Enterprise-ready, plug-in chatbot solution for Facebook Messenger and Google RCS Business Messaging
Nowadays everything is about conversations, messages and other new ways to get your customer’s limited attention. To make things more complicated, the communication channels are pretty much fragmented. Your business could have a Facebook Messenger platform, but if your customer doesn’t know about it, or even worse has some aversion towards the platform, you are stuck with the regular marketing and communication tools, e.g. emails.
We’ve created a smart conversational platform for businesses to let them fully utilize their omnichannel experiences so that they can have a better chance to win their customers’ loyalty than others.
Mito has joined forces with Google as part of their Early Access Partner program to bring their RBM platform which connects to the operators’ Rich Communication Services (RCS) to the Mitobot Platform.
Mitobot has an easy-to-use UI to build your own bots and now supports Google’s RCS Business Messaging (Google RBM) along with Facebook Messenger.
But how did we get here? Let me briefly explain.
AI-powered chatbots were promised
…and all I got was this lousy T-shirt.
It seems like chatbots were to 2016 as cryptocurrencies were to 2017. All the hype to the moon and back. But the term and the underlying technologies are not new, only that Facebook Messenger and other messaging software gained a lot of traction in the last few years, and they have started to open their platform to the world to create bots. This has summoned dozens, if not hundreds, of startups which aimed to create smart chatbots with NLP (Natural-Language Processing) and other AI-related technologies, such as voice and image recognition and machine learning.
What caused some kind of disappointment and misunderstanding is that customers thought the bots are really smart and powered by AI, but in fact, most of the bots simply resembled a traditional IVR map of the typical question and answer flows of the company’s customer support department. Also traditionally the most used languages in the world (English, Chinese, Spanish) got their NLP support very soon, but what about the minority languages? Let’s say Baltic languages. Without surprise, support is kind of limited to the extent of some basic parsing of dates, locations and other stuff. That’s not enough, but we believe that NLP and voice will and can have their respective places in the UX flow, so we continuously search for new possibilities to use them.
Our approach
We’ve started creating our first chatbot engine -simply called Mitobot- 2 years ago. Originally, the first version of the platform was developed as a pilot application for Deutsche Telekom and its subsidiaries in Macedonia, Albania, Croatia, Poland, Czech Republic, Slovakia, and Hungary. The initial success of the platform has proven that the ideas and key elements of the chatbot platform was worth the time to elevate it to the next level. So in 2017, we’ve revamped the backend infrastructure to create a unified microservices-based, robust, modern and enterprise-ready API which can serve thousands of conversations per second.
Our basic components
Early on in the core development of the platform, we had chosen a microservice based architecture over a monolithic one. This way we can decouple important parts of the system and improve on them as our requirements change over time. Microservices also allow us to easily scale different parts of the system during high load situations in production. The actual needs of the business customers can vary, therefore it is easier to configure a microservice architecture for them.
Unified REST API
The main component in our architecture is the central API written in Go language. It handles the predefined, static conversational flows in a unified format. Every conversational platform has its unique format to describe the message types (eg. simple text message, images, complex templates), but the basics can be unified. This helps us create essentially the same experiences for the users on any platform. The API also allows outside systems to control some aspects of the conversation by altering context (eg. updating the last purchased item from the business), sending trigger messages or broadcasting to single/multiple users.
Communicators
The Communicator instances are there to provide access and control over the conversational platforms that our application connects to. Every platform (eg. Facebook Messenger, Whatsapp, Viber or Google RBM) has at least one communicator instance or more depending on the anticipated overall load. The Communicators are there to receive and send messages and provide relevant statistical events during the conversation.
The future of mobile messaging: RCS
Our solution was proven to be stable and ready-to-use, so we were thrilled when we’ve been given the challenge late 2017 to extend the platform to support RCS.
What’s RCS exactly? It’s Rich Communication Services — basically the next version of SMS, in which the telco operators provide a similar experience to what you’re already used to with Facebook Messenger and in iMessage, from reading receipts to group chats and more.
The good thing is that it’s hardware agnostic, so unlike Apple’s iMessage, RCS is designed to work across all phones and software, you just need mobile data or WiFi. If you don’t have it, then the message will gradually fall back to a standard SMS.
We have been registered to the Early Access Program by Google to be able to reach our goal in less than a months time. Then we’ve successfully delivered a working demo of our integration with the Google RBM to be showcased on the Mobile World Congress 2018 in Barcelona by Deutsche Telekom.
The aim of the demo was to showcase a Customer Service Bot that is able to handle most typical interactions the customers can have with Deutsche Telekom’s services. The starting point of the demo was an onboarding experience where a user would see an onboarding message once opening Android Messages on a newly purchased mobile device. All of the content — messages, and flows — has been created with the help of the Mitobot administration panel without using any scripts or integrations.
The demo shows a future product to be released once the RBM platform and all of the dependencies are ready for production use.
Challenges of the integration
During the implementation phase, our team had to familiarize itself with the Google Cloud infrastructure built around the RBM platform. The other challenge was to understand that the platform is currently undergoing heavy development for the initial public launch in the U.S. with parts of Europe following later this year. Regular updates had to be implemented on a last-minute basis during our project and the changes and new features are still being introduced regularly.
Beyond the demo
We have decided to shape the original approach in a way for it to become much more than a chatbot platform. We want to provide a unified service layer for conversations across any messaging channel. The RBM platform is only one of these integrations.
With our product we’re ready to tackle most of the challenges a business faces in the messaging and IT landscape:
- Unified API is established to connect all integrated conversational platforms and to add conversational features to your product and enhance its feature set.
- CRM integration can be easily created to collect and reply to customer conversations.
- Customer engagement can be extended in conversations across your choice of messaging channels.
- Smart bots behind the conversations can automate common tasks while allowing human intervention any time.
- Elastic Stack (ELK) is used to log and analyze conversations and events, so your data team can have all the data.
- GDPR and EU-U.S. Privacy Shield Framework compliance — our solution is designed from the ground up to comply with EU regulations.
- Cloud-ready — we can deploy the chatbot into your IT infrastructure or to Azure Cloud with Kubernetes.
- No limits regarding chatbots. Do you operate several brands in multiple countries? No problem, you can use integrated admin panel for all of them.
- We’re one of the partners Google has listed on their business page.
Our platform was made to be a plug-in solution in digital transformation projects. Are you ready to chat?
Our dedicated, full stack team continuously adds more and more features to the system. Our business analysts can support both the business and the IT departments with fast and feasible solutions regardless what kind of special modification or integration is required. Also, our extended team of UX, copywriter and analytics specialists can help you to create and measure the best possible user flow to your customers needs.
Interested? We’re happy to schedule a demo with you and your colleagues, please feel free to get in touch with us.
The author(s)
Product Owner
