Case Study: Citibot reduces wait times by up to 90% with Amazon Lex

How Citibot’s chatbot search engine uses AI to find more answers

Citibot, a provider of AI-powered chat solutions for local governments, used Amazon Lex to power text‑based interactions that help residents get city information and report issues. During the COVID‑19 pandemic Citibot saw a surge in calls, longer wait times and abandoned calls, and needed a scalable, self‑service general search capability that could leverage curated FAQs and up‑to‑date content from CDC and municipal websites to provide reliable answers faster.

To solve this, TensorIoT extended Citibot’s Amazon Lex bot with a custom fallback search intent and Lambda fulfillment that queries an Amazon Kendra index populated via a Web Connector, S3 storage, and DynamoDB‑managed crawls. The integrated Amazon Lex + Kendra solution delivered specific, document‑extracted answers, supported feedback tuning and city‑specific filters, and cut wait times by up to 90%, reducing the volume of interactions handled by city staff and providing 24/7 contactless access to information.

Citibot

Bratton Riley

Co-Founder and CEO