In the month of May 2024, I delivered a speech during the Public Technology Institute (PTI) AI and Cyber Summit held in Washington, D.C. It was a pleasure to listen to several exceptional presentations from municipal governments across the nation.
An impressive practice that stood out at the PTI event — and the most remarkable technological initiative by a local government in the previous year — was showcased through the “Monty” project in Montgomery County, Maryland.
The team that presented on behalf of Montgomery County was headed by Shayna Taqi. Taqi functions as the chief officer of organizational transformation for Montgomery County within the Department of Technology and Enterprise Business Solutions (TEBS). With more than two decades of experience in leading transformative change projects, Taqi’s department is tasked with implementing strategies to enhance the understanding and adoption of large-scale innovative technological efforts. Currently, Taqi spearheads Montgomery County’s inaugural generative AI (GenAI) initiative. This technology has bolstered the county’s customer service platform, enabling residents and visitors to effortlessly access information regarding county services through a virtual chat interface on the county’s 311 website.
Having earned her B.A. from American University and receiving recognition as a Certified Public Manager from the Regional Executive Development Program at the Metropolitan Washington Council of Governments, Shayna Taqi is at the forefront of this groundbreaking project.
Following the PTI event, I made contact with Shayna to conduct an interview for this blog. My objective was to amplify the awareness of the Monty project on a national and global scale.
Dan Lohrmann (DL): Brief us on the inception of the Monty project. What were the features of Monty 1.0 at its commencement?
Shayna Taqi (ST): The original website chatbot, Monty 1.0, was rolled out in January 2021 to assist the county’s 311 team in managing the surge in service requests from residents during the COVID-19 crisis. Powered by Zammo.ai artificial intelligence technology and utilizing multi-turn conversational mapping in English and Spanish, the Monty 1.0 chatbot handled resident inquiries across approximately two dozen frequently asked 311 topics.
DL: How did the project’s vision evolve post-pandemic?
ST: During the height of the COVID-19 outbreak, Montgomery County residents were confronting extensive wait times when contacting an MC311 customer service representative. Recognizing the urgency of establishing an alternative communication channel swiftly, particularly for time-sensitive issues such as rental and financial aid, the Monty 1.0 chatbot was swiftly configured and launched on the MC311 website within a few weeks by a dedicated project team within TEBS, in collaboration with Microsoft and Zammo.ai. Monty 1.0 initially addressed the most commonly requested services by residents. Modifications were also made to the telephony system launch options to expedite response to urgent calls. Within just two months post-launch, the hold times decreased from 5 minutes to 2 minutes, and call abandonment (hang-ups) rate dropped from 25 percent to 11 percent.
Despite the successful mitigation of the escalated service inquiries encountered by the county’s 311, feedback from residents regarding the Monty 1.0 chatbot was not consistently positive since it failed to address all queries as comprehensively as a live customer service agent could. The task of expanding Monty’s knowledge database was labor-intensive and necessitated advanced technical skills, impeding 311’s ability to swiftly enhance and sustain an informed chatbot. Consequently, residents noted that the chatbot’s capabilities were constrained and struggled to understand basic requests and phrases. In direct response to this feedback, the Department of Technology and Enterprise Business Solutions for the county initiated a trial in May 2023 to evaluate the employment of generative artificial intelligence technologies as a solution to these challenges.
DL: What benefits does Monty 2.0 provide currently? How many languages can it accommodate for residents?
ST: Operating on generative artificial intelligence technologies, the Monty 2.0 chatbot enables natural conversations with residents without the need for multi-turn mapping. This chatbot can promptly address queries relating to over 3,000 311 topics — a hundredfold increase from its predecessor. The chatbot supports conversations in more than 140 languages, including the seven most widely spoken languages in Montgomery County (English, Spanish, French, Mandarin Chinese, Korean, Vietnamese, and Amharic).
DL: What technologies, including GenAI, contribute to the creation of Monty?
ST: The foundation of the Monty 2.0 chatbot lies in a synapse pipeline, consisting of data-driven workflows that feed 311 knowledge data into the Microsoft Azure platform for processing. Here, the 311 knowledge data is transformed and indexed for utilization by Microsoft’s GenAI Cognitive Search service. Zammo.ai tools facilitate the chatbot application, workflows, and user interface experiences.
Essentially, a chat-based request is initially captured and transmitted to the Microsoft Cognitive Search service through a Zammo.ai workflow. The service selects and prioritizes indexed 311 knowledge data based on the request. This data is then dispatched to OpenAI via Zammo.ai workflow, where it is structured into a coherent response using ChatGPT GenAI technology. The response is relayed back to Zammo.ai and delivered to the resident, completing the interaction within seconds.
DL: How can individuals access Monty? What are the primary advantages for Montgomery County residents utilizing this service?
