{"slug":"signal-iduna-builds-a-gemini-powered-knowledge-assistant-for-customer-service-agents","url":"https://findausecase.com/use-cases/signal-iduna-builds-a-gemini-powered-knowledge-assistant-for-customer-service-agents","title":"SIGNAL IDUNA Builds a Gemini-Powered Knowledge Assistant for Customer Service Agents","description":"German insurer SIGNAL IDUNA, working with Google Cloud, BCG and Deloitte, built an AI knowledge assistant on Vertex AI using Gemini 1.5 Pro, Document AI layout parsing, Google's Gecko multilingual embeddings and a PostgreSQL pgvector store. Grounded in more than 2,000 internal documents covering 600+ tariffs, the assistant lets agents ask natural-language questions, cutting core processing time by about 30% and raising the case closure rate from 73% to almost 98% in an internal experiment with 20 employees.","company":"SIGNAL IDUNA","industry":"Insurance","country":"Germany","aiCapabilities":["Generative AI","Document Intelligence"],"technology":["Vertex AI","Gemini 1.5 Pro","Google Cloud Document AI","PDFPlumber","Gecko multilingual embedding model","Cloud SQL for PostgreSQL (pgvector)","BigQuery","Looker"],"deployment":"Unknown","problemStatement":"Prior to introducing its AI knowledge assistant, SIGNAL IDUNA's service agents had to manually search thousands of internal documents across hundreds of different tariffs to answer customer questions, and 27% of inquiries required further escalation to other departments or specialists, delaying resolutions and increasing costs.","solutionApproach":"SIGNAL IDUNA, with Google Cloud, BCG and Deloitte, built a generative AI knowledge assistant on Google Cloud's Vertex AI platform, using Gemini 1.5 Pro's long-context capabilities. A hybrid of Document AI's Layout Parser and PDFPlumber extracts text and tables from PDF policy documents, which are chunked and embedded with Google's Gecko multilingual embedding model and stored in Cloud SQL for PostgreSQL with the pgvector extension. A query-augmentation step uses Gemini Pro 1.5 to rewrite and expand user questions, a query cache and the Vertex AI ranking API support retrieval, and the Gen AI evaluation service in Vertex AI, together with BigQuery and Looker dashboards, evaluates response quality.","businessValue":"In an experiment with 20 employees, the AI knowledge assistant reduced core processing time (information search and response formulation) by approximately 30% and increased SIGNAL IDUNA's case closure rate by about 24 percentage points, from 73% to almost 98%, with the system achieving an average response time of about 6 seconds.","evidence":{"band":"high"},"sourceUrl":"https://cloud.google.com/blog/products/ai-machine-learning/how-signal-iduna-supercharges-customer-service-with-gen-ai/","dates":{"publishedAt":"2026-10-10T05:46:04.315Z","publishedAtSource":"pipeline","updatedAt":"2026-10-10T05:46:04.315Z"},"license":"Open for reading and citing with a link to https://findausecase.com/use-cases/signal-iduna-builds-a-gemini-powered-knowledge-assistant-for-customer-service-agents. Bulk republication requires permission."}