We replaced traditional "search-and-click" navigation with an intelligent assistant that instantly surfaces precise answers from across 60+ information-dense webpages. By grounding the AI in official NALC HBP data, the tool eliminates hallucinations and provides plain-English responses backed by strict policy citations. This zero-friction approach transforms a complex website into a conversational experience, drastically improving findability while reducing support call volume
The primary objective of the BPR Assistant was to drastically reduce member confusion and lower call center volume by modernizing how users find information. NALC members needed a way to navigate complex union documentation instantly without waiting for support staff, so we aimed to replace simple keyword search with an intelligent, conversational "expert assistant" that answers questions in plain English while maintaining strict fidelity to official NALC policies.
In the healthcare and union space, the "hallucinations" common to generic AI tools like ChatGPT are unacceptable, creating a need for an intelligent agent that understands natural language but never fabricates information. We faced the technical hurdle of seamlessly embedding sophisticated AI into a WordPress environment while ensuring the system fetches answers exclusively from official NALC resources and verifies every claim with clickable source links.
We delivered a custom WordPress plugin powered by Google Cloud Vertex AI that creates a secure pipeline to index and search NALC data in real-time. Unlike standard chatbots, the BPR Assistant provides reliable, citation-backed answers with direct links to source materials, successfully streamlining the user journey and allowing members to self-serve information that previously required a support call.
