Personalised news research

Source discovery and brief generation for editorial teams.

Editors needed to follow many sources, identify relevant developments and turn them into briefs for different audiences.

A pipeline combining search, article collection, topic selection and LLM-assisted summaries to support news research and publishing.

• Processing thousands of news sources generates massive data streams. We needed infrastructure that could handle this volume while maintaining lightning-fast response times.

• Every user has unique needs. For example, a fintech blogger cares about different trends than a healthcare marketer. Our AI had to learn these nuances and deliver spot-on recommendations across dozens of industries.

• Generic AI suggestions feel robotic. We needed our content recommendations to match each user's unique style, tone, and audience expectations.

• Some users want hourly updates during breaking news cycles. Others prefer weekly digests. Building a system flexible enough for everyone was crucial.

• A Python and PostgreSQL pipeline collects and processes incoming news articles, providing the material used for topic selection and brief generation.

• Developed an intuitive preference system where users define their interests in plain language. Our AI translates these into sophisticated filtering rules that improve over time.

• The implementation uses Claude and Gemini models to analyse retrieved news material and produce summaries and suggested content.

• Created a flexible notification system that adapts to your workflow, whether you need real-time alerts or prefer batched updates.

• Integrated AI-powered writing tools that help you quickly transform insights into publish-ready articles optimized for both readers and SEO.

Outcome

A workflow connecting source discovery, personalised briefs and editorial preparation.