We have seen significant need from our customers and partners to deliver a conversational assistant tailored to their brand, personalized to their customers and made available across a broad range of conversational canvases and devices. Continuing Microsoft open-sourced approach toward Bot Framework SDK, the open source Custom Personal Assistant solution provides full control over the end user experience built on a set of foundational capabilities. Additionally, the experience can be infused with intelligence about the end-user and any device/ecosystem information for a truly integrated and intelligent experience.
We strongly believe our customers should own and enrich their customer relationships and insights. Therefore, any Virtual Assistant provides complete control of the user experience to our customers and partners through open-sourcing the code on GitHub. The name, voice and personality can be changed to suit the organization’s needs. Our Virtual Assistant solution simplifies creation of your own assistant enabling you to get started in minutes and then extended using our end to end development tooling.
The scope of Virtual Assistant functionality is broad, typically offering end users a range of capabilities. To increase developer productivity and to enable a vibrant ecosystem of reusable conversational experiences, we are providing developers initial examples of reusable conversational skills. These Skills can be added into a conversational application to light up a specific conversation experience, such as finding a point of interest, interacting with calendar, tasks, email and many other scenarios. Skills are fully customizable and consist of language models for multiple languages, dialogs and code.
At this time we are working closely with initial customers and partners in an open-source repository to bring to life the first experiences.
What's in the box
The Virtual Assistant builds on the Enterprise Template capabilities which provide foundational capabilities for conversational experiences including base conversational intents in multiple languages, Dispatching, QnA and conversational insights. The following Assistant related capabilities are provided at this time, further capabilities are planned and we'll be working closely with customers and partners to help inform the roadmap.
The Virtual Assistant extends across a broad number of industry scenarios, some example scenarios are shown below for reference purposes.
Automotive Industry: Voice enabled Personal Assistant integrated into the car providing end users the ability to perform traditional car operations (e.g. navigation, radio) along with productivity focused scenarios such as moving meetings when your running late, adding items to your task list and proactive experiences where the car can suggest tasks to complete based on events such as starting the engine, traveling home or enabling cruise control. Adaptive Cards are rendered within the Head Unit and Speech integration performed through Push-To-Talk or Wake Word interactions.
Hospitality: Voice enabled Personal Assistant integrated into a hotel-room device providing a broad range of Hospitality focused scenarios (e.g. extend your stay, request late checkout, room service) including concierge and the ability to find local restaurants and attractions. Optional linking to your Productivity accounts open up more personalised experiences such as suggested alarm calls, Weather warnings and learning of patterns across stays. An evolution of the current TV personalisation experienced in room today.
Enterprise: Voice and Text enabled branded Employee Assistant experiences integrated into enterprise devices and existing conversation canvases (e.g. Teams, WebChat, Slack) enabling employees to manage their calendars, find available meeting rooms, find people with specific skills or perform HR related operations.
Virtual Assistant Principles
Your data, your brand and your experience
All aspects of the end user experience are owned and controlled by you. This includes the Branding, Name, Voice, Personality, Responses and Avatar. The source-code to the Virtual Assistant and supporting Skills are provided in full enabling you to adjust as required.
Your Virtual Assistant will be deployed within your Azure subscription. Therefore all data generated by your assistant (questions asked, user behaviour, etc.) is entirely contained within your Azure subscription. See Cognitive Services Azure Trusted Cloud and the Azure section of the Trust Center more specifically for more information.
Write it once, embed it anywhere
The Virtual Assistant leverages the Microsoft Conversational AI platform and therefore can be surfaced through any Bot Framework channel – e.g. WebChat, FaceBook Messenger, Skype, etc.
In addition, through the Direct Line channel we can embed experiences into Desktop and Mobile Apps including devices such as Cars, Speakers, Alarm Clocks, etc.
Enterprise Grade Solutions
The Virtual Assistant solution is built on the Azure Bot Service, Language Understanding Cognitive Service, Unified Speech along with a broad set of supporting Azure components meaning that you benefit from the Azure global infrastructure including ISO 27018, HIPPA, PCI DSS, SOC 1,2 and 3 certification.
In addition, Language Understanding support is provided by the LUIS Cognitive Service which supports a broad set of languages listed here. The Translator Cognitive Service provides additional Machine Translation capabilities to extend the reach of your Virtual Assistant even further.
Integrated and Context Aware
Your Virtual Assistant can be integrated into your device and ecosystem enabling a truly integrated and intelligent experience . Through this contextual awareness more intelligent experiences can be developed and deliver further personalisation than otherwise possible.
3rd Party assistant integration
The Virtual Assistant enables you to deliver your own unique experience but also hand-off to the end-users chosen Digital Assistant for certain types of questions.
Our Virtual Assistant architecture is flexible and can be integrated with existing investments you may have made into device based Speech or Natural Language processing capabilities and of course integrate with your existing back-end systems and APIs.
Adaptive Cards provide the ability for your Virtual Assistant to return User Experience elements (e.g. Cards, Images, Buttons) alongside text base responses. If the device or conversation canvas has a screen these Adaptive Cards can be rendered across a broad range of devices and platforms providing supporting User Experience where appropriate. Examples of Adaptive Cards can be found here with information on Rendering options in the documentation here.
In addition to the base assistant, there exists a broad set of common capabilities which require each developer to build themselves. Productivity is a great example where each organisation would need to create Language Models (LUIS), Dialogs (Code), Integration (Code) and Language Generation (Responses) to enable popular Calendar, Task or email experiences.
This is then further complicated by the need to support multiple languages and results in a large amount of work required for any organisation building their own assistant.
Our Virtual Assistant solution includes a new Skill capability enabling new capabilities to be plugged into an custom-assistant through configuration only.
All aspects of each Skill (Language Model, Dialogs, Integration Code and Language Generation) are completely customisable by developers as the full source code is provided on GitHub along with the Virtual Assistant.
The Virtual Assistant solution is available in this GitHub repository which is updated regularly by the Virtual Assistant team. More detailed documentation is available in the same repository and issues/feedback can be provided directly through the GitHub feedback mechanisms. If you have a key scenario that you'd like to discuss with us please fill in this form and we'll be in touch.