We are all aware that AI is everywhere around us, from self-driving cars to the simplicity of asking our phones to make a call without even touching it. It's transforming the way we live, work, and play, making everything more efficient and accurate, and solving complicated tasks with what looks like effortless speed. But it's good to keep in mind that, exciting as AI is, it's not a one-size-fits-all solution. It can't solve every problem with just a flick of its digital wand.
Embarking on the AI journey, especially when it comes to developing an AI chatbot for hospitality, means balancing high hopes with the facts of what's actually achievable. In the hospitality industry, where enhancing guest experiences is always the goal, the ability to provide the right information quickly and accurately is critical.
Here at Dharma, our adventure in building our very own AI chatbot taught us quite a bit. It became clear that creating a chatbot that really meets the hospitality industry's needs isn't something you can do overnight. There's no magic button that suddenly fills machines with intelligence. What it takes is a thoughtful mix of technical skill, deep understanding of the hospitality world, and a dedication to never stop improving. That's how we're developing a chatbot that doesn't just do its job but becomes an important resource, ready to offer its users exactly what they need, right when they need it.
While there are numerous AI chatbots available in the market, here at Dharma, we dared to take a bold step by embarking on a journey to develop our very own AI assistant. By creating our own chatbot, we have the flexibility to design it in a way that aligns perfectly with the needs of our industry, ensuring seamless integration with our knowledge base and providing personalized experiences for our guests. Our goal is to revolutionize the way hospitality businesses engage with their guests, delivering exceptional service and enhancing overall guest satisfaction.
We kicked off the journey in early 2021 by constructing our very own Business Data Knowledge Base, which will eventually serve as the main hub of information for our AI assistant. Since then, we have been relentless in our efforts to further create AI tools that can be at our clients’ fingertips in helpful ways in the vacation rental industry - from AI chatbots and beyond, but let’s start by breaking down our journey of bringing our virtual assistant to life.
Though the path has had its ups and downs, we are still dedicated to the task at hand. Our vision goes beyond our own benefit—we aim to create a virtual assistant that will become of great value not only to us, but also to boutique hotels and vacation rental companies.
Having a good knowledge base is key to making an AI assistant work well. It's like giving the assistant a big book of everything it needs to know to understand what people are asking for, find the right info, and give helpful answers to guests and team members. This base of knowledge is what helps the AI be excellent at helping our team members to find the correct answers to their questions quickly.
In the world of hosting guests, a knowledge base is where all the details about how things work, the rules, how to fix common issues, and especially, everything about the rooms and properties are kept. Think of any detail such as how many beds are in a room, how big a unit is, what amenities are there, or how to use certain things in the room - it's all in there.
For example, you can just ask the AI assistant something like, “Is there a sofa bed in unit #483?” and get your answer quickly - and it is an info hub that gets constantly updated, so team members never need to wonder if this information is up-to-date or correct. Having this AI tool at your fingertips makes everything run smoother and helps guests get the info they need right away, ultimately leading to those sought-after 5-star reviews!
The main goal of bringing an AI assistant into the hospitality scene is to boost guest service to a whole new level. Think of the AI assistant as an invaluable team player, always on standby to help out with tasks, whether they're urgent or just need that extra attention.
The AI assistant is available for both staff and guests alike, answering their questions quickly through chat. For staff, the AI answers their questions regarding certain processes “How do I generate a new access code for a guest?” or as a source for detailed information such as “How do I report a maintenance issue?” either for their own knowledge or to answer a guest’s inquiry. When helping the guests, the AI is useful in answering questions related to their stays like “At what time can I check-in?” or to know more about a property, asking the AI “Does the property have a gym I can use?”.
Building and implementing our AI assistant into operations has not been a straightforward process. Our goal is to develop a tool that can help in the daily operations tasks not only for our staff, but also for other companies in the hospitality industry, and during this process we have faced (and continue to face!) different challenges that, in the end, have helped us to refine our tool. But this is what also makes the journey a fun one! Challenges help you improve and transform your weak points into strengths. Let’s walk through some of the challenges we have faced so far:
When it came to collecting, storing and managing information for our AI into the knowledge base, we realized that having the data in one place wasn't cutting it for accurate AI responses. We had to step up our effort and constantly analyze and update the data to make sure there were no gaps in the AI assistant’s answers. Although this is an issue most of the chatbots face during their development, it is crucial to deal with this challenge from the beginning.
