Four core areas of CPaaS that AI can augment

Channel handling, channel intelligence, human-in-the-loop, and the back end.

Sketching out the CPaaS + AI Domains, shown here for Customer Support.

Communications-Platforms-as-a-Service (CPaaS) domains offer four core areas of focus that AI innovation efforts can be applied to. They are:

1. Channel Handling: Make the channels work better — think if this in terms of user-interface design. How well does this voice sound? How clear is this video? Does the text sound robotic?

2. Channel Intelligence: Make the message and content understood and conveyed by each channel as rich, customized, and valuable as possible.

3. Human-in-the-loop: Present AI-generated information to enterprise users such as contact center agents, and let them make the final decision on how to use it in any given situation. This is a great place to start when it is necessary to build confidence and trust in AI-systems in your enterprise.

4. Back End: Think traditional back-end integrations, but with AI. CRM, ERPs, etc.

Let’s take a closer look at each…

Channel Handling

A lot of work is being done these days in improving speech-to-text (STT) and text-to-speech (TTS) so that humans interacting with voice channels experience interactions that feel as if they are happening with another human. This is essentially a quest to overcome the “uncanny valley” and also offer nuanced but important experience improvements like localized dialects. Similar work is happening with video, while many text channels (SMS, WhatsApp, various forms of web chat) are already able to make use of existing AI platforms for their channel handling needs. Here are a few examples:

1. Voice over the “Uncanny Valley”: Computer-generated representations of humans can run up against the concept of the “uncanny valley” — a psychological phenomenon where a human is put off by an almost-but-not-quite-human image. Tom Hanks’ characters in the movie Polar Express are a great example of this. In voice channels today, it is critically important for the AI-driven representation of the human to sound truly human to prevent the user from jamming “0” or yelling I WANT TO SPEAK TO A REAL PERSON!

2. Localized Dialects: An AI system can convert a user's speech to text while accurately capturing regional accents and dialects, making the interaction more relatable and natural. Perhaps more importantly are Speech to Text (STT) systems that can speak back to a user in their regional (country-level) dialect.

3. Enhanced Video Interactions: AI algorithms enhance video quality in real-time, ensuring clear and uninterrupted video calls, even in low bandwidth situations.

Channel Intelligence

This is all about making a seamless interface experience (channel handling) be an intelligent experience by imbuing the content carried across the channel with AI-backed features that make it so the automation-backed channel experiences meet and potentially exceed the expectations of the human interacting with it. Examples of this include:

1. Real-time language translation: Enables seamless communication between users speaking different languages, broadening the service reach. This could also fall under Channel Handling, but there are opportunities to take this farther.

2. Recommendations / next-best-step: Analyze user behavior and suggest relevant products or services during a chat interaction, enhancing sales opportunities.

3. Sentiment analysis: Not a new concept but certainly one that can improve to handle the ineffable long-tail of human feelings and emotion.

4. Extremely context & history-aware chatbots and IVRs: Mass-customize flows and interactions based on the customer’s previous interactions and current mood, providing a personalized experience.

Human-in-the-loop

This can take many forms, but it is essentially a pattern where an AI system presents information to a human, and the human must make a determination on how to use the information in their interaction with the human customer on the other end of the channel. In CPaaS contexts, this is usually a contact center agent. This focus areas is often a great entry-point for AI-curious organizations. Some example uses cases are:

1. Email & call transcript summarization: Summarize long customer emails and past call transcripts and present the key points to the agent, allowing them to respond more efficiently and accurately.

2. Consolidated view of multi-channel interactions: Seamless hand-off between channels is awesome, but how do you bring it all together for an agent? During a multi-channel interaction, AI can integrat data from various sources and provide a consolidated view to the agent, aiding in quicker decision-making and better end-customer experience.

3. (More!) Next-best-action: An AI system analyzes customer interactions and provides the agent with suggested responses, which the agent can then customize based on the specific context.

4. Simplifying complexity: During a complex support call, potential solutions are created on-the-fly based on historical data, and the agent selects and explains the best option to the customer.

5. Compliance is critical: Monitor ongoing conversations for compliance issues and alert the agent or manager if any guidelines are breached, allowing the agent to rectify the situation promptly.

Back End

“Below” the level of the CPaaS channel and contact center platforms, are the enterprise systems that are ultimately being interacted with them. In a CPaaS context, this might look like:

1. Surface maximum CRM value to the agent: An AI system in the CRM can automatically update customer records based on interaction data, providing contact center agents with up-to-date information for informed decision-making.

2. (Even more!) Recommendations & next-best-actions: Algorithms analyze customer behavior patterns to identify opportunities for cross-selling and upselling, which are then communicated to agents during customer interactions. If you haven’t figured it out already, there are numerous areas in a CPaaS stack to integrate this kind of intelligence.

3. Analyze for Strategy Refinement: Data analytics tools generate insights from customer interactions, helping the business to refine its service strategies and improve overall customer satisfaction.

4. Classification: Classification in AI systems is deserving of its own future posts. A simple example here is that AI systems can automate the classification and routing of customer support tickets based on the nature of the inquiry, ensuring faster resolution times.

5. Bring it all together: AI-driven data integration platforms merge customer data from various sources, providing a comprehensive view that supports better strategic decisions. There are emerging paradigms that make for new approaches to “system integration.”

6. Don’t forget about keeping it all running: Predictive maintenance algorithms analyze ops data to forecast and prevent potential system failures, ensuring continuous service availability.

It is important to note that work in this area, when viewed through a CPaaS lens, could be happening explicitly or implicitly, or both. For example, a CRM team might be doing innovation work that introduces data manipulation to customer records that then has a ripple effect when a contact center agent views the record and makes “informed” decisions based on what they see. This is just the tip of one of many icebergs in the ocean that is the need for enterprise AI governance…

What areas of the intersection of CPaaS + AI do you see? Drop a comment below, or get in touch if you’d like to talk more about this.