Automated conversations at scale with RCS and Vertex AI
RCS (Rich Communication Services) Business Messaging has always been an attractive channel for brands due to the range of features that it offers that are not available with SMS.
The only problem was that RCS was not supported on Apple devices, so RCS would always have to be used in conjunction with SMS, or with a messaging app like WhatsApp, Messenger, or Viber to ensure maximum reach for key messages.
All that changed with the announcement from Apple that they will be supporting RCS in 2024. This is expected to coincide with the iOS 18 update scheduled for around September of this year.
Even before the announcement in November 2023, RCS was enjoying somewhat of a boom. Our recently published Messaging Trends report shows global growth in RCS messages on the Infobip platform in 2023 was a massive 358%.
RCS Business Messaging is now set for exponential growth, and will no doubt become one of the primary channels that businesses, government agencies, and all types of organizations use to communicate with customers and citizens.
As we have already seen with WhatsApp and Facebook Messenger this will include introducing AI and chatbots to automate a range of important functions on the RCS channel including user-initiated chats for customer service queries, marketing communication, and both transactional and sales support.
Learn how major UK supermarket chain ASDA has partnered with Infobip to incorporate RCS to support messaging across the full online customer journey, including order confirmation, delivery times and substitutions and in the process becoming the largest sender of RCS traffic in the country.
Does RCS support chatbots?
Yes, RCS already supports chatbots. And they can go far beyond what is possible with SMS chatbots as they can include interactive features like structured conversation flows, images and carousels, GIFs, suggested replies and actions, and the option to transfer to a human when required.
What types of chatbot does RCS support?
Due to the flexibility and many inbuilt features of the RCS channel, it can be used to deploy all sorts of chatbots, from simple menu-based bots that guide users to the information they require, to more advanced conversational chatbots that are trained to speak human (more on this later).
Some examples of RCS chatbots include:
- Menu or button-based chatbots: These chatbots present users with a set of predefined options or buttons to choose from, much like a Voice IVR menu, which guides them to the answers they require. guiding them through a structured flow.
- Rule-based chatbots: An extension of button-based chatbots, these bots use conditional logic to build structured conversation flows. They effectively behave like interactive FAQs that respond to user inputs based on pre-programmed rules.
- Transactional chatbots: Designed for specific transactions, such as booking appointments or processing payments, these chatbots are designed to guide users through the process quickly and easily.
- Conversational chatbots: These AI-powered chatbots use machine learning and natural language processing (NLP) to understand and respond to more complex and unstructured user queries. They can be trained to deal with a wide variety of queries and requests and can mimic the experience of dealing with a human agent by understanding context and being able to switch from one function to another without starting a whole new process workflow.
For each type of chatbot, the unique features that RCS provides can be used to create a more effective and engaging user experience. By incorporating advanced machine learning capability, businesses can give their customers controlled access to huge resources of information and services via the RCS interface on their mobile phones.
One platform that offers a complete environment for creating conversational RCS chatbots is Vertex AI.
What is Vertex AI?
Vertex AI is a machine learning (ML) platform developed by Google Cloud. It has been designed to help software developers and data scientists to easily build, train, and manage ML models and AI applications by providing all the tools and services required for a range of AI models in a single platform.
Some of the key features of Vertex AI that make it well-suited to building chatbots for RCS include:
- AutoML: This feature allows you to train models on tabular, image, text, or video data without writing code or preparing data splits.
- Unified workflow: Vertex AI provides a single environment for the entire machine learning workflow, including data preparation, model training, and deployment.
- Custom training: The platform offers complete control over the training process, including the use of preferred ML frameworks, custom training code, and hyperparameter tuning options.
- Model garden: Vertex AI includes a collection of pre-trained models and assets that you can use to test, customize, and deploy.
- Generative AI: It includes access to Google’s large generative AI models for multiple modalities, which you can tune and deploy in your applications.
- MLOps Tools: Tools to automate and scale projects throughout the ML lifecycle on managed infrastructure.
What is the difference between Google AI and Vertex AI?
Google AI and Vertex AI are both part of Google’s suite of artificial intelligence services, but they serve different purposes and are designed for different types of users.
While Vertex AI is a specific product that falls under the Google Cloud platform, Google AI refers to the company’s broader AI initiatives and research that encompasses everything from the company’s guiding AI principles and research papers to their open-source AI tools for image analysis, natural language processing, and much more.
