Municipality of Panama Improves Inquiry Response Times by 80% with Grupo TX’s Generative AI Solution on AWS

Construction projects in the Municipality of Panama are governed by complex regulations, and failure to comply can result in fines and costly delays. Citizens, engineers, architects, and builders frequently contact the municipality’s citizen service center with related questions. Many of those inquiries required staff to conduct in-depth research or consult with experts in the Department of Works and Construction, leading to long wait times and limiting the number of citizens that employees could help each day.

Inconsistent responses from staff members created a greater risk of regulatory noncompliance due to information gaps. Communication challenges were compounded by the frequent use of colloquial Spanish, making it difficult for staff to understand and communicate technical information clearly. The municipality looked for ways that technology could help them improve the customer experience without exceeding fiscal constraints.

To better equip staff members with accurate answers about construction regulations, the Municipality of Panama turned to long-time technology partner Grupo TX, which designed a specialized generative AI solution. With the contact center’s technology already built on Amazon Web Services (AWS) and integrated into the municipality’s main operating platform, Grupo TX was able to embed a conversational chatbot directly into the system used by employees, accelerating adoption. The generative AI-powered chatbot on AWS uses a Retrieval-Augmented Generation (RAG) architecture, enabling staff to query the comprehensive regulatory knowledge base and receive instant answers. Designed in Spanish, the chatbot meets local needs and accounts for informal or colloquial phrasing. “It can be difficult to interact with traditional artificial intelligence and ask questions the way technology understands,” says Eduardo Javier Serra, manager of generative AI at Grupo TX. “With generative AI on AWS, we can build a system that understands natural conversations and provides accurate, context-specific answers.”

Employees take advantage of a web application interface to access the chatbot, which uses Amazon Lex to interpret questions and then routes the employees through a two-tiered response system. If the query matches a known frequently asked question, the system returns a direct answer. For less common questions, the system escalates the query to Amazon Bedrock, which relies on a pretrained large language model to generate a tailored response based on the full regulatory corpus. All documentation is securely kept in Amazon Simple Storage Service (Amazon S3), with AWS Lambda facilitating backend interactions between the chatbot and the information repositories. 

Working closely with AWS solution architects, Grupo TX built the solution using the AWS Well-Architected Framework (WAF) with fine-grained identity and access management roles and AWS Key Management Service (AWS WAF) encryption, ensuring security and compliance. To maintain appropriate use in a government setting, the chatbot also includes guardrails that prevent it from generating responses on sensitive or political topics, keeping the tool focused on its core mission of supporting regulatory guidance. 

Since launching the generative AI chatbot, the municipality has more than doubled the number of inquiries each customer service representative can handle per day. Average response time for retrieving regulatory answers has dropped from about 15 minutes to less than 3 minutes—an 80 percent improvement—freeing staff to serve more constituents with greater consistency and ease. Access to regulatory knowledge has also been democratized. Where once only a few specialists had access to the right documents or details, all front-line contact center employees now have instant access to accurate, searchable guidance.

The chatbot’s performance has remained reliable, with 99.999 percent uptime and consistently fast processing even during peak periods. This AI-driven system has already helped simplify operations for staff, improve the constituent experience, and set the stage for future enhancements. “It used to be that sometimes only a few people could find information about specific regulations,” says Luis Carballeda, director of Works and Construction at the Mayor’s Office of Panama. “Now, thanks to the newly implemented chatbot, it has become much easier.”

Serra adds, “Cost optimization was a priority for the municipality, and we built that into the solution’s structure from day one by having the chatbot pull the most frequently needed information from FAQs and only escalating to Amazon Bedrock for less common questions. The chatbot has delivered significant staff efficiency gains, enabling the municipality to achieve a return on investment of 150 percent in just six months.”

Looking ahead, the municipality may expand its use of the technology to support additional services and provide new channels through which citizens can access information. “There’s a great opportunity to improve service to citizens with technology and specifically AI,” says Antonio Alvaro, director of technology services at Grupo TX. “By identifying a high-impact area and providing gains early, organizations can make better-informed decisions as they expand to other use cases.”