Cicso Voice Assistant App
Name of Organization / Company: Cisco Systems India Private Ltd., Bangalore, Karnataka, India
Category: Award for Innovation in Business Information Apps
Entry Title: Cisco Voice Assistant
Essay
Cisco Voice Assistant (Voice is the new UI)
As part of Cisco’s digital transformation, we the Data & Analytics team in Cisco IT have built a voice-enabled mobile application, as a 'workforce innovation' initiative.
Voice search is the next big frontier in consumer behavior, business visibility, and technological evolution. Gartner predicts that half of all online search will be voice-based by 2020. With the rapid increase in bots within Cisco Ecosystem, Cisco Voice Assistant positions itself as a platform for information convergence through a minimalistic UI i.e. Voice. It is designed and targeted for audience who are quite literally on the move and need access to information from a particular data service or to instruct an enterprise system to take certain actions. It is aimed to enhance productivity, efficiency, user experience and most important of all provides access to information for the differently abled workforce as well.
Design:
Cisco Voice Assistant is created as an Artificial Intelligence (AI) API service that processes to decipher a natural language question and get the answer. While it is suited for automating high frequency and low complexity answers,
it also factors in the context and persona of user and is capable of having relatively longer conversations with the user.
It is currently integrated with an iOS App, an Android App and a chatbot on Cisco’s Collaboration Platform – WebEx Teams. It will be extended to Amazon Alexa / Google Assistant, email and web widgets.
Cisco Sales Community are the early adopters of this service. It offers suite of unique AI features, targeting users’ personas including Cisco’s Executive Leadership, Sales Leadership and Sales teams, in an innovative way to keep track of their business on the go. It provides action assistance on bookings attainment and profitable growth across World Wide Sales. This becomes more imperative to bring simplicity in their daily experience given Cisco is undergoing digital business transformation. The ability to ask questions in plain English and receive a compiled business report, with near real time data, with contextual follow up questions is much appreciated by Cisco users.
Workflow:
User’s voice input is converted to text using Native Apple and Android speech recognition API service (Benchmark accurate for different dialects and accents) at the mobile front-end, which then passes the text in a call to a Python REST API service hosted on OpenShift Cisco Cloud Platform.
The text is run against a Machine Learning (ML) recognizer that was trained using RASA NLU (Natural Language Understanding). For training data, labelled samples were generated based on possible patterns. A knowledge corpus of FAQs and human created/curated Q&A were also added.
The model identifies intent and entities in the text. The application service then makes a corresponding series of API requests to fetch the data, forms an answer and returns it to the front-end, where it is voiced out.
Prometheus collects health metrics of the application service (No. of requests, errors and response time) and infrastructure (CPU and RAM) in a time series database. In the event of an unusual behaviour, Prometheus fires alerts to Alertmanager which then sends out email notification to the admin.
Details about errors and responses are logged into MongoDB for usage analysis.
TIBCO Mashery deployed on Cisco Cloud Platform is used as the API gateway that maintains secure connection and manages traffic, both inside and outside Cisco network.
Features/Capabilities:
• Role-based access
• Faceted search which allows Switching between Context parameters
• Recommendation of Follow up questions
• One shot Summary
• Telemetry and Feedback enabled for Model Refinement
• Adoption Metrics dashboard
• Health Monitoring and Alerting for Application and Infrastructure
• API management with Mashery Business Impact:
User experience – Hands-free. Convenient to interact. Near Real-time response. No. of current active users (Sales) – 3000
Credits
Savitha Kumaravel Pragathi Prabaharan Jatin Chaudhary Senthil Kumar D
Links: