Speech Technology’s Key Role In Mobile Cloud: Interactions LLC


Speech technology is a primary driver in the advancement of the mobile cloud and related big data apps. While voice assistants, such as Siri (Apple), Assistant (Google) and Cortana (Microsoft) have captured considerable attention, another company, Interactions LLC, has been a leader in bringing voice assistants into the world of meaningful enterprise usage. We recently spoke with exec’s from Interactions who explained their very focused business strategy and growth plans.

Interactions’ Business Objective

The company’s objective is to automate and improve communications between companies, large enterprises in particular, and their customers. This starts with the most basic form of communications, speech. As Phil Gray, EVP Business Development for Interactions explained, if you call into Hyatt Hotels to make a reservation, you’re actually speaking with a hosted Virtual Assistant provided by Interactions.

Most people have dealt with speech recognition over the telephone, for example when calling into customer service for a product they’re having a problem with. They’re undoubtedly familiar with the situation where the system has trouble processing something they say and, after the second or third try, responds, “Please hold on and I’ll transfer you to a representative.”

Interactions is devoted to making the automated customer care situation as seamless as possible and eliminating this type of frustrating contact. In the course of accomplishing this objective – providing solutions with an ultra high degree of reliability – the company has invested a good deal of time and money since 2004 (over $118 million of VC capital raised, including over $50 million in 2016), and of great significance, in late 2014, it purchased one of the leading repositories of advanced voice technology, namely the AT&T Watson operation.

An Array of Offerings

To create the seamless, reliable end user experience with automated tools Interactions has developed an array of speech-related technology capabilities:

  1. Automated speech recognition (ASR)
  2. Natural language processing (NLP)
  3. Text-to-speech (TTS), and
  4. Voice biometrics (VB)

The company refers to its primary offerings as Virtual Assistant Solutions. These solutions operate across a range of communications modalities, starting with speech. The modalities include:

  1. Voice
  2. Chat
  3. Outbound
  4. Texting
  5. Mobile

Regarding Mobile, for example, Interactions states, “Interactions Mobile Virtual Assistants integrate all available smartphone features for a rich self-service experience.” Thus an end user can make a contact by voice or text and receive a response utilizing any of the smartphone contact capabilities. Interactions also states that it uses proprietary “acoustic models that predict how words sound in a given environment, such as when talking on a mobile phone.”

Regarding Outbound, the system enables more than just simple notifications or announcements, since it allows the end user to respond and launch into a full conversation.

Jane Price, VP Marketing, pointed out the breadth of situations that Interactions deals with. She states that, “For Hyatt, for example, we’re handling reservations, which is mission critical. With Humana, we are dealing with elderly patients who have to complete forms over the telephone, that can involve a 20-minute long session that we have automated.”

“Human Assisted Understanding”

Interwoven with these technologies is a key distinguishing capability that the company has developed, which it calls – Adaptive Understanding (AU). AU incorporates Human Assisted Understanding (HAU).

Interactions has added what it calls iProxy between the applications and the ASR engine. The proxy is designed to distinguish, using machine intelligence, when a call cannot be handled adequately by the ASR engine. It can route the call to a live analyst in real time, who can then add human understanding to satisfy a caller’s request. Gray states, “We’ve been using a combination of human and artificial intelligence for a decade.”

Why This Is Difficult – And How Scalable Is It?

The company is handling over one billion transactions per year. The company currently focuses on eight primary verticals, of which Gray mentions hospitality, insurance and telecoms as areas where they have been making significant penetration. Other segments include: healthcare, financial services, utilities, retail and technology.

What about barriers to entry? Gray emphasizes that there are significant barriers to providing “an omni-channel assistant.” In addition, a significant barrier is the amount of detailed knowledge required of the tasks being aided in each vertical. Gray states, “We have to be deeply integrated into our customers’ operational systems.”

