PGA TOUR & AWS Using Agentic AI to Scale Sports Content Creation
AWS and the PGA TOUR say a new automated content system is producing more than 1,200 pieces of content per week, with near-instant publishing, lower costs, and expanded player-by-player coverage.
The PGA TOUR is using agentic AI to automate sports content creation at a scale that would have been difficult to manage through traditional editorial workflows. In a new AWS case study, the TOUR said its “Automated Content System,” developed with CapTech and built on Amazon Bedrock, now generates more than 1,200 pieces of content per week while reducing content production costs by 95% and keeping average generation cost below 25 cents per item.
The setup is designed around a distributed network of specialized AI agents rather than a single model handling the full workflow. AWS said research agents gather and verify information, writer agents generate content in different formats and styles, validation agents check factual accuracy and brand compliance, and editor agents prepare outputs for specific channels and audiences. The goal is to move from manual content production toward a system that can research, create, validate, and publish with minimal human intervention.
The content itself spans a wide range of formats. According to AWS, the system can generate round recaps, tournament recaps, betting profiles, player previews, field previews, round-by-round summaries, tournament summaries, and short-form social content. It also supports coverage for all 150-plus players in a tournament field and even shot-by-shot commentary across more than 30,000 shots in a golf event.
The PGA TOUR said manual workflows created delays that caused it to miss peak engagement windows, limited coverage mostly to top players and major storylines, and introduced cost inefficiencies through dependence on outside vendors. The new system is built to close those gaps by publishing within minutes of live action, expanding the depth of coverage, and bringing more recurring content generation in-house.
Under the hood, the research layer connects directly to the PGA TOUR API through custom Python functions that pull real-time player statistics, tournament and leaderboard data, historical performance records, and shot-by-shot tracking information. AWS said those agents can work dynamically from natural language prompts or programmatically through fixed content workflows, depending on the use case. Tournament recaps, for example, can trigger a standard sequence of data gathering and analysis steps to maintain consistency across outputs.
Accuracy and brand control are a major part of the architecture. AWS said validation agents extract factual claims from drafts, compare them against authoritative data sources, and generate correction recommendations when needed. Separate checks handle brand voice, preferred terminology, prohibited language, and context-specific style requirements. Multimedia agents also review visual assets against brand standards.
The model strategy is also built around efficiency. AWS said the PGA TOUR selects different models for different agent roles based on capability and cost, using lower-cost models for simpler tasks such as tool selection and fact checking, while reserving more advanced models for higher-value writing and creative work. That modular setup also allows the system to scale specific functions independently, depending on workload.
What emerges is less a single AI tool and more a new publishing workflow for sports organizations trying to keep pace with live events, fragmented platforms, and growing demand for individualized coverage. The PGA TOUR said the system now supports faster publishing, broader player-level storytelling, and content customization for other golf organizations, including the USGA.
The recommendation is to start with a narrow pilot around a specific content type, codify brand and data standards early, and expand gradually as teams build confidence in the workflow. For sports properties dealing with more events, more athletes, and more platforms than editorial teams can manually support, the PGA TOUR’s system offers a clear example of how agentic AI is beginning to reshape content operations.
