August 8, 2025

Making Weather More Accessible with AI-Powered "Weather Agent"

In recent years, artificial intelligence has garnered tremendous attention, gradually integrating into our daily lives through smartphone voice assistants, robotic vacuum cleaners, automatic translation apps, facial recognition systems, recommendation engines, autonomous driving features, and more.

Weathernews has developed a weather-specialized AI service that launched January 2025. We spoke with Takuya Kageyama from the Mobile Internet Business Division's development team about Weathernews' newly developed AI service, the "Weather Agent."

AI-Era New Services Transform Weather Forecasting

- What is the "Weather Agent"?

On January 30, 2025, Weathernews launched a new feature called "Weather Agent" that leverages generative AI within our weather app "Weathernews." The "Weather Agent" is a chat-based function where AI responds to complex weather-related questions, delivering weather information through natural conversation and providing users with an entirely new experience beyond traditional weather forecasts.

Beyond offering easy-to-understand weather forecast explanations, we've enhanced the system to handle diverse meteorological information including pollen forecasts, cherry blossom updates, and severe weather alerts. The chat interface uses emojis and friendly language to create an approachable atmosphere, delivering responses that naturally guide users toward relevant weather information.

Since this is a chat function, we expect users may occasionally ask non-weather-related questions. We've designed the system to handle these situations appropriately by clearly communicating our specialization in weather information or presenting follow-up questions like asking which specific region's weather they're interested in, steering the conversation back to meteorological topics.

- Were there any difficulties during the development process?

Improving answer accuracy was a major challenge in developing the "Weather Agent." Initially, the AI model occasionally provided incorrect responses based on its training data or failed to properly reference essential meteorological information.

To address this issue, we modified the system to enable the AI model to directly access and utilize Weathernews' proprietary and highly accurate meteorological data in its responses.

This approach allowed us to create a meteorology-specialized AI agent that relies on Weathernews' trusted data rather than model assumptions, resulting in a system grounded in reliable information from our specialized meteorological institution.

- Why did you decide to develop AI-powered services now?

At Weathernews, we actively integrate AI into our daily operations, including organizing company-wide AI hackathons for all employees 1. AI technology has advanced significantly and dramatically improved our operational efficiency. Through hands-on experience with AI tools, our team became confident that the technology had reached a quality level capable of delivering genuine value to users, which sparked our service development initiative.

We also decided to launch this service publicly because we wanted app users to experience the weather-specialized AI agent we had been developing and testing internally. Thanks to collaboration from our testing team in advance during the "Weather Agent" development, we were able to complete the content development in just one month after deciding to release the service.

Rich Meteorological Data: What Makes "Weather Agent" Amazing!

Takuya Kageyama, Development Department, Mobile Internet Business Division, Weathernews
Takuya Kageyama, Development Department, Mobile Internet Business Division, Weathernews

- How will the "Weather Agent" change our lives?

For instance, it can handle detailed inquiries like "How many days ago was the last rainfall in Chiyoda Ward?" or "What are some sunny tourist destinations within a 2-hour drive this weekend?" The system provides visually intuitive responses by incorporating not only text but also images and graphs for clearer explanations.

With the introduction of "Weather Agent," accessing weather-related information becomes simple and intuitive, making daily weather checks more convenient than ever.

The strength of "Weather Agent" lies in its ability to reference and respond with comprehensive information based on user questions, drawing from Weathernews' extensive internal resources including weather forecasts, disaster alerts, historical meteorological data, natural phenomena information, and suggestions for related app features.

For instance, when checking weather for a travel destination, it provides not only basic forecasts but also practical, lifestyle-oriented details such as clothing recommendations and safety precautions. If you have any weather-related questions or concerns, we encourage you to try the "Weather Agent."

Additionally, we've built a flexible system that allows non-engineer staff managing app operations to easily update the sample questions displayed at the top of the chat interface, helping to boost user engagement.

- How will the "Weather Agent" evolve in the future?

We collect logs from "Weather Agent" interactions and store user questions anonymously for a designated period.

At Weathernews, we leverage these dialogue logs to enhance our daily service quality. We continuously analyze whether the system delivers appropriate responses to user inputs, constantly monitoring service performance. We also categorize similar questions and examine dialogue patterns and themes to develop a deeper understanding of user needs. Based on these analytical insights, we regularly update our content to provide the information users genuinely want.

Recently, we introduced AI-generated commentary based on user reports to our rain cloud radar, beginning to offer this as "local voices" 2. Moving forward, we plan to integrate AI into other popular existing features as well, with the goal of enhancing user satisfaction.

We also plan to implement additional functional improvements for "Weather Agent," including voice support that users have requested, as generative AI technology continues to advance.

Toward Providing AI Agents Specialized in Customer Business: Weathernews' Vision for the Future of Weather × AI

Currently, we offer "Weather Agent" as an individual service limited to paid members, but we aim to deliver AI-powered services to a broader audience in the future.

Weathernews provides meteorological information services across multiple industries including maritime, aviation, transportation, and rail sectors, and we plan to expand AI-powered solutions for corporate clients moving forward.

While information requirements and technical terminology vary by industry, Weathernews brings 40 years of corporate service experience and has developed deep expertise through comprehensive understanding of customer operations. Building on this foundation, we want to create specialized corporate "Weather Agents" tailored to each industry, helping to enhance operational efficiency.

AI technology has advanced rapidly in recent years, including the development of AI-based weather forecasting systems. At Weathernews, we're actively working to evaluate these AI forecast models daily, understanding their unique characteristics compared to conventional weather prediction models and identifying their respective strengths and limitations.

Given that meteorology involves processing vast amounts of data, we believe it offers excellent compatibility with AI technology. Weathernews will continue leveraging AI across various applications, pursuing business innovation and creating new value for our users.

Footnotes

  1. 1: 80% of Employees Participate in AI Hackathon, Accelerating New Value Creation Through Generative AI ↩︎
  2. 2: AI generates accurate comments from surrounding weather conditions, enhancing weather radar with 'local voices ↩︎