Writer’s New AI Models Transform Images, Charts, and Graphs into Textual Content

As generative AI continues to capture attention, pinpointing actionable business use cases amidst the buzz can be challenging. Writer, a San Francisco-based startup, is dedicated to developing generative AI writing tools tailored specifically for enterprises. Today, the company unveiled an exciting new feature for its Palmyra model, named Palmyra-Vision, which generates text from images, including graphs and charts.

Co-founder and CEO May Habib explains that the company strategically focuses on multimodal content, allowing for text generation from visual inputs. “Our goal is to leverage multimodal input while providing insights that are delivered through text,” Habib stated.

Guided by this vision, Writer has opted to analyze images, deferring any plans for producing them for the time being. While Habib acknowledges the potential for future developments in creating charts and graphs from data, the current focus is solely on extracting text from such images.

To achieve Palmyra-Vision's impressive results, Writer employs a multi-model framework, with each model assigned specific tasks to accurately interpret the image and generate text with a high level of precision, reportedly achieving four nines of accuracy.

This innovative tool offers numerous applications, such as automating text generation for e-commerce sites that feature thousands of dynamic images, ensuring the website stays updated with the latest products without manual intervention. Additionally, it can effortlessly extract key insights from charts and graphs. Another significant use case is in regulatory compliance; for instance, a pharmaceutical company might leverage Palmyra-Vision for automated FDA compliance checks on advertising copy, aligning it with relevant regulatory documents.

The product also includes functionality to interpret and summarize handwritten notes into text. However, Habib emphasizes that training the model for specific contexts, like medical or insurance, is crucial to ensure precision.

Habib advises that human oversight is essential when utilizing these tools. She asserts that reviews are critical, as AI models can sometimes produce inaccurate information or hallucinate. While most customers are receptive to this recommendation, she envisions a future where automated workflows will facilitate consistent human review across users—a goal the team is actively pursuing.

To date, Writer has secured $126 million in funding, according to Crunchbase, and is exploring partnerships with major cloud infrastructure providers to scale its operations. The company recently concluded a $100 million Series B funding round led by Iconiq last September.

The new Palmyra-Vision feature is now available, providing enterprises with enhanced capabilities in image-to-text generation.

Writer raises $100M for its enterprise-focused generative AI platform