Twilio, a company chartered in 2008 to support standards-and API-driven telephony apps, has launched “Autopilot”, a development platform that serves, in part as the Rosetta Stone for conversational intelligent assistance. Since its IPO in mid-2017, Twilio has been on a high-visibility charm offensive that has moved beyond the confines of the app developer community by identifying, describing and then fulfilling tough challenges that businesses confront as they look for the most practical and cost-effective ways to integrate the most advanced technologies surrounding messaging, routing, voice processing and natural language understanding.

In the case of “Autopilot” (whose name, in this age of autonomous vehicles, belies its actual functions), Twilio provides developers – who are often in business units of large enterprises and brands – a set of tools and test facilities to build custom bots, conversational IVRs and home assistant apps. It is tackling the “write-once-render-everywhere” challenges that have confronted developers since the days of multiple personal computing and mobile operating systems.

SIGNAL: AutoPilot’s Coming Out Party

According to Nico Acosta, Director of Product Management for Understanding and Conversational AI at Twilio, work began on the Autopilot services about two years ago. The company was determined to build a platform that simultaneously creates operational efficiencies while providing better customer experiences. Inherent within that goal, was the ability to identify where chatbots and intelligent assistants provide the right service and the right time, and to deploy at scale in well-defined workflows, said Acosta.

Autopilot is offered as a natural language service intended to leverage current investments in enterprise contact center infrastructure. The service includes Twilio’s own natural language processing (NLP) engine, a conversational application platform and an omnichannel hub to optimize conversational applications for IVR, messaging, and personal intelligent assistants.

Its design principles are to enable businesses to:

• develop intelligent bots faster by providing the Natural Language Understanding and Machine Learning resources that allow developers to focus on business logic and customer experience.

• personalize the tone and conversational style of the interaction, thanks to style sheets that allow developers to select the tone of voice, language and standard messages triggered “success or error”.

• use their own data, based on actual conversations, to train bots

• smoothly transfer or “hand off” conversations to agents when necessary.

• build applications once and deploy across any channel. Autopilot’s responses are adapted to provide the best experience on all channels: IVRs, SMS, Chat, Alexa, Slack and Google Assistant.

Developers have access to a set of programmed conversational patterns to automate customer data collection and information gathering and then offers the ability to handoff or escalate to agents if the interactions requires more complex actions. By building their own NLP, Twilio is investing in machine learning to build a training set of conversational data allowing developers to focus on business logic and customer experience.

Currently, Autopilot is available in public beta in the Twilio Console, but can be accessed via a widget in Twilio Studio and integrated into Twilio Flex.

A Defining Moment for Conversational Commerce

Twilio has a history of creative/disruptive offerings that give enterprises a glimpse of how the future should take shape. For example, back in 2012, as an upstart development platform for “over-the-top” (OTT) mobile services it came very close to transforming AT&T Mobile into a reseller, triggering a glimpse of what might have been in an API-driven world of enhanced telephony services. Much like land-line telcos and mobile network operators (MNOs), Contact Centers have entered a twilight period. Incumbents – Genesys, Cisco, Avaya – had already adjusted to the threat of the cloud-based cohort that includes NICE/InContact, Five9, RingCentral, Verint/Contact Solutions and 8×8, only to be buffeted again when Amazon Web Services, Salesforce and now Twilio introduced an expansive view of cloud-based telephony in which contact center functionality is just one of many components. The point has not been lost on the likes of Vonage, whose acquisition of New Voice Media rounds out a broad portfolio of services built on IP-Telephony and a slew of advanced services.

Twilio’s introduction of Autopilot has to be understood in the broader context of a new hierarchy where Contact Center as a Service (CCaasS) is a subset of a more general Communications Platform as a Service (CPaaS). Whether you’re Salesforce, Amazon or Twilio, you have great incentive to differentiate and offer real value to business customers. In the case of Autopilot, Twilio shows how enterprises can extend the value and life-span of old guard components like IVRs, contact centers and mobile apps. It is creative disruption.

There are bound to be kinks in the process. For example, there are significant differences in the “conversational user interfaces” involved in SMS or messaging (where emojis and carousels replace words and selections) and IVRs (where spoken words can be much richer indicators of sentiment and other context). The idea of a single library of applications is a good one, however, and must be taken very seriously as brands look for cost-effective way to support the slew of new services, modalities required to support the variety of new devices (intelligent endpoints) that people carry with them, install in their homes and use in their cars.

Categories: Intelligent Assistants, Articles