The California Department of Motor Vehicles is seeking a Low-Maintenance Omnichannel Smart Assistant/Chatbot solution to provide secure, human-centric, contextual, personalized, and expedited customer information, transaction processing, and technical support through conversational user interfaces transaction processing, and technical support through conversational user interfaces for our customers and internal teams by July 2024..

 Like most consumer-focused commerce websites, the DMV provides web customers with a chatbot to answer questions, find information on the website, and assist in connecting customers to a live contact center agent. This chatbot is named Miles. Based on keyword input from customers, Miles delivers link-based topics and calls to action that customers select in order to learn more or start an online transaction. Based on structured and unstructured customer satisfaction and engagement data, there are several challenges to the current solution:

• Customers expect a human-like, personalized interaction with Miles, seeking answers to questions specific to their DMV account, case, or transaction, which Miles cannot provide.  
• Currently, Miles provides links based on keywords. Customers seeking more personalized answers relevant to their situation are given links back to pages they likely have already visited, creating a feedback loop.
• Content is not customizable based on channel. While customers may have the same questions no matter the channel, answers must vary by channel because the type of interaction is different.
• Low chatbot customer satisfaction score of 2.1 out of 5 stars (Medallia)
• Miles gets confused by typos or anything else it wasn’t programmed for and will require customers to speak with a live agent if it is unable to parse language.
• From January 1 through December 31, 2023, Miles was able to resolve 12% of customer issues (Medallia)
• Live agents may not be immediately available for customers, introducing wait times for customers who are already frustrated with the inability to get the information they wanted when they wanted it.  
• From January 1 through December 31, 2023, 80% of Miles' comments related to these topics are tagged with negative sentiment: Question, Answer, DMV, Agent, License, Help, Person, Chat, Website, Time (Medallia)
• Other than Medallia customer satisfaction survey data, reporting is unusable as it consists of 60 pages that must be manually compiled and reviewed on a daily and weekly basis, then fed into a third-party tool to attempt to extract insights. As a result, engagement, retention and conversion data is not available.

Management of the Miles chatbot is a shared responsibility: one division owns the content strategy, and another owns the system update process. Due to the resources, bandwidth, and administrative constraints of an inherently manual process, the Miles system does not always reflect the most current information for customers. This results in:

• Website customer frustration  
• Decreasing customer satisfaction
• Increased need for live agent intervention / higher agent workload
• Longer time for agent/customer resolution  
• Staff analysts are subjected to customer satisfaction comments, including profanity
• Lack of agency and empowerment for staff, feeling that the process is broken
• Escalations and conflict resolution among the team
• Lost productivity
• Reputational damage to DMV 

The California DMV is seeking a solution that would meet the following needs: 

Requirements: 

•  Natural Language Processing for accurate understanding of complex user queries and intent, not simply linking to the website content that they’ve already read. 

• Seamless integration with existing systems and databases. 

• Omnichannel support for a consistent user experience across platforms. 

• Personalization to cater to individual user needs. 

• Regular updates and maintenance to stay current and effective. 

• Low maintenance management for easy upkeep and reduced need for manual intervention while maintaining optimal performance. 

• Ability for business and content teams to tune or train the bot(s). 

 Constraints: 

•  Adherence to data privacy regulations and security standards. 

• Cost-effective implementation within budget constraints. 

• Scalability to handle varying user loads without performance issues. 

• Compatibility with diverse devices and operating systems. 

• Availability in English, Spanish, and Chinese (Mandarin); some applications will require additional languages as a function of the statute. 

• Compliance with standards defined by the Americans with Disabilities Act (ADA) and the Web Content Accessibility Guidelines (WCAG) 2. 

Outcomes: 

• Improved customer satisfaction through quick, relevant, and contextually accurate responses. 

• Reduced workload on human agents for routine queries. 

• Increased efficiency in handling a large volume of user interactions. 

• Analytics for continuous improvement based on user interactions, conversions, website journey, chat handoff, and feedback. 

• Cost savings through automation.