Planning

SoftServe, with its strong expertise in AWS solutions, stepped in to address ServiceTrade's challenges. After conducting a thorough analysis of ServiceTrade's business operations and infrastructure, SoftServe identified AWS — using Amazon Bedrock, LangChain, RDS with vector extension, EC2, and Streamlit — as the ideal solution.

The goal was to create a PoC using AWS to demonstrate the potential for improved cost-effectiveness and operational efficiency.

Specifically, the three business goals ServiceTrade wanted to accomplish through its partnership with SoftServe included:

Project

The ServiceTrade and SoftServe teams included the client’s chief technology officer (CTO) and vice president of AI, and SoftServe’s account executive, delivery director, project manager, business analyst, lead data scientist, and associate data scientist.

From the start, SoftServe provided ServiceTrade clear guidance about the boundaries and intent of the PoC. SoftServe also counseled ​​ServiceTrade through the AWS funding process. Within weeks of the PoC kickoff, the client had already seen how quickly and efficiently SoftServe managed the project.

The success criteria for this project included:

• Deployment of an LLM-enabled environment on Amazon Bedrock.
• Equipment summaries that contain meaningful insights.
• Complete documentation for the architecture, codebase, and operational procedures for the LLM-enabled environment.
• Scalability for further implementation and functionality extension.

SoftServe initiated ServiceTrade’s project with a collaborative approach, engaging closely with the client through design thinking sessions to identify high-value use cases and delineate the scope for PoCs. The PoC phase comprised of these key deliverables:

• Comprehensive report about generated activities
• Roadmap for a production and pilot project​
• Technical solution design and documentation
• LLM-enabled environment capable of generating activities
• PoC deployed to AWS EC2

The fast pace of the delivery allowed SoftServe to give ServiceTrade the summarizing portion of the project and extract the equipment maintenance schedule using a retrieval augmented generation (RAG) component.

Results

Delivered in less than six weeks, the implementation of the first Gen AI solution opens the door for ServiceTrade to help its customers transform their operations by improving technician productivity and effectiveness.

Specific solution outcomes include a:

• Validated approach of equipment history summarization through Gen AI, which drastically reduces and optimizes the time technicians spend on it. The anticipated reduction to review specific equipment history by field technicians enhances overall operational efficiency. This is vital because field technicians are ServiceTrade’s customers’ scarcest resources​​.
• Validated approach for a recommended maintenance schedule from the manual using RAG. The automated maintenance schedule extraction reduces the likelihood of errors and significantly enhances accuracy and adherence to those maintenance schedules. Additionally, the recommended maintenance schedule is now aligned with other ServiceTrade solutions, such as “tasking.”​
• Prompt evaluation pipeline.

SoftServe's client-centric and conversational approach also enabled ServiceTrade to be an active participant throughout the project. Consequently, the client understood the changes and will use the new capabilities of its AWS-powered platform for other already-identified use cases.

Conclusion

SoftServe's expertise in AWS and Gen AI solutions, combined with its empathetic and problem-solving approach, enabled ServiceTrade to overcome its challenges, paving the way for continued growth and success.

ServiceTrade co-founder and CTO Brian Smithwick states:

Get insights from Smithwick in a Q&A conversation, where he explains ServiceTrade’s Gen AI journey, outcomes, and lessons learned. To read the interview, click here​​.