ST: The Monty 2.0 chatbot is accessible on the county’s 311 website. This chatbot enables Montgomery County residents to promptly access essential information concerning numerous county services in a vast array of languages, eliminating the need to navigate extensive pages of data or contact a customer service representative. Residents can simply type or speak their query in the chat window, and Monty 2.0 handles the rest. The chatbot excels in addressing clear and concise questions, enabling the county’s 311 customer service representatives to address complex resident issues with more depth.
DL: How has the user response been thus far?
ST: Feedback on the Monty 2.0 chatbot has been largely positive since its launch in March 2024, based on the analytics gathered by the county using Zammo.ai feedback mechanisms. Resident feedback is automatically collected through thumbs up/down buttons and comment forms during each interaction with the chatbot. Monty 2.0’s usage has more than doubled compared to Monty 1.0, as evidenced by year-over-year comparisons. Generally, residents have lauded the chatbot’s expanded capabilities and enhanced functionality in comparison to the Monty 1.0 version; this feedback has also been instrumental in identifying previously unnoticed deficiencies and suggesting additional features. TEBS continues to…Absorb insights from resident suggestions and refine upgrades to the Monty 2.0 chatbot, encompassing enhancements to prompts, workflows, knowledge base information, and other backend elements. This proactive approach aims to trim down the occurrence of erroneous and unresolved queries from residents.
DL: Could you elaborate further on resident feedback?
ST: Resident feedback typically revolves around three key areas: enhancements to our knowledge base, ameliorations to the chatbot user interface, and requisitions for fresh content and functionalities. This feedback serves as a critical compass in pinpointing glitches and imperfections within our knowledge base, notably broken links, outdated content, and associated issues. Furthermore, we leverage resident feedback to steer our prompt engineering backend initiatives and uplift the chatbot’s performance, particularly its vast language model infrastructure. Concurrently, we maintain a resident-driven enhancement backlog to shape forthcoming iterations of the Monty chatbot.
The most common enhancement appeals we receive pertain to functionalities directly within the chat window. For instance, residents seek capabilities like submitting county service requests via the chatbot, transitioning conversations from the chatbot to a 311 customer service agent, and accessing location-specific service details.
While exploring additional service levels such as 911, state, and federal aid remains on the horizon, Montgomery County perpetually seeks to enhance service delivery for residents. On another note, preliminary efforts are underway to overhaul the county’s web pages.
DL: Your team showcased at the recent PTI AI and Cyber Summit in Washington, D.C., impressed me by enabling team members to present segments of the project. They exuded enthusiasm for their roles and took pride in their work. How did you foster such a positive atmosphere? Do you blend public and private sector talents?
The Monty 2.0 Project Team comprises the following professionals, image provided by Montgomery County:
Gail Roper, Chief Information Officer, Program Sponsor
Shayna Taqi, Project Coordinator
Manu Daniel, Technical Project Supervisor
Michael Zanfardino, Functional Project Manager
Tushar Parekh, Developer
Sri Harsha Kotagiri, Developer
Colin Cox, Product Manager
Skyler Grubbs, Change Management Lead
Lauren Der, Communication and Marketing Lead
ST: The in-house development of the Monty 2.0 chatbot was orchestrated by TEBS, with extensive assistance from our Microsoft and Zammo.ai collaborators. Over 10 months, our dynamic project team fostered a special bond, commencing from the project’s conceptual phase to the chatbot’s deployment. This project stood out among various others for team members who were temporarily reassigned from multiple departments within TEBS to contribute expertise in project management, business analysis, and software development. To maximize efficiency, we embraced an agile, sprint-oriented approach throughout the project.
Our sprints were intensive, entailing collaborative design, development, and validation of chatbot components among multiple team members. Rigorous vetting, involving numerous functional testing rounds and a resident focus group, refined our end product. Team members operated flexibly, assuming various roles based on project demands and phases, fostering an environment of continuous learning and courageous innovation. This diversity of skills, receptivity to feedback, and group camaraderie were pivotal to the project’s success.
DL: What lies ahead for Monty? What are your aspirations for the next one to two years?
ST: The project team is presently engrossed in significant enhancements to the Monty 2.0 chatbot, integrating county GIS functionalities to better cater to resident-specific inquiries such as voting center locations, snow removal status, and related matters. Collaboration with our Microsoft and Zammo.ai partners aims to bolster the chatbot’s performance through backend upgrades. These enhancements encompass the incorporation of Microsoft Azure Prompt Flow, a tool for refining prompts guiding the Monty 2.0 chatbot’s GenAI components. The nascent product ownership team is also scrutinizing UI/UX enhancements and knowledge base improvements to address resident-reported bugs comprehensively and enhance chatbot adoption. Furthermore, we aim to leverage the team’s expertise to tackle additional GenAI initiatives within the county domain.
Since the project’s inception, we have recognized the value of continual advancement and operational efficiencies gleaned during the COVID-19 era. We take immense pride in the team’s agility in supporting residents promptly while modernizing our 311 technology and enhancing Montgomery County’s customer service model and capabilities holistically.
About Author
