Copilot AI Agent Improves Product Search for Customer Service Teams

A global supplier of seals and O-rings with a broad product catalog enables its customer service team to source highly specific components across a wide variety of applications and specifications.

When a customer comes to the company looking for a seal or O-ring, the request often involves a detailed set of parameters: durometer rating, operating temperature range, industry application, material compatibility, and more. Answering these questions accurately requires cross-referencing product guides, supplier catalogs, and specification sheets from multiple sources, some hosted on supplier websites, others stored internally as PDFs or spreadsheets. The more experienced a team member, the faster they could navigate these resources and match a customer to the right product. But the process was time-intensive regardless, and the depth of knowledge required meant that response speed varied across the team.

The company came to Emergent Software looking to explore how Microsoft Copilot and AI could help. Their goals were straightforward: get their team trained and comfortable with Copilot, evaluate whether an AI agent could help speed up product searches and surface better recommendations, and do so in a way that was practical and grounded in how their team actually worked.

Emergent's Copilot Quickstart engagement was the right fit for where the company wanted to start. The Quickstart is designed to help organizations move from curiosity to confidence with Microsoft Copilot, covering both hands-on user training and the behind-the-scenes environment readiness work that sets a Copilot deployment up for success.

A key part of the environment readiness process is understanding where an organization's data lives and how permissions are structured. Copilot surfaces information based on what a user already has access to, so ensuring that data governance and security policies are properly in place before deployment is essential. Emergent reviewed the organization's data environment, assessed how their Microsoft 365 permissions were configured, and walked their administrators through what they were enabling and how to manage it going forward.

Alongside the Quickstart, Emergent built a custom Copilot Studio agent tailored to the client's product search use case. The agent was designed to help customer service team members quickly find seal and O-ring recommendations by querying a defined set of supplier knowledge sources, including public supplier websites and internal product documentation. Rather than relying on a broad, open-ended search, the agent operates within a curated scope of trusted sources, giving the team more confidence in the results it surfaces.

Building the agent involved more than a single pass. Emergent worked through three to four iterations of refinement with an internal pilot group, gathering structured feedback using a user acceptance testing process developed specifically for generative AI. Because generative AI responses are naturally variable rather than fixed, the testing approach was designed to evaluate quality and usefulness rather than exact output matching. Testers submitted prompts, rated the responses, and provided feedback on what was helpful and what needed adjustment. Some refinements came from agent instruction updates, such as formatting responses to bold supplier and product names and organize recommendations in a consistent structure. Others came from user training, helping team members understand how to phrase their questions in ways that produced the best results.

The agent was made available to the full organization following the pilot, allowing the organization to continue gathering feedback and expanding adoption at their own pace.

The team now has access to a purpose-built AI agent that can accelerate product search and surface supplier recommendations faster than a manual lookup process. For customer service team members fielding specification-driven inquiries, the agent provides a focused starting point that draws on trusted sources rather than requiring them to navigate multiple catalogs and websites independently.

The agent does not replace the expertise of an experienced team member, nor was it designed to. What it does is help the team respond more efficiently to a meaningful portion of incoming questions, freeing up time and reducing the knowledge gap between newer and more tenured employees when it comes to routine product lookups.

The pilot group's response was positive enough that the company chose to make the agent available company-wide, a decision that reflects the team's confidence in what the tool delivers. Emergent continues to support the client as they expand their use of Microsoft's technology ecosystem. Following the Copilot Quickstart engagement, the company moved their Microsoft 365 and Azure licensing to Emergent, and the two organizations are in active discussions around additional cloud initiatives. What began as an introduction to Copilot has grown into a broader technology partnership.