Give users relevant answers quickly with intelligent search
summary
Amazon Kendra indexes diverse sources (50+ native connectors or custom connectors) and returns extractive answers plus document lists, with relevance tuned by field boosting (1–10), sortable/facetable metadata, and query-time dynamic relevance tuning based on user attributes. Incremental learning updates rankings from clicks or thumbs feedback within ~24 hours; demo uses AWS SDK + Lambda + S3 to automate tests. Production cases include Magellan Rx call-center gains, Wall Street Journal’s 2020 voter Q&A, and RI DLT chatbot serving ~85,000 users day one.
AI-written summary · about
chapters (17)
- 0:00Introduction to intelligent search
- 1:27Barriers to finding relevant information
- 3:57Amazon Kendra overview
- 8:19Natural language vs keyword search
- 10:04Setting up Kendra search
- 12:53Optimizing search results
- 14:01Demo data ingestion and attributes
- 15:28Displayable and searchable fields
- 16:32Sortable and facetable fields
- 17:56Manual tuning with boosting
- 20:05Dynamic relevance tuning
- 20:47Incremental learning explained
- 21:33Testing with Python SDK
- 23:20Healthcare use case Magellan Rx
- 24:28Media use case Wall Street Journal
- 25:30Government use case Rhode Island
- 26:58How to get started with Kendra
related talks
28mFinding answers faster with intelligent search
30mFind Accurate Information Faster with Intelligent Search
31mModernize your call center with intelligent search using Amazon Kendra
26mEmpower your business use cases
40mLevel Up: Everything you need to know about search, ...
35mYou've got questions. We've got answers! #bbuzz
29mSolve common business problems with AI and ML
22mYottaanswers.com – Delivering Billions of Smart Answers Efficiently