Fexa Uses Generative AI to Streamline Customer Service Orders

Executive Summary

Mission Cloud designed a solution to enhance Fexa's facilities management platform using generative AI. The goal was to reduce maintenance service calls by providing an interactive chatbot interface that offers troubleshooting suggestions. Fexa wanted to streamline the customer reporting process and empower customers to resolve common issues on their own. Mission Cloud used Amazon Bedrock, Anthropic Claude 2 and Amazon Kendra for the AI model, while Streamlit facilitated the front-end interface and LangChain served as an orchestrator to bring everything together. The solution saved time and money for Fexa and its customers, inspiring further innovation and optimization within the platform.

The AI-powered chatbot provides immediate solutions for common problems, reducing the need for maintenance service calls. Fexa and its customers can benefit from streamlined processes and cost savings. The success of this project has opened doors for further innovation and optimization within Fexa's facilities management platform.

Fexa Uses Generative AI to Streamline Customer Service Orders

Background: Leveraging a Trusted Relationship for New Solutions

Fexa and Mission Cloud have established a strong and successful partnership, initially centered around cost optimization and cloud operations through Mission Cloud's managed services. Over time, Fexa expanded its use of Mission Cloud's services to accommodate its growing business needs. During their collaboration on an enterprise data warehouse Migration Acceleration Program (MAP) engagement, Fexa recognized an opportunity to leverage generative AI to enhance its facilities management platform and provide a more streamlined customer experience.

Fexa's platform serves as a comprehensive solution for retail and restaurant chains, offering efficient management of location repair and maintenance. Renowned brands such as Tesla and Crate & Barrel rely on Fexa's state-of-the-art software to ensure timely and cost-effective handling of their maintenance operations. The platform, hosted on AWS, enables clients to manage repairs, maintenance, work orders, vendor contracts and more.

With the anticipation of significant growth in the next three years, Fexa wanted to scale its operations efficiently and securely and turned to Mission Cloud's expertise in generative AI. 

Challenge: Using Generative AI to Streamline Service Requests

The goal was to leverage AI capabilities to reduce the number of maintenance service calls by providing an interactive interface that offers valuable suggestions to avoid costly work orders. One of the primary challenges Fexa wanted to overcome was the time-consuming and inefficient process of customers creating work orders for maintenance and repair requests. Customers would have to spend valuable time explaining their problems in work orders only to discover there was a simple solution they could implement without a technician, or their incomplete description would result in the wrong service call being made. These costly inefficiencies would lead to delays in receiving answers and resolutions. Additionally, Fexa found that many of these reported issues could be resolved with simple solutions, rather than requiring extensive maintenance work.

The customer wanted to implement a solution that would allow customers to quickly resolve common and straightforward issues without overwhelming the maintenance system with unnecessary work orders. This would provide customers with a seamless and efficient way to troubleshoot and resolve problems on their own, without the need for extensive technical knowledge or tools. Fexa already had a frequently asked questions (FAQ) feature in place, but they wanted to leverage AI to make it more comprehensive and applicable to a wider range of cases.

One of the unique considerations in this project was the need to integrate the chatbot solution seamlessly into Fexa's existing workflow and work order system. The chatbot had to be able to generate troubleshooting suggestions on the fly, based on beginner-friendly data sources that were technically relevant to the specific problem at hand. The challenge was finding a balance between providing helpful and accurate information while ensuring the responses were tailored to a novice audience, rather than assuming advanced technical knowledge.

The complexity of this challenge would require the Mission Cloud team to invest significant effort in setting up the environment, performing business-specific tasks and handling the necessary administrative work to ensure a smooth coding process. While it would have been easier to build a technically advanced chatbot that catered to experienced users, the team had to navigate the intricacies of finding and using data sources suitable for beginners. This requirement added complexity and would make the data science aspect of the project more challenging, but the team began working on a solution that met Fexa's requirements.

Solution: Combining Bedrock, Kendra and Claude 2

The first step was to find a significant data source for training the AI model. The Mission Cloud team scoured the internet for general maintenance knowledge related to various mechanical systems, including HVAC, plumbing and carpentry. 

The team then designed a cloud architecture that would support the AI-powered chatbot solution. They utilized Amazon Bedrock, along with Anthropic Claude 2, which served as the primary Language Learning Model (LLM) for the project. Claude 2's large context window allowed for more comprehensive prompts and answers, while the seamless integration between Amazon Kendra and Amazon Bedrock accelerated the development process. To support the front-end interface, the team employed LangChain, an open-source Python technology, and Streamlit, a Python virtual user interface that could be deployed on the web. This combination facilitated the integration of the AI chatbot into the platform.

To ensure the solution was highly repeatable and easily manageable, the team leveraged Amazon CDK, a code-as-infrastructure platform. By coding the environment using Amazon CDK, they enabled the client to deploy and manage the solution independently, minimizing the need for manual configuration.

A unique aspect of this project was integrating a work order form into the AI chatbot. The client wanted the details provided in the work order form to influence the chatbot's responses. To accommodate this requirement, the team created a mock work form that captured relevant data and provided initial checklist tasks for clients to test. Clients could then engage with the chatbot and ask questions related to the specific checklist items.

Results: Enhanced Functionality and Accelerated Service

This solution enables both Fexa and its customers to save time and money by allowing self-serve functionality and empowering clients to resolve common maintenance issues independently. Clients will no longer need to wait for a technician or incur unnecessary service call expenses for simple problems.

The client's response to the solution was overwhelmingly positive. Key stakeholders expressed their satisfaction, stating that the solution aligned well with their goals and expectations. The success of the project sparked discussions about future enhancements and expansions. Fexa is now exploring opportunities to leverage the solution for vendor support, aiming to streamline more complex issues that require the expertise of a technician.

The AI-powered chatbot provided faster solutions for common problems and inspired Fexa to consider additional functionalities and improvements. The success of this initial implementation has opened doors for further innovation and optimization within Fexa's facilities management platform.

AWS Services Used

During the engagement with Mission Cloud, Fexa leveraged a variety of AWS services, including: 

• Amazon Bedrock
• Amazon Kendra
• Anthropic Claude 2
• Amazon S3
• Amazon Elastic Container Service (ECS)
• AWS Cloud Development KIT (CDK)

About the Customer

Fexa is the cutting-edge facilities management platform that revolutionizes how retail and restaurant chains handle location repair and maintenance. Tesla, Crate & Barrel, Five Guys Burgers, and Dollar General place their confidence in Fexa’s state-of-the-art software to get the job done on time and on budget.