Generating Headlines at Scale: Adapting to Diverse Media Brands

Modulai collaborated with Expressen to enhance their AI-powered headline generation tool. By using previously published articles as examples, we developed a data-driven solution that generates headlines and subheadings tailored to each brand’s unique style and tonality. This approach is expected to yield even better results as more advanced language models become available, with minimal additional development efforts.

Background

Expressen had developed an AI-powered tool to help their editors create headlines and subheadings quicker, as well as to inspire them with fresh ideas. While the tool was functional, Expressen recognized that scaling the system to accommodate different tonalities and brands was a challenge. Expressen sought our expertise to create a more data-driven solution that would adapt to the unique requirements of various brands without relying on numerous prompts.

Solution

We leveraged Large Language Models (LLMs) to generate the content. By utilizing few-shot learning, we developed a single prompt that could be used to generate output tailored to each brand’s unique style and tonality. This was achieved by providing the model with historical examples of published content from each brand, allowing it to learn and adapt to the desired style without the need for multiple prompts.

The result

By implementing few-shot learning with historical examples, we helped Expressen create a more data-driven and adaptable headline generation tool. The solution proved to be simple yet effective, and it is expected to remain relevant as AI models become more advanced and cost-effective in the future. The collaboration between Modulai and Expressen demonstrates the potential for AI-powered tools to streamline content creation processes across various brands and tonalities.