Overview
Provide grounded summaries with citations from top results.
Problem
Users need concise answers; scanning multiple articles slows them down.
Solution
Optional AI-written snippets summarize top-k results and cite sources to maintain trust.
How it works
Toggle synthesis in Models. Require citations and cap tokens. Measure impact on deflection, AHT, and satisfaction; keep the retrieval grounding strict.
Who is this for
Support Lead
Admin
Expected outcomes
- Faster answers without sacrificing accuracy
- Higher user satisfaction
Key metrics
Time-to-answer (widget)
Baseline
20 seconds
Target
8 seconds
CSAT on self-serve flows
Baseline
3.9 score
Target
4.5 score
Gallery
Case studies
Travel app improves CSAT by 0.6
Grounded synthesis reduced bounce and improved clarity.
Travel SMB NA
Security impact
- Transient prompts built from retrieved articles; no PII
Compliance
- SOC2
- GDPR (data processing transparency)
Availability & next steps
Enterprise