AI: SkyNet is Bad, but a Great Customer Experience is Good!

For those of us who grew up watching The Terminator, the concept of artificial intelligence invokes a knee-jerk reaction of fear as we anticipate doom for humanity once these sentient machines inevitably discover their own unfettered will to rule. While that fear is likely just an overblown result of Hollywood storytelling (I mean, right?!), the actual, real-world impact has been much less diabolical so far. And the potential is intoxicating. Can we leverage large-data-driven AI for the early identification of disease…and recommend a hyper-personalized prevention/treatment plan too? How about AI to optimize the production and distribution of the world’s food supply to eradicate hunger? What if AI could run a city’s network of traffic lights and public transportation schedules to optimize throughput and mitigate travel delays at rush hour? Or perhaps, we can use it to optimize our electric grid to maximize up-time, minimize outages and lower the carbon footprint? The possibilities are endless. 

While the future of AI might be bright, most of us aren’t shocked to see that many early adopters are starting to leverage AI first to simply improve their customer experience across their customer journey. While that may not sound as lofty as solving for world hunger, let’s be honest–we all really love a great customer experience when we’re the customer. We’re glad to have our concerns solved in a single chat or phone call. We’re happy to be recommended the perfect product (and coupons!) from a brand we trust at just the right moment. We’re glad when a company we buy from can proactively solve problems before we even know they exist. 

Most in the business know that Twilio is the industry leading CPaaS (Communication Platform as a Service) that powers much of the world’s communications today. Twilio makes a compelling case for their place at the table when it comes to AI and customer engagement, but their strategy is one of partnership with best-in-breed, emerging AI technologies rather than head-on competition. Smart move in an environment where there are lots of entrants into the marketplace who have the IP, but not the reach; who have the product, but not the global infrastructure to turn it loose. Twilio understands that they can let the market battle it out, but that mostly ALL of those solutions will still need to come to them if they want their products to interact with the world. I’m sure that, over time, we’ll see a consolidation in this CX/AI space, but right now, market leaders are emerging use case by use case. Over the next few months, we’ll explore a few of these leaders and discuss their optimal uses. We’ll also discuss some of our customer successes with these tools, highlight a few of our accelerators for integrating them, and discuss common challenges. 

Today, let’s focus on Deepgram. What is it? Deepgram offers high throughput, low latency audio intelligence, including both text-to-speech (human-like speech) and a compelling speech-to-text translation service. It’s the latter we want to focus on today.

According to Deepgram, their transcription technology can:

• Run multiple audio streams without losing accuracy. This means that you can transcribe all speakers on a call simultaneously. 

• Provide much faster transcription–they boast a latency less than 300 milliseconds, meaning the word text is output within 300 milliseconds of the end of the spoken word. They claim that they are up to 40x faster than traditional transcription services–10x faster than Amazon AWS.  

• Return more accurate transcriptions–30% less word error rate (WER)

• Lower the barrier to entry–boasting pricing that is 3-7x cheaper than their competitors

OK, so fast, reliable and affordable transcription sounds good–but why do we need that? Certainly transcription itself is useful for things such as quality assurance and training, but it’s bigger than just that. Most AI solutions (along with all legacy business process management (BPM) solutions) are text-driven technologies. Not many can use audio directly as an input, so getting to a near-real-time transcription is an important first step in being able to make discernments about or take action upon the content of a live conversation. So, in some ways, using a technology like Deepgram for transcription is a key that unlocks the rest of the AI world for you to leverage. Having near-real-time access to things such as transcript, summary, sentiment, intent and key-driver analysis and topic detection can allow you to enable things such as: 

• Super intuitive call routing and call escalations that can reduce resolution time and prevent customer frustration.

• Leverage real time customer sentiment to make the customer journey more interactive, responsive, intuitive…and better!

• Facilitate context-driven, timely and effective next-best-action recommendations for agents while they are live with a customer. 

• Sentiment-informed call summarization to facilitate seamless handoffs and escalations

• Better and more useful insights for post-call analysis leading to improved customer satisfaction over time.

• Improved and enriched knowledge base with more suggestive capabilities. 

• Deeper, more insightful inputs into your CDP (customer data platform) and marketing automation tools for ongoing nurturing and engagement.

Deepgram speech-to-text pairs perfectly with Twilio Voice. Like mostly all things Twilio, it will require some development effort to integrate the two offerings–but that’s why we’re here.  Ascent CX has a Twilio Voice + Deepgram Jumpstart offering that will get your transcription services working affordably, and in less than two weeks! Within the next few weeks, we’ll show you a demonstration of Twilio Voice + Deepgram in action–stay tuned!

Email us at info@ascentcx.com for more information.