From the source
Gen Digital built Norton Revamp's AI-curated News Engine on Web Search for 
- The professional reputation management platform launched in March 2026, built by a small group of engineers, and the post walks through how the News Engine was assembled.
- The team says it evaluated custom crawlers, third-party news APIs and static knowledge bases before choosing AgentCore Web Search, because crawlers need ongoing maintenance and lag real time, news APIs add vendor contracts and take data outside AWS, and static bases go stale.
- Web Search is a fully managed, MCP-compliant capability available as a managed connector on the AgentCore Gateway; an agent sends a natural-language query and gets back snippets, source URLs, titles and publication dates.
- It powers three features: a goal-aware news feed that refreshes each session, content suggestions written in the user's voice with a three-step Analyze, Refine, Post coaching flow for LinkedIn and X, and trend-aware goal setting that learns from the past four weeks of activity.
- In the event-driven architecture, Amazon Nova generates search queries from the user's goal and recent activity, an AI Learning System ranks results by relevance, recency and source credibility, and Nova Lite turns the ranked results into content suggestions.
- The stack also uses
Stability AI for image generation and Claude through Amazon Bedrock for complex reasoning.
aws.amazon.com