While generative AI dominates the news, it might be challenging to discover real commercial application cases. Writer, a San Francisco company, develops enterprise-focused generative AI writing tools. The business launched Palmyra-Vision, a new feature that creates words from pictures like graphs and charts, today.
According to co-founder and CEO May Habib, generating text from photographs is part of their aim to focus on multimodal content. “We are going to focus on multimodal input, but text output, so text generation and insight,” Habib told Eltrys.
Following that guiding light, the business opted to analyze photographs rather than make them (for now). She may build charts and graphs from data, but they’re not doing it yet. This release generates text from those photographs.
According to Habib, the Palmyra-Vision findings are produced via a multiple-model technique, with each model detecting what is in the picture and creating text with four nines of accuracy.
This may be used to generate text from hundreds of changing photos to feed an e-commerce website with the newest products without a person or to automatically understand important lessons from charts and graphs. Compliance checking is another. For instance, a pharmaceutical corporation may employ Palmyra-Vision to automatically evaluate ad content for FDA compliance, as seen below.
Finally, the system can decode and summarize handwritten notes into text, but Habib says it needs to train the model for medical or insurance use cases to ensure accuracy.
Habib advises against utilizing these technologies without human assessment in the process. She thinks this is crucial since any model might hallucinate or misrepresent reality, so humans must verify the findings. They propose this to every client, and most understand it, but she feels it will take a more automated approach to make it happen consistently across consumers, which they are working toward.
According to Crunchbase, the business has raised $126 million and is exploring partnerships with significant cloud infrastructure platforms to grow. The $100 million Series B, headed by Iconiq in September, was its latest.
The newest Palmyra image-to-text release is ready now.
Current Writer models can create text from visuals like charts and graphs.
