Signal 2019 - Twilio helps contact centers get on top of conversations

Phil Wainewright Profile picture for user pwainewright August 6, 2019
At Signal 2019 in San Francisco today, communications services platform Twilio introduces new conversational computing APIs

Business man hand with mobile phone and app icons © warakorn -
(© warakorn -

Communications services platform Twilio is announcing new conversational functionality for customer contact at the opening of its Signal 2019 annual conference in San Francisco today. These will make it possible for developers to build applications that can open a multi-channel conversation from a mobile notification — for example to respond to a change of delivery or flight time — and introduce real-time sentiment and content analysis of voice conversations, to help handle calls more effectively.

Today's conference keynote shows off the new Conversations function in one of those complex live demos that is a signature element of Twilio events. A mobile notification comes through of a canceled flight. Now what do you do? Instead of having to go to a website or phone a call center to work out what to do next, wouldn't it be great to just respond to the notification and launch a message thread that gets you on the next best flight? That's the promise of Conversations, says Scott Fallon, VP of Product Marketing, who briefed us ahead of today's announcements:

[Customers] don't want to shift over to another channel, they just want to respond. You can get these three- and four-way conversations going and you just want to resolve it right there.

Bringing mobile notifications into the broader web of messaging and contact channels is something Twilio's customers have been asking for to improve the experience for their customers, explains Fallon. People often try to respond to a notification even though the traditional setup isn't able to handle replies.

A lot of what's driving this, we're seeing that customers want to do this with communications that heretofore have been notifications. In many cases the customer will respond back, even though the company didn't intend a conversation, so it just falls on the floor — no one deals with it.

The Conversations API allows more cases where customer wants to have a back-and-forth conversation.

Those conversations can include multiple participants across several different channels, he adds, including group SMS, MMS, Chat and WhatsApp. For example, a financial adviser speaking to a provider may want to check a fact with the client and then get approval from compliance colleagues to send out a proposal. The Conversations API makes it possible to join up and orchestrate these different elements while keeping a record of the complete interaction. Fallon sums up:

They're often long-lived conversations and it was really cumbersome. We've abstracted all that away.

The second announcement is Media Streams, an API that brings AI services, such as natural language processing, keyword spotting and sentiment analysis, to live voice calls. It's already common to use machine learning services in contact center applications to provide contextual help and support during message exchanges. The new service now extends that real-time analysis into voice interactions.

The new API can support capabilities such as real-time transcription, voice biometrics and sentiment analysis. So for example, a supervisor might be alerted based on sentiment analysis that an agent needs support to help resolve a call. Fallon explains:

Sentiment analysis can alert a supervisor of the need to jump in ... It's much more cost-effective and efficient. You can have an effect on that customer conversation in real time.

Twilio is partnering with Google Cloud and Gridspace at launch to integrate their machine learning resources into the API. It also plans to add access to resources such as Lex and Transcribe from Amazon Web Services. Other third parties will offer services such as voice biometrics. Ecosystem partnerships like these are an important part of the offering, says Fallon.

One of the benefits to our customers is we're able to develop relationships with other leaders, to get them to more natively support what we're doing.

The point here is, you don't have to figure all this out yourself.

My take

There are two interesting trends highlighted here. One is the increasing convergence of messaging channels, bringing mobile notification into the same conversational computing ambit.

The other is the introduction of AI support during voice conversations. We're starting to see here the mainstreaming of AI augmentation — the use of AI to support people at work and, if on the track forecast by investor Gordon Ritter in his notion of 'coaching networks' — also collaboratively learn from their behavior.

A grey colored placeholder image