How CNN Weather is reimaging forecasts with AWS AI

For more than 46 years, CNN has served as a trusted resource for global news. Now the storied media brand is undergoing a digital transformation, as well as an expansion into new verticals. CNN Weather, the first of these new standalone lifestyle extensions, launched in May. Now available in the iOS App Store in the United States, the app combines expert-led meteorology and climate expertise with a modern, mobile-first design. Built on Amazon Web Services (AWS), the CNN Weather app delivers real-time weather summaries using AI/ML, alongside its trusted journalism.

“We defined real-time journalism on cable TV, and now we’re becoming a digital-first media company that reaches audiences across every screen. Expanding beyond breaking news, we’re providing more utility and leaning into other moments that impact us all, like weather,” said Guy Griggs, SVP of Advertising and Client Partnerships, CNN. “Most people don’t want to spend tons of time tracking radar or parsing through dense meteorological data. The CNN Weather app puts the information you need at your fingertips while embodying the CNN standard. With AWS AI, we’re confident the experience remains reliable, scalable, and real-time.”

Automating user-friendly insights

In developing the weather app, CNN prioritized delivering compelling visuals and a personalized experience. Translating raw forecast data and climate expertise into simple weather takeaways, the app generates easily digestible insights that help its audience prepare for what’s ahead and stay informed during severe weather events.

“People want to know whether they need to take an umbrella or if they should adjust their travel plans. We’re using AI to turn huge, complex sets of data into conversational summaries that really resonate with audiences,” explained Griggs.

CNN Weather features vertical video and short written updates from meteorologists, alongside AI-driven weather summaries powered by Amazon Bedrock. It distills raw forecast data encompassing hourly temperatures, precipitation probabilities, wind speeds, dew points, and more into a single sentence that tells someone what the weather means for their day just by glancing at their phone, instead of bouncing between confusing data sources.

Unlocking new opportunities at scale

Behind the simple output is a highly complex system. Before any AI model is invoked, raw forecast data is run through a threshold-based deterministic compression engine. This step transforms swaths of meteorological data conveyed in dense JSON files into a prioritized narrative of what actually matters, cutting inference cost and latency while ensuring the model focuses on the most impactful weather conditions. It detects severe weather, classifies wind based on configured thresholds, surfaces extreme temperatures, and filters precipitation events.

The system then sends the compressed weather data to Amazon Bedrock for natural language generation. The service queries Anthropic Claude, which generates a forecast output that follows CNN’s editorial standards, including AP style, language guidelines, and a severity-appropriate tone, within the 74-character limit.

The weather API runs as a containerized FastAPI application on Amazon Elastic Container Service (Amazon ECS) with AWS Fargate serverless compute engine, eliminating the need to manage underlying servers. The service is configured for high availability and scale, with automated scaling and application load balancing. The apps’ deployment pipeline is also fully managed on AWS, orchestrating end-to-end processes, managing artifacts, handling encryption, and more.

Griggs shared, “We have a longstanding partnership with AWS, which makes up a lot of our tech stack, so building this app was a natural extension of the work that we’re already doing. We trust AWS and know the speed and reliability it delivers. We tested a lot of different AI partners and AWS rose to the top in terms of the capabilities and scale. That was really important as we launched this new experience.”

The CNN Weather app is now available for iOS and coming soon to Android. Next, the team anticipates launching a new health and wellness app in the fall, applying key learnings from this experience.

Learn how to build generative AI applications at scale on AWS, and get started with Amazon Bedrock.

Further reading

• Unifying global streaming metrics for Warner Bros. Discovery on AWS
• Transforming sports storytelling with generative AI on AWS
• Warner Bros. Discovery reinvents linear television advertising with AWS
• Warners Bros. Discovery announces agentic AI-powered advertising technology built on AWS, its preferred cloud provider
• Why we’re launching the CNN Weather app