The world of data and analytics strives to deliver data to the consumer in the right form at the right time. While analytics has enabled the data consumer with various capabilities like Standard Report with KPIs, Dashboards, Reports delivered through Alerts & subscriptions, Self Service Analytics – users still need to understand the tool, get familiar with the report names and learn which report will help them find answers. The KPI definition and reports need to be designed as per the user profile - otherwise the user will have to run the report, put the right filter and find answers to questions.

The artificial intelligence (AI) maturity and capability to learn about the consumer sets the future in the right direction – Voice Digital Assistant brings the entire AI capability into action, where it answers questions and conveys what it can process along with its thoughts in simple Natural Language conversation. Voice Digital Assistant can learn about the enterprise user while it picks up new capabilities (like setting meetings, sending emails, status of various business processes, read KPIs). The art of possibilities with Enterprise Voice Assistant is endless as we get introduced to Voice User Interface (VUI).

With Digital Voice Assistant at the forefront of AI, let’s fast forward few years from now and look at some real-life examples given in Table 1.

Table 1: Voice Digital Assistant use cases & examples

Voice: Reshaping data consumption and user experience

Typically, users are forced to learn or change when any new application or user interface is launched. User needs to learn how to login, the menu structure, how to deal with the messages, the different clicks and status, etc. Voice User Experience (UX) implementation is the other way around -- the designers should build a voice product for the humans that can interact naturally.

There is always a collection of data sets and large number of use cases per role. Traditional reports and dashboards try to generalize these use cases and deliver them periodically. Here the user must define the problem statement, find the right data set and derive the answer for the business question. This is time consuming - the effectiveness of data and decision mostly depends on the user skill level.

The Digital Voice Assistant will learn over a period and will truly be an assistant who understands the user needs. From checking if the data is available to printing the required charts will be just a question away. For eg. Voice (Voice assistant), print the sales forecast once the data is ready. (as simple as saying “Voice, set a reminder for 5 minutes”)

Future-ready Enterprise Voice Assistant

Voice Skill will evolve over time. The voice assistant will run Business Intelligence (BI) reports and query the database. It will happen as the developer enhances the skill to add more Intents and Utterances, and each role is defined better. This needs to be seamless to the data consumer. Development and Deployment models of Voice Skills will mature rapidly.

Going beyond the Voice Skill, BI & Visualization tools will continue to serve the data consumer and will need better integration with Voice Skills. Most of the tools are already talking about Natural Language Processing (NLP) within their tool capability, which is a good sign.

Data needs to expand beyond tables and charts - Industry is moving towards specialized storage for every type of data like Files, Images, XML, NoSQL and any other unstructured data.

Data Virtualization, aptly called "Information Fabric", is going to take center stage in delivering the data across enterprise. BI tools fail when the user needs anything that is unstructured - that need will be served by the Data Virtualization technology. This will serve data in any form, from any part of the enterprise, from any location - where the user just consumes without connecting to multiple systems.

Coming back to Voice Skill - If Cloud platform adoption is complete, the capability to build a server-less function in any Cloud and integrating this with any BI tool and Data platform is already available (See Figure 1). This high-level architecture for any enterprise brings all the capabilities together and delivers a seamless experience to the user through the Digital Voice Assistant. Developing Voice Skill to access various capabilities across the enterprise can be classified as 2 categories – Access the BI tools in the enterprise, Access the Data in the enterprise.

In the architecture, Voice Skills can directly access all the BI components and similarly access all forms of data through the Data Virtualization Layer - so it only needs to understand the data (Not where it is coming from or how to access). Adaption and implementation of a Data Virtualization Layer will enable the Voice Skill to deliver the experience even if the underlying source system changes. This simplification of access and focus on Voice Skill will deliver the most effective consumer experience and answer the questions in the right time.

Figure 1: High-level architecture of enterprise voice assistants 

To demonstrate the approach to implement a simple skill like Total sales – a simple query like below can be attached to the Sales intent & utterance. This will deliver the most important KPI just like telling ‘Today’s weather’.

Future of Voice and natural conversation

Speech Synthesis Markup Language (SSML) Version 1.1 is already in action and is emerging as the standard XML based language for Voice User Interface. Few years back, cloud was looked as being locked in with a vendor and now it’s considered as only Internet - accessible to everyone from everywhere. The same evolution will happen for Voice Assistant as well. In the next few years, integration across voice assistants is inevitable.

Recent developments from the voice assistant providers are in-line to the above expectation:

• Microsoft Cortana announced that Cortana app in Android & IOS is being dropped and being integrated to Office 365¹
• Siri is focused on delivering enterprise automation²
• Alexa for Business is already in action³
• Samsung announced its Bixby can be made available across devices with just 1 line of code - Phone, Watch, TVs, Refrigerators & any other appliances⁴

How to drive enterprise Voice strategy and adaption

User interfaces have evolved from Command line interface, Graphical, Touch and now more humane Voice Interface. Enterprises needs to define a User-focused Voice Strategy with clear objectives and strong organization structure to deliver it.

Following are the aspects that should construct the Enterprise Voice Strategy

1. Users, common use cases, user experience, adaption to individual users and the environment restrictions
2. Executive sponsored Voice Organization Structure with strong governance
3. Data Platform, Data available across organization and how all will be made available
4. Determine if you are building a custom voice assistant or use 3^rd party options
5. Voice Experience Delivery platform
6. Robust on-going data collection and continuous learning process
7. Focus Groups, user survey-based testing methods
8. Monetize your investment

Voice product should deliver most of the common features that are commercially available in the market to excite enterprise consumer and drive adaption. Below are the most important features and this is not the exhaustive list.

1. Conversational behavior – go beyond question & answer
2. Support multilingual – People have become more global with fusion of languages
3. Privacy at the core
4. Identification of environment noise, music and user voice
5. Knowledge graphs to improve over time
6. Cloud native and platform agnostic
7. Integration with the enterprise ecosystem

With improvement of cloud, processing power, 5G and power of Data & AI – Voice is set to take the human-machine interaction to the next level very soon.

References

1. https://support.microsoft.com/en-gb/help/4531683
2. https://www.theverge.com/2018/9/24/17895794/apple-salesforce-partnership-ios-siri-enterprise-trailhead
3. https://aws.amazon.com/alexaforbusiness/
4. https://bixby.developer.samsung.com/

Karthikeyan Subramaniam

Managing Consultant, Analytics and AI Consulting, Wipro

Karthikeyan has 16 years’ experience in designing, developing, and delivering innovative data solutions across merchandising, supply chain, vendor management and store operations. He helps large enterprises embrace data and transform into business outcome-driven organizations.