Automating Hospitality Industry With A Scripted Chatbot

Notice: This article is maintained for historical purposes. The Bixly Automate/Nebri OS platform that Bixly built, and this article features, is no longer available. Checkout Serverless Providers as an alternative.

Terry Woodward, of http://artilectmarketing.com, reached out to us in 2015 with a challenge. It was our first close encounter with the Hospitality Industry, specifically, Hotel guest hospitality. Here’s an interview with Terry.

Adam: Terry, we were excited to see you use Nebri for the hotel industry. Can you describe what you were looking for when you found Nebri, and how it has helped you develop your application?

Terry: Historic hotels provide a window into our past as part of a fun cultural experience.  Multi-day stays are common and guests look forward to a unique local experience.   Recent advances in natural language processing together with the emergence of increasingly sophisticated chatbot functionality have opened the door to ‘Concierge’ type applications where bots can enhance guest experience.

To provide a convenient on-the-go Concierge experience without the need for an App install, the TwilioSMS/MMS service was chosen to implement chatbot messaging for hotel guests and the Python pyAIML framework was selected to implement the chatbot logic.

Since guests have a lot on their mind when checking in to a hotel, having an invitation to use a Concierge service show up a few minutes after the car is parked and suitcases have been delivered to the room turns out to be a convenient way to begin a stay.

Scheduling  SMS welcome and goodbye notifications to occur in the future based on trigger events required some type of event workflow approach to queue and manage the messages.  Initially Python celery was explored as an option to manage this but it soon became evident that while celery would work just fine to take care of the mechanics of the messaging, there would need to be logging, administrative tools, and other support tools built to manage the process and it would have to be self-hosted.

Nebri … Discovering the Python based workflow/event management framework Nebri in a Python Weekly newsletter suddenly simplified the whole approach to scheduled messaging with the added bonus of a hosted cloud environment with developer ease of use similar to services like Twilio.   In addition, Nebri has a convenient email-to-task parser that can take incoming natural language based emails and transform the data into trigger data for customization python workflow scripts.   This feature made it possible to use a simple built-in Drupal guest registration webform to generate a built-in email to the Nebri system to schedule guest invitation messages – taking advantage of the natural buffering inherent in email systems (a REST endpoint could have also easily been set up within Nebri).    The Nebri administrative dashboard made it easy to develop custom workflow rule scripts, to install and use Twilio API support within Nebri, and to monitor the system performance.  This resulted in to rapid development and deployment of the workflow application.

Adam: That’s perfect! This is why we created Nebri. We were looking for process-heavy solutions like BPM, CEP and so on. That search ended up in sadness, let me tell you. I like the Drupal integration that you mentioned also. How does this tool compare to existing tools?

Terry: Initially, when looking for a good way to implement workflow with scheduling I explored the python Transitions module and the IFTTT saas system.   The Transitions module was fairly low level but needs a backing store and would need logging, administrative tools for managing the workflows and so on so a fair amount of work would be done to create a framework effect.   The IFTTT concept had the Admin side of things covered but was high level requiring a type of DSL approach to configuration and wasn’t Python.   NebriOS turned out to be the goldilocks fit – all of the low level developer control of Python Transitions (can even install Transitions within NebriOS and use it!) – yet with a built-in backing store and the easy administration and the configuration flexibility seen in the saas approach of IFTTT.

Adam: Have you gotten good feedback?

Terry: So far, the automated Hotel Concierge has been a hit with the guests.  Context such as intelligent interactions based on time of day and local events have resulted in multi-day repeated use of the SMS based Chatbot.   Guests have commented that the technology feels like texting a friend for recommendations.   NebriOS has been a key framework element in rapid, contextual deployment of this system.