glowing which also we have automated a lot during our journey with help of quick videos and inside the product.
There seems to be some confusion with regards to the level of overlap between AI & NLP – there are some who question whether NLP integration can be legitimately claimed as a significant AI-based feature when today companies and startups are simply plugging into pre-existing libraries. How do you look at this and could you help clear the confusion?
AI is a field of computer science which was created mainly to solve tasks easier for humans but hard for computers. With new advances, AI is matured enough that we are able to solve problems that are hard for humans now but easier for computers. The eventual goal of replicating humans is much further away but definitely, new advances are approaching that direction.
NLP is just a part of AI focused on solving problems around languages.
There are a lot of plug-and-play solutions available in the market which are very cost effective as compared to building our own systems. The answer we need to find is what to do with these tools.
It is more like deciding whether to go for cloud or to host your own servers. For businesses I always advise to go for tools if they are mature enough and solve your needs. Most of these tools also provide probabilistic confidence on the solution where UX plays an important role in improving via feeding back data combined with human assistance.
Only drawbacks are with tools which do not provide a functionality to provide learning with data.
Our experience with NLP is for texting in international messaging where messages are well formatted and are not very long and for these kinds of messages, NLP out of open tools such as Google works amazingly well.
There are a lot of advances still needed in NLP which could thread sentences from the context as currently NLP tools provide line by line translations without it.
In short, NLP definitely is part of AI strategy and companies are making decisions not just in NLP just as plug-and-play but also in a lot of other contexts for AI and that is the best move until you find that no external tool will answer your need or have concerns on data privacy (e.g. in GDPR).
Since Glowing.io is described as a ‘bot and human-assisted AI’ platform – in which hotel processes is there a bot enabled and where does human intervention come in? And what was the research process like in going about designing it?
I cannot name hotels who are currently automating conversations for various reasons. Human-assisted is a rule-based algorithm which mainly comes in when there are discussions happening on mission-critical tasks like reservations or when probabilistic confidence is lower than a certain percentage for understanding context or a reply of that particular context is not handled by the bot.
A good example of where the bot is used is during checkout. As it’s extremely important for hotels to know someone is checking out since they can schedule cleaning and upsell room for early check-ins but guests generally don’t want to go through the hassle of formally checking out.
There are other instances where guests ask questions like ‘where is the buffet’ etc. Most of the usage of bots are in what we call “goal-based conversations” where the discussion is not open-ended but the guest has a specific task in mind or wants to know a definitive piece of information.
Research process involved taking into account a lot of first-hand experience in understanding conversations happening over Glowing as well as visiting properties and finding out how requests are taken care of in terms of operations. A lot of focus while automating conversations is always given for operational efficiency in achieving tasks.
Since Glowing.io supports interactions in multiple languages, have you also implemented NLP for non-English languages and what challenges did you face in the process?
We have not implemented NLP in non-English languages natively and therefore translate and then attempt to understand the context. It is harder to scale NLP to other languages as models and the whole system becomes drastically different and each language looks like an opportunity to create a huge organization. This is still a problem which needs to be solved and most solutions right now focus on keywords based approach.
The second part to this interview will be published tomorrow. Make sure you’re subscribed to the AI Boss newsletter to keep up with the latest in AI developments across the world!