As for scalability, Price points out that, “The worldwide spend on customer service agent activities is $467 billion per year, providing a tremendous growth opportunity.” Since the company believes that it has major growth opportunities in its basic business, including international expansion, it is currently not focusing on other areas. One of these others would be addressing the SMB (small-to-medium business) market. Also they don’t provide ancillary support to agents, such as access to detailed company databases for responding to queries – their focus is determinedly on the quality of communication with the end user, and the end user experience.

Interactions’ Purchase of AT&T Watson – Curo Platform

In December 2014, the company purchased the AT&T Watson technology platform. Interactions acquired various patents and the working engineering team, while AT&T received a minority interest in Interactions. Prior to that time Interactions had had a license arrangement with AT&T for Watson technology.

AT&T Watson (or Dr. Watson) was a major project of the carrier in speech and multi-modal communications technology. The capabilities included not only Speech Recognition and Voice Response but also conversion of machine-generated information to voice or text for human use. It could also combine speech with other forms of expression, such as a spoken command and a tap on a computer screen or a gesture, as well as doing language translation.

Interactions has since extended the development of the platform and in June 2016 announced release of an upgraded and rebranded version, now called Curo Speech and Language Platform. The company describes Curo Speech capabilities as, “automatic speech recognition (ASR) and text-to-speech (TTS) solution that incorporates neural networks and deep learning to modernize customer care interactions.” Interactions offers the Curo capabilities to other vendor companies. In August it announced that it was partnering with Genesys, one of the leading suppliers of contact center solutions.

Price states that AT&T is a customer. Is the connection with AT&T impeding Interactions from doing business with other carriers? Price says no, and points out that in the company’s recent raise of $50+ million, Comcast was an investor.

Impressive Business Model

The company’s predominant business model is impressive – we’re tempted to say, almost astounding – because the continuing charges for its Personal Assistant service are success-based. As they state:

“Interactions only charges for successful transactions that add business value to our clients. A successful transaction is defined as the completion of a discrete task, such as authenticating a caller or completing a reservation.”

The company hosts the service in its cloud and therefore can identify “successful transactions.” Interactions does charge an upfront fee for development of the customer’s specific Virtual Assistant application. Price notes that the upfront fees are typically recovered out of savings in year one of the service. In addition, some services are sold on a different basis, such as the Curo platform, which is sold through distribution channels.

Gray states, “Companies have put a simple IVR (independent voice recognition) system between their customer and the agent, but the agent did most of the work. Now they can use our Virtual Assistant and the automated system does the work.”

Price adds that they have five offices and nearly 350 people. Revenues for 2015 more than doubled versus 2014. For 2016, revenues are up by about 80%. Growth is being driven by new customer acquisition, increased usage by their embedded base, low churn, and add-ons such as chat and voice biometrics usage.

Our Take: Whose “Virtual Assistant Is It?”

Overall, it would be hard not to be impressed with the growth of Interactions and the company’s ability to stick to a basic mission. Part of our interest in the company is that we have been following the evolution of Personal Assistants, since the first ones came on the market in the early 1990s and have been following the evolution of speech processing technologies for even longer.

Interactions’ approach raises the question in our minds: Whose assistant is it?

Siri or Assistant are the personal assistants to the end user. The interaction is between the user, fulfilling their requests, and the rest of the world. However, Interactions, as a personal assistant is different.

In Interactions’ case, there is a two-way communication going on. In one sense they are a personal assistant to the agent, who is somewhere off in the background, helping to lighten the load on the agent. As we understand the company’s marketing pitch, that is the Personal Assistant they are referring to – one that helps the agent.

But Interactions could also be viewed as the personal assistant to the caller – after all the caller actually interacts with the system, just as an individual makes requests of Siri. Of course, the caller doesn’t ordinarily have any idea whose automated system technology they may be interacting with.

We’ll be following Interactions closely, in part to see if there comes a point where end users – who’ve had a great experience with a Hyatt, for example – begin to get the message to other chains whose systems don’t work as well.

The company is certainly aware of the needs in the area of customer experience. It released a survey in late 2015 finding that “customers continue to express frustration with” voice recognition systems and actually try to circumvent the systems and reach live agents.

Visit their website: www.interactions.com