The world of artificial intelligence continues to capture attention and headlines, with its potential to transform various industries. However, amidst all the hype, it can be difficult to find practical and effective uses for generative AI technology. That’s where Writer, a San Francisco startup, comes in with its innovative enterprise-focused generative AI writing products. And today, the company has announced a new feature for its Palmyra model – Palmyra-Vision – that can generate text directly from images, including graphs and charts.
“We have strategically chosen to focus on multimodal content, with a specific emphasis on text output. This new capability is part of that strategy,” explains May Habib, co-founder and CEO of Writer, in an interview with TechCrunch.
This strategic direction has led the company to analyze images rather than produce them, at least for now. Habib mentions that they may introduce the ability to create charts and graphs in the future, but currently, their focus is on generating text from images.
The process behind Palmyra-Vision involves multiple models working together to produce highly accurate results with a 99.99% accuracy rate, according to Habib. This technology has a wide range of potential uses, from generating text for eCommerce websites that has to keep up with constantly changing images, to automatically interpreting key takeaways from complex charts and graphs. Another use case is compliance checking, where Palmyra-Vision can analyze ad copy for FDA compliance.
Habib does emphasize the importance of human review in the process, as no model is error-proof. “Models can hallucinate or simply get facts wrong, and it’s crucial to have human input and review in the workflow,” she states. While recommending this to all their customers, Habib acknowledges that a more automated workflow may be necessary in the future as the company continues to scale.
With $126 million in funding to date, Writer is currently in talks with major cloud infrastructure platforms to partner and further expand their reach. Their last funding round, a $100 million Series B, was led by Iconiq in September last year.
The latest update to Palmyra, which includes the image to text feature, is now available for use.
