Case study / 01
Cloudflare Daily
News Intelligence Agent
A Cloudflare-native intelligence system for scheduled collection, provider observability, story clustering, D1/R2 storage, article extraction, drafts, publishing, and traceable reporting.
Type
AI intelligence system
Platform
Cloudflare
Status
Built MVP / Cloudflare-native system
Stack
Agents SDK · Durable Objects · D1 · R2
Overview
newsharness is a Cloudflare-native agent that discovers, curates, and delivers daily news intelligence.
It orchestrates end-to-end pipelines using Cloudflare primitives with observability and traceability at each step.
System architecture
Sources
Provider normalization
Story clustering
Extraction + enrichment
Drafts + publishing
D1 analytics + R2 artifacts
Problem
News is fragmented, noisy, and fast-moving. Manual aggregation and publishing do not scale.
What it does
- Scheduled collection
- Provider normalization
- Story clustering
- Article extraction
- Draft generation
- Publishing workflows
- Traceable reporting
What this proves
- Cloudflare Agents can orchestrate practical content pipelines.
- D1/R2/Durable Objects can support inspectable AI workflows.
- AI-assisted reporting should be observable and traceable.
- Separation of concerns improves reliability and maintainability.
Next improvements
- Richer evals
- Better personalization
- Public Ask the Work integration
- Stronger reviewer workflow
- Source credibility scoring