How much does Vertex AI cost?
Unlike platforms like Tensorflow and PyTorch, Vertex AI is not an open-source platform so there are costs involved, depending on which service you use.
Here’s a general overview of the pricing structure:
- Training: You pay for the computing resources used during the training of your models. The cost varies based on the machine types and the duration of the training.
- Prediction: When you deploy models to make predictions, you’re charged for the prediction requests. This includes online predictions, batch predictions, and use of AutoML models.
- Storage: There’s a cost associated with storing your datasets, models, and other resources on Google Cloud.
- AutoML: For AutoML models, you pay for training the model, deploying it to an endpoint, and using the model to make predictions.
- Generative AI: If you use Generative AI features, you’re charged based on the number of characters in the input and output for text, and per image or second for media content.
It is important that you check the official pricing page for detailed information on costs and pricing related to your own use case.
How Vertex AI can be used to build effective RCS chatbots
Building an RCS chatbot with Vertex AI will enable developers to create intelligent bots that can deal with a range of queries, transactions, and customer support issues across the entire customer lifecycle. Rather than building separate chatbots for each use case, all can be serviced in a single messaging thread on RCS.
The ability to process and understand both structured data from a customer data platform and the unstructured data that the chatbot receives via multiple customer interactions means the chatbot can provide meaningful and accurate information in a conversational and human way.
Ideal use cases for RCS and Vertex AI
As a messaging tool RCS is ideal for chatting and exchanging media with friends, family, and with businesses that use the channel. However, it can become so much more than that with a platform like Vertex AI working miracles in the background.
Here are just a few use cases where the combination of RCS and Vertex AI can create something exceptional.
1. Making a lot of data easily accessible
The term big data is thrown around a lot these days, but what does it actually mean for the average person who just wants to find out a bit of information specific to them from a business?
With the ability to do real-time analytics and data mining operations, a Vertex AI powered chatbot makes it possible for verified users to access detailed information just by sending an RCS message. It’s almost like having your own personal GenAI model at your disposal 24/7.
For example, if you are getting your finances in order and have a bunch of questions for your bank like:
- How much have I spent on mortgage repayments in the last five years?
- How much interest did I pay on my credit card last month?
- When do I qualify for a lower rate on my mortgage?
2. Recommendations on steroids
Vertex AI can be used to create powerful and hyper-accurate recommendation systems for specific customers. With access to a customer data platform, it can respond on the fly to questions sent by RCS that require instant analysis of purchase history, preferences, and any other available data. Its machine learning algorithms also mean that these recommendations will improve over time, for every single person.
3. Image and video recognition
You can easily send images and video by RCS. Vertex AI supports sophisticated image and video recognition models that can be used in all sorts of scenarios, from categorizing products to finding items similar to an uploaded image.
For example, if you spot something you like in a magazine you could ask “Find me similar coats to this one for under $200”.
In the security and investigation sector it can also be used by field officers to submit photos to cross check against a database of known people, for example football hooligans that are banned from entering stadiums, or people linked to illegal gambling operations.
4. Personal travel assistants
How useful would it be to have someone else make all your travel arrangements and bookings, taking instructions from simple RCS messages?
Vertex AI can make this a reality.
When a customer of a travel company says “Book me the first flight on Monday morning from LA to Dallas”, the NLP chatbot will understand the ‘book flight’ intent, extract relevant information like flight origin, destination, and the required dates, and then find matching flights by making API calls to the airline systems. The chatbot could even prebook the customer’s preferred seat and meal options based on previous knowledge.
And it’s not just flights – hotels, restaurants, and local attractions can all be retrieved by Vertex AI’s search model, ranked, and organized according to the user’s own requirements to provide a personalized and easy search experience.
Why choose Infobip as your RCS partner?
Infobip is an official Google partner and has been offering RCS Business Messages since 2017. We have been repeatedly recognized as an industry leader in the RCS and omnichannel communication market and continue to innovate with services incorporating AI and machine learning.
Crucially, we recognize that RCS is often one of several communication channels that our customers use. As experts in omnichannel messaging and with an unmatched portfolio of telecom partners across the globe, we can ensure that RCS messages, and all other channel messages are safely and securely delivered to customers everywhere.
The combination of our RCS expertise and partner network, the power of Google’s Vertex AI, and your own industry knowledge and customer insights guarantees conversational RCS interactions that will set you apart in your own market.