At this year’s SIGNAL, we explored the balance between artificial intelligence and authentic engagement at the intersection of developers, innovation, and communications. From customer presentations to developer hands-on deep dives on Twilio's core products, we gave attendees the connections, tools, and resources for what they build next.
In a few days, we’re hosting the Blueprint to AI-Assisted Contact Centers with Twilio Flex and Google Contact Center AI. To get you excited, we thought we’d share this session from SIGNAL that explored how to use AI at different stages of customer interaction.
Chris Kendall, Twilio Solution Architect, kicked off this session by asking, “What would a perfect contact center look like?”
One of most important measures is customer satisfaction. We obsess over this at Twilio—communicating at the right time, on the right channel, with the right context—with the goal of creating that customer experience nirvana.
The challenge for today’s businesses to do all of that just right is becoming increasingly difficult. Customers expect you to be available on all channels at all times, and the cost to deliver that experience, especially using traditional means, is increasing. As customer expectations grow, and as companies are unable to keep up, the gap widens.
AI in the Contact Center
Contact centers are complex, imperfect environments. They are also a hive of activity, filled with repetitive work and lots and lots of data. Which makes it the ideal environment for AI and automation. If AI can deliver what we want it to, we can actually decrease costs and simultaneously increase customer satisfaction.
Sounds like a pretty good deal, right?
However, there are some roadblocks along the way. We still need human agents to create intimacy with our customers. Intimacy drives loyalty, which drives customer satisfaction. And satisfied customer spend more money. Ultimately, the question is not whether we should be using AI, but how do we apply it appropriately while still keeping that human touch.
Along with keeping the human touch, we also need to keep our human agents happy. Happy agents create better experiences for customers and stay with your contact center longer, reducing churn and costs. Which is where agent-assist AI can be so useful.
Twilio Flex, now generally available, was built to allow complete customization all the way through the stack; AI can be applied at every layer. Tangibly, you can focus on using AI before the agent interaction, such as by building chatbots or using sentiment analysis to improve routing. Or you can use AI to assist agents during an interaction or for agent training in conjunction with workforce optimization (WFO) analytics.
One of the top uses of AI in the contact center is intelligent routing, and Fabio Cardenas, CEO of Sundown AI, joined the session to explain how to get it right.
Sundown AI builds and implements machine learning algorithms for routing. These algorithms are great for optimizing an existing routing strategy because they increase happiness between customer and agent pairings without changing any of the front end systems. As Cardenas explains, the goal is optimization, not modernization.
Happiness routing applies collaborative filtering to match tasks with the most appropriate agent.
They take into account sentiment, CRM data, forecasting information from workforce management systems, and more, increasing the amount of data to feed these models so they can make better predictions.
So, what do you need to build a solid model for routing? A LOT of data! Specifically: at least 20,000 relevant interactions, 50 agents, and one to three months of iterating on the model.
See how Sundown’s solution works with Twilio Flex step-by-step (from minute 10:23) in the session video.
Let’s Have a Conversation About Conversations
Adam Champy, Senior Product Manager at Google, closed out the session with a conversation about conversations. The contact center is a conversational interface of back and forth communication. When “How can I help you?” is the opening line, context and understanding lead to a true conversation. Doing this well with AI has been a challenge since the concept was conceived.
The truth is, people connect to a contact center at critical moments. They’re usually not calling a contact center because something is perfect. Maybe they want to buy something, which may be a moment that’s happy (and in that case, you want to keep them happy). Or perhaps they want to complain about something or return a product, and it’s important to respond to those moments correctly.
Whether the customer is talking with a virtual system or having an engagement with an agent, you want these to be personalized and enriched conversations. You don’t want to give your customer a wrong answer when they’re reaching out in a moment that matters.
Natural language understanding and synthesizing speech in a human way enables richer conversations with virtual systems. Google has researched this extensively. If you sound like a computer, users will treat your systems like computers. They’ll ask very short queries with very limited scope. If you sound and engage more human-like then you’ll get deeper and more rich conversations on the virtual side.
Google + Twilio + You
Google and Twilio are working together to use machine intelligence to make customer interactions not only better for customers and agents, but better for businesses. If the systems are too complex or too costly, they won’t be good for businesses.
Their products of focus are:
- Virtual agent - enabling chat bots to have rich conversations on any channel
- Agent Assist - helping agents during an interaction, giving them super powers
- Topic modeling - understanding contact center data using analytics and insights
If you’re currently asking any of these questions:
Can we answer any question faster and better?
Can we understand each customer and our contact center better?
Can we offer a multichannel, personalized customer experience 24/7?
AI is available for you to do this in real and transformative ways today.
Watch the demo (at minute 19:47), see what’s possible, and tune in for the webinar to dive deeper.
Join Product Managers from Twilio Flex and Google Contact Center AI in a live webinar and learn how to build your AI-assisted contact center roadmap.
November 8 @ 10 AM PT/1 PM ET. Register here. You can also access the webinar on-demand after Nov 8.