To tackle the information update challenge, we came up with a solid plan: We set up regular review routines and got our operations staff involved. They helped gather the latest information using a template document we created, and then they submitted it in a form. Lastly, we put in place an internal process to double-check the accuracy and relevance of the data before putting it into our knowledge base.
Another challenge we faced was determining the most effective format for storing and organizing our information into the knowledge base. Initially, we thought content tables would be the best way to organize our articles, but during testing, we discovered that the AI was not retrieving the information correctly. To address this, we underwent a trial-and-error process, making modifications to the storage format based on testing and feedback. Through our efforts, we were able to refine the storage format, ensuring efficient data access and enabling the AI to generate accurate answers.
One aspect we can’t leave behind is that we also had to make sure the articles in which we have stored the information had a proper and logical flow of information, so the AI can sort the information out and match it to the users' questions in a most effective and accurate way. Overcoming this challenge required dedication to solve the gaps of information in the AI’s answers with the goal of enhancing its performance.
When it comes to the AI’s performance, we faced some challenges beyond just giving inaccurate answers. We had to fine-tune a few things, like picking the right keywords for quick data retrieval. Plus, we had to decide whether the AI should generate short or long answers to the users. And one important thing is that it should be able to recognize who it's chatting with, as there's a difference between how it shares info with staff and how it does with guests. Our AI assistant can be pretty smart that way too!
A crucial part of any new process is gathering users’ feedback, even though it can be uncomfortable at times, as it may reveal areas that require a complete redo, for example, when we had to change our articles’ format, so the AI could be more accurate and logical with its answers. Feedback has been invaluable in the process of improving our AI chatbot.
We actively encourage staff to share their thoughts, suggestions, and concerns with us, so that we can continuously refine and enhance the assistant’s performance, ensuring that it meets the needs and expectations of staff and guests. After all, we're committed to making an AI assistant that people want to use because it actually works - not yet another chatbot that doesn't solve our problems or give generic and basic answers that lead us on a wild chase for the real answers.
It’s important to understand since the beginning of the AI journey, that as any other tool, incorporating AI into operations is a process. At Dharma, we have developed the Early Adopter Program: In a nutshell, it is a 5-step guide that will help you through the process of implementing and taking advantage of AI in your hospitality business.
From filling in your own knowledge base to testing both semi-automated mode, in which the chatbot will be able to propose responses that your staff can review and customize before they reach your guests, and fully automated mode when the AI is capable of handling guests related tasks like reservations or handling their inquiries, freeing your staff to focus on more complex tasks.
Integrating AI into our daily operations isn't just a win for our guests; it's a huge boost for our staff too. Tools like AI take the recurring tasks off their plates, freeing them up to focus on more complex issues that guests might have. This is especially crucial for a growing company where team members often take over multiple tasks. Remember, when our staff is happy, our guests are too!
As a real-world example, at Dharma, we held a meeting in which one of our agents in the Guest Support team expressed that there are moments in her shift when she has to look over so many time-consuming tasks that make it harder for her to complete other tasks. By incorporating an AI assistant, she can dedicate more time to the call from that guest who is requesting an extension of their stay and sending a quote to a potential guest, relying on the AI who will handle that text from the incoming guest who is asking about an early check-in. In just ten minutes, we have a stress-free guest service agent and three guests satisfied by the responsiveness provided by the company’s staff, that leads to a happy guest who will score their experience with us with a 5-star review.
As we've built our AI, one thing has become clear: patience and consistency are key. Developing our own AI has been a process of trial and error, requiring perseverance. We're committed to delivering an AI that brings value to the hospitality industry. Every challenge is an opportunity for growth, and we're excited about the possibilities ahead. For now, Q&A interactions with our AI is just the beginning of the journey; we have started the phase in which we have a foundation to also include Sales, Guest Services, Issue Management and more in the AI interactions, we are eager to continue growing and share these improvements with everyone in the industry as well!