A context aware digital assistant helping users get accurate information every time and anytime
Client
Customer is one of the eminent marketing organizations in the US with a distribution channel of over 400,000+ agents spanning across the entire country serving many large insurance companies. A leader in products focused on health insurance and life insurance in the US.
Business Needs
Client’s internal staff who serve several agents need to respond with answers on a live call in less than 30 seconds. Moreover, the agents may ask new questions leaving behind earlier questions. So, the internal staff need to remember all the questions and share answers appropriately and quickly. The client was looking for a solution that complemented existing web portal and helped staff to get the data quickly without navigating multiple steps.
01Discover
We conducted stakeholder interviews to understand the experiences and challenges faced by internal staff while supporting agents in relation to sharing accurate and timely information.
We discovered that the internal staff faced some friction to navigate several pages and take multiple actions to retrieve information that is being asked by the agent. Few key needs of the staff uncovered were:
- Cut down investigation time to find information
- Streamline and reduce the workflow
- Reduce onboarding time taken by new staff
- Possibility of contextual insights the internal staff can take to the agent even in that short conversation during the support call
Persona
We identified senior internal staff as one of the key persona and documented the empathy map.
02Define
We needed to find a solution that is seamless and friction-less, which gives accurate and contextual information in as less steps as possible. It was decided after brainstorming to design and implement a digital assistant based on AI which understands natural language and gives contextual information. The users can chat with the digital assistant in natural language and get needed information thereby reducing friction and time.
We derived a conversational UI suitability test tool using Google framework to score the use case suitability for a conversational interface.
Our Analysis
Test for conversational interface suitability for a sample business scenario
(More the yes’s, better the suitability to be solved using CAI)
Functional Concept
A Unified Digital Assistant that can scale to support multiple business cases across applications & channels with following capabilities.
03Design
As a first step, conversation designs were created for multiple scenarios which was validated by stakeholders to ensure the optimal conversational flow meeting the user goals.
Sample conversation designs
We designed a very intuitive virtual assistant which was easy to use for both new users and expert users. The assistant helps new users via guided steps which provide contextual hints and buttons leading them towards their end goal without having the user to apply cognitive efforts to remember next steps. Whereas, expert users can ask direct questions in natural language to accomplish their end goal bypassing all the intermediate steps. Finally, a design system was created with reference colors, fonts, icons and UI widgets for developers.
Dynamic menu is displayed based on the usage pattern to guide users in a more personalized way
Users have flexibility to ask direct questions in natural language or take help of contextual menu to accomplish their tasks
Highly customized UI widgets to display different types of rich data making it easier for user consumption
Key Capabilities
- Guided dialog
- Keywords based direct search
- Context aware
- Potential for personalized responses
- Granular insights on usage behaviors
- Bot intelligence
- Finding people
- Find specific details on people
- Nickname based search
- New applications for various departments
- Natural language understanding
- Multi-turn conversation