Normalized claim
Time: 30% decrease
The experiment showed that core processing time was reduced by approximately 30%.
SIGNAL IDUNA: Gemini-powered AI knowledge assistant for faster insurance customer service
Use case typeKnowledge managementUpdated Jun 13, 2026
SIGNAL IDUNA, a German health insurer, built an AI knowledge assistant to help service agents resolve complex customer inquiries faster and more accurately across thousands of internal policy, procedure, and tariff documents. The assistant supports natural-language question answering for health insurance service operations and was developed with Google Cloud alongside BCG and Deloitte.
- Organization
- SIGNAL IDUNA
- Industry
- Insurance
- Location
- Germany
- Published
- March 2025
Reported outcomes
Time: Approximately 30% lower
Time & speed
Impact: Approximately 73% higherImpact: Approximately 98% higherImpact: Approximately 24% higherTime: Approximately 6 seconds
Catalog median for time & speed deployments: −50% across 295 reported metrics. Compare benchmarks →
Why do we believe this?Outcome claims, sources, and evidence checks
Normalized claim
Quantified impact: 73% increase
Case closure rate increased from 73% to almost 98%, or about 24 percentage points.
Normalized claim
Quantified impact: 98% increase
Case closure rate increased from 73% to almost 98%, or about 24 percentage points.
Normalized claim
Quantified impact: 24% increase
Case closure rate increased from 73% to almost 98%, or about 24 percentage points.
Normalized claim
Time: 6 seconds increase
Response quality improved based on expert evaluations, with average response time around 6 seconds.
Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks
- Customer
- SIGNAL IDUNA
- Provider
- GCP
- Maturity
- Exploring
- Linked source
- Google Cloud Blog
5 Pro, Vertex AI ranking API, Vertex AI Gen AI evaluation service, Vertex AI Experiments, Google Cloud Document AI Layout Parser, BigQuery, Looker, and Google Cloud SQL for PostgreSQL with pgvector
Customer identity supportedSource describes one deploymentMaturity supported
Primary read
Use case focus
Showing 3 of 3
- 1Knowledge management
- 2Customer service
- 3Retrieval augmented generation
- SIGNAL IDUNA built a RAG-style AI knowledge assistant on Google Cloud.
- The system uses Gemini 1.5 Pro, Vertex AI ranking API, Vertex AI Gen AI evaluation service, Vertex AI Experiments, Google Cloud Document AI Layout Parser, BigQuery, Looker, and Google Cloud SQL for PostgreSQL with pgvector.
- PDFs are parsed and chunked, vectors are stored in Cloud SQL, queries are augmented and reranked, responses are streamed to service agents, and evaluation metrics are tracked in BigQuery and Looker.
Architecture
The assistant uses Document AI Layout Parser and PDFPlumber to extract text from PDFs, then chunks and embeds content with Gemini/Gecko-based processing and metadata enrichment. Vectorized chunks are stored in Google Cloud SQL for PostgreSQL with pgvector. User questions are expanded through query augmentation with Gemini 1.5 Pro, retrieved from cache or vector store, reranked with Vertex AI ranking API, and assembled into final prompts. Responses are streamed to agents, while Gen AI evaluation service, Vertex AI Experiments, BigQuery, and Looker support automated evaluation and dashboarding.
Sources & evidence1
Evidence: Medium50/100Evidence strength
- Customer explicitly identified
- Deployment status explicitly supported
- Quantified outcome available
- Technical implementation details available
Type: Blog PostPublished: Mar 13, 2025Publisher: Google Cloud BlogEvidence: VendorConfidence: Medium
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Customer evidence
Provider evidence
AI patterns
Browse the catalog
Was this useful?
Community
Comments
No published comments yet.