Cutting Contact Centre Handle Time with AI-Grounded Knowledge Search

This client runs one of Canada's largest contact centre operations, supporting customers across multiple national brands. Every extra minute an agent spends hunting for an answer multiplies across thousands of daily interactions, and much of that support runs offshore, where cost tracks directly to handle time. The knowledge infrastructure underneath those agents was working against them.

• Knowledge lived in disjointed systems across brands, with no single place for an agent to find a reliable answer

• Agents fell back on tribal knowledge and outdated personal bookmarks when search failed them, so answer quality depended on who picked up the call

• One of their knowledge platforms was scheduled for decommissioning, putting a hard deadline on migrating and consolidating its content without disrupting live operations

• Long search times and inconsistent support content drove up the length and cost of offshore interactions

• Support content had no centralized authoring or management, so keeping answers current across brands was manual and unreliable

• Designed and deployed an AI-native knowledge management platform on Azure, integrating Azure AI Search with Azure OpenAI to return contextual answers grounded in Rogers' internal knowledge base

• Built generative answers that summarize lengthy support articles into what the agent needs in the moment, with every AI response traceable to approved source content

• Applied custom prompt engineering and tuning to sharpen relevance and mitigate hallucination, prioritizing utility inside live frontline workflows

• Migrated knowledge content off one of their platforms ahead of its decommissioning

• Centralized content authoring and management in Contentful, giving content teams one governed place to maintain support knowledge

• Secured access through Azure AD and SSO, and laid the platform foundation for unified knowledge system across brands

• 30 to 40% reduction in agent search time

• 10%+ decrease in agent handling time

• Generative, summarized answers live in frontline workflows, each traceable to approved source content

• Content migrated and consolidated before platform decommissioning

• Centralized content management operating in Contentful

• A secure, auditable platform positioned to unify knowledge across their brands

• Handle time fell, and with it the cost and duration of offshore support interactions. In a contact centre this size, seconds per call compound into real money.

• The tribal-knowledge workaround ended. Agents retrieve one governed answer instead of guessing from bookmarks, so answer quality no longer depends on tenure.

• Customer experience leaders gained visibility into the metrics that matter, including CSAT and handle time, with the platform instrumented to track them.

• The client now owns the foundation for one knowledge system across multiple brands. Consolidation is a migration exercise on existing rails, not a new build.

• Traceability is built in. Every generated answer points back to approved content, which is what makes generative AI defensible in a regulated, customer-facing operation.