AP's NLG Revolution: 14x Earnings Reports Surge

The Challenge

In the mid-2010s, the Associated Press (AP) faced significant constraints in its business newsroom due to limited manual resources. With only a handful of journalists dedicated to earnings coverage, AP could produce just around 300 quarterly earnings reports per quarter, primarily focusing on major S&P 500 companies. This manual process was labor-intensive: reporters had to extract data from financial filings, analyze key metrics like revenue, profits, and growth rates, and craft concise narratives under tight deadlines. As the number of publicly traded companies grew, AP struggled to cover smaller firms, leaving vast amounts of market-relevant information unreported.[1][4]

This limitation not only reduced AP's comprehensive market coverage but also tied up journalists on rote tasks, preventing them from pursuing investigative stories or deeper analysis. The pressure of quarterly earnings seasons amplified these issues, with deadlines coinciding across thousands of companies, making scalable reporting impossible without innovation.

Quantitative Results

• 14x increase in quarterly earnings stories: 300 to 4,200
• Coverage expanded to 4,000+ U.S. public companies per quarter
• Equivalent to freeing time of 20 full-time reporters
• Stories published in seconds vs. hours manually
• Zero reported errors in automated stories post-implementation
• Sustained use expanded to sports, weather, and lottery reports

Implementation Details

Timeline and Rollout

The AP's NLG journey began in 2014 with a pilot partnership alongside Automated Insights, focusing on automating quarterly earnings recaps. By July 2015, AP publicly announced the expansion, stating that the majority of U.S. corporate earnings stories would be produced via automation technology. This phased approach allowed for iterative improvements: initial pilots covered select companies, refining templates before full deployment across thousands of firms.[1][4]

Full implementation hit stride in Q3 2015, scaling to 4,200 stories per quarter by late 2016. The technology integrated seamlessly with AP's content management systems, pulling data from trusted feeds like Zacks. Today, as of 2025, Wordsmith remains core, now owned by Stats Perform, and has evolved into broader AI initiatives like AP's Local News AI projects funded by the Knight Foundation.[2][6]

Technology Stack and Process

At the heart is Wordsmith NLG, a template-based engine that converts spreadsheets of financial metrics into natural language. Key steps include: 1) Data ingestion from SEC filings or partners; 2) NLG template execution—e.g., 'Company X reported $Y revenue, up Z% year-over-year'; 3) Human review for nuances like executive quotes or context; 4) Publication. This ensured stories indistinguishable from human-written, with no style shift noticed by readers.[3][5]

Challenges like data accuracy were mitigated via rigorous verification protocols, and ethical concerns (e.g., transparency) were addressed by disclosing automation internally. Integration with AP's global wire service allowed syndication to thousands of outlets.

Overcoming Hurdles

Initial skepticism about AI quality was overcome through human-AI collaboration: editors customized templates for AP's voice. Scalability issues during peak earnings weeks were resolved by cloud-based processing. Long-term, AP trained staff on oversight, turning reporters into 'story directors.' This model has since expanded to non-earnings areas, proving NLG's versatility.[1]

Current Status and Evolution

As of 2025, the system covers thousands of companies quarterly, with expansions into sports recaps, weather alerts, and election data via five new AI tools in AP's Local News initiative. No major disruptions reported, and it continues driving efficiency in a resource-strapped industry.[2]