Brief news stories generated from structured data
An agent writes short, formulaic stories from structured data such as earnings or game results, with editors setting templates and checking samples.
The use case
News agencies and local outlets cannot assign a reporter to every quarterly earnings report or high school game, yet readers want those results. An automated writing system takes structured data from a trusted feed and writes short stories in a fixed style, which editors design and test. The Associated Press has used automation for earnings stories for years. Editors still own the templates, check samples regularly, correct errors quickly, and add reporting to stories that need context. Stories should be labeled as automated, and anything unusual in the data goes to a human.
Example workflow
- Choose a trusted structured data feed
- Editors write and test story templates
- System generates stories as new data arrives
- Editors sample-check and fix templates
- Unusual results go to a reporter
Tools
- Automated Insights Wordsmith
- AP
- Claude
Risks and guardrails
- Bad data produces wrong stories at scale
- Formulaic text lacks context
- Labeling automated content
- Quick correction process needed
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- AP: artificial intelligence at the Associated Press: www.ap.org/solutions/artificial-intelligence
- Claude (official page): claude.ai (checked 2026-11-09)