Writer’s newest models can generate text from images, encompassing charts and graphs as well
Writer, a San Francisco-based startup, is carving out a niche in the realm of generative AI writing products tailored for enterprise use cases. Their latest offering, Palmyra-Vision, is designed to generate text from images, including graphs and charts, as part of their focus on multimodal content.
May Habib, CEO and co-founder of Writer, explains that their strategic emphasis on multimodal input aligns with their goal of delivering insights via text output. While the company may explore image creation in the future, their current focus is on analyzing images to generate text.
Palmyra-Vision leverages a multiple model approach to achieve accurate results, with each model assigned a specific role in identifying image content and generating corresponding text. According to Habib, the platform boasts an impressive accuracy rate of four nines.
The applications of Palmyra-Vision are diverse and impactful. In e-commerce, the tool can automate the process of generating text from a multitude of dynamic images, ensuring that product descriptions remain up-to-date without requiring constant human intervention. Additionally, Palmyra-Vision can extract key insights from charts and graphs automatically, enabling quick analysis of data visualizations.
In industries like pharmaceuticals, compliance checking is crucial. Palmyra-Vision can assist companies in performing automated FDA compliance checks on advertising copy. By comparing the content of an ad against FDA regulations outlined in associated documents, the tool ensures that promotional materials adhere to regulatory standards, minimizing the risk of non-compliance.
