On Tuesday, Adobe launched an artificial intelligence assistant in its Reader and Acrobat apps that can summarize PDFs and other documents and answer questions about them.
The AI assistant, which is still in beta, is currently live in Acrobat. "In the coming days to weeks, Reader will also receive these features," according to a press release.Adobe intends to launch its subscription service once the assistant has finished testing.
Adobe said that the AI assistant can help users extract key points from long PDF documents, by generating a brief overview of the content to simplify the information processing process. Through a "conversational interface", the assistant can also answer questions about the content of the document, and recommend that the user may raise questions about the document.
Adobe also noted that the AI assistant is also capable of generating citations so that users can check the source of their responses and can create content for text in different formats (such as emails, presentations, and reports), as described in the press release.
Similarly, other AI models such as ChatGPT offer PDF reader features that speed up the analysis of long documents, but these services require users to upload PDF files. Adobe's AI assistant, on the other hand, is a built-in feature.
In an interview on CNBC's "The Street" program on Tuesday, Adobe CEO Shantanu Narayen mentioned that this new tool reflects the company's goal of opening up access to billions of PDF files.
"Imagine you open a 100-page document and you want to know its summary quickly, you want to have a conversation with it, ask questions," Narayen says." You want to be able to correlate what's here with other documents you may have and all the information in your organization."
Last week, OpenAI, which developed ChatGPT, unveiled a new tool that generates photorealistic HD video based on text prompts. Responding to a question about whether OpenAI's model, called Sora, would have an impact on Adobe's domain, Narayen said that Adobe is also developing its own video model and plans to apply the technology to "a variety of tools and workflows" in a "responsible way".