Helping get more value out of phone calls

by Max Smolaks 8 August 2019

Popular cloud-based communications platform Twilio has enabled its customers to apply machine learning to live voice calls.

Media Streams is a brand-new set of APIs that makes it easy for developers to layer AI-based functionality, like natural language processing, keyword spotting and sentiment analysis, on top of calls in real-time.

The APIs can also tap into AI services from third-parties: some of the first examples include real-time transcription with Google Cloud, call center optimization with Gridspace, and chatbots built with AWS and Amazon Lex – using the same natural language understanding technology that powers Alexa.

Media Streams has been launched in public beta at Signal, Twilio’s annual conference in San Francisco.

The company enables customers to use standard web languages to build voice, VoIP, and SMS apps via web-based APIs. It was established in 2008 and went public in 2016 on a valuation of $2 billion – which means some people will inevitably refer to it as a unicorn. Twilio’s customers include AirBnB, Dell, eBay, Salesforce, Twitch, Uber and Lyft, to name a few.

Back in the nineties, building corporate communications platforms was a complex and expensive task, requiring proprietary hardware from a handful of major ‘unified communications’ vendors. But then the cloud came along, and now anyone can assemble their own communications stack from widely available software tools, hosted on cheap servers in public cloud data centers. Such virtualized stacks can then be easily integrated into websites, apps and services.

Most of Twilio’s communications platform lives within AWS.

The new Media Streams APIs expand Twilio’s functionality, reimagining it as a high-quality data source. They give developers instant, secure access to a raw audio stream of every Twilio-powered phone call; this should enable developers to build ever more advanced services that rely on AI, like real-time sentiment analysis that quickly tells you what thousands of customers think.

“Advancements in natural language processing, sentiment analysis and other AI technologies have created huge opportunities for businesses to improve their customer engagement in real time,” said Chee Chew, chief product officer at Twilio.

“Businesses need access to conversations as they are actually happening. With Media Streams, we’re making it easy for businesses to apply their own AI to voice calls or integrate third party solutions with a few lines of code.”