Main content

PDFs get smart as Adobe brings generative AI to Reader and Acrobat

Phil Wainewright Profile picture for user pwainewright February 20, 2024
Summary:
Generative AI comes to the world's PDF documents today as Adobe announces its beta of an AI Assistant that will summarize and find answers in Acrobat and Reader.

Adobe AI Assistant screenshot
AI Assistant in action (Adobe)

If you've ever had a feeling of dread when opening an email or shared folder to discover a slew of weighty PDFs to read and digest, today's announcement from Adobe will feel like a lifesaver. The creator of the ubiquitous PDF file format is introducing AI Assistant for English users in beta, harnessing generative AI in Acrobat desktop and web applications to instantly produce summaries and insights from long documents, answer questions about the content, and format the resulting information for sharing in emails, reports and presentations. Within the next few weeks, all users of the free Adobe Reader desktop app will also be able to get this AI help, bringing generative AI to users of an estimated total of 3 trillion PDF documents in existence worldwide. Once out of beta, the AI Assistant will be a paid add-on for Acrobat and Reader subscribers.

The new capabilities are available immediately to customers using Acrobat Individual, Pro and Teams or trialing Pro, and will be rolling out to users of Reader over the coming weeks. They include:

  • AI Assistant — using a conversational interface, the chatbot recommends questions based on a PDF’s content and answers questions about what’s in the document. Adobe proprietary technology adds citations showing the source of these answers.
  • Generative summary — provides a quick, easy-to-read overview of the content inside long documents.
  • Clickable links — help customers quickly find what they need in long documents.
  • Formatted output — AI Assistant can consolidate and format information into top takeaways or other output ready to copy and paste into emails, presentations, reports, and other sharing channels.
  • Beyond PDF — AI Assistant can also be used with other kinds of document formats, including Word, PowerPoint, meeting transcripts, and so on.

Adobe is flagging today's announcement as just the beginning, with future AI Assistant capabilities in the roadmap ahead set to include:

  • The ability to work across multiple document types and sources to surface information as needed.
  • AI-powered authoring, editing and formatting, such as rapid generation of first drafts, and help with copy editing, for example changing voice and tone, compressing copy length, and suggesting content design and layout.
  • Applying capabilities from Firefly image generation and the Adobe Express design tool to adjust the presentation of documents.
  • New AI-supported digital collaboration, providing analysis of feedback and comments, suggesting changes, or highlighting and even helping to resolve conflicting feedback.

Adobe is keen to emphasize that its AI Assistant has the necessary safeguards in place, including enterprise-grade security and information governance for large business customers, and has been developed in line with the vendor’s AI Ethics processes. The assistant currently uses the Microsoft Azure OpenAI Service but Adobe insists it is not tied to any specific LLM, and will use a selection of technologies as needed to address a range of customer use cases. Third-party LLMs are not trained on Adobe customer data. The AI Assistant builds on the existing technology first introduced in 2020 to power Acrobat Liquid Mode, which uses knowledge of PDF document and content structure to turn the fixed page structure of traditional PDFs into more responsive reading experiences for use on mobile devices. This already allows for the creation of intelligent outlines, collapsible sections and text search, and has been foundational to the subsequent work on the AI Assistant.

My take

After last year's very impressive launch of Firefly, which brought the hugely disruptive impact of generative AI on image creation and editing to Creative Cloud and Experience Cloud, it was only a matter of time before Adobe extended the technology into the Document Cloud. Making the new AI Assistant available, initially at no cost, to the huge population of Adobe Reader users, will bring generative AI squarely into the enterprise mainstream for document querying and summarization. IT leaders might want to take note that this is yet another backdoor entry point for generative AI to seep into everyday use within the enterprise, as how many knowledge workers aren't dealing with PDF documents on a daily basis? The urgency of dealing with an explosion of 'shadow AI' across the organization just stepped up another few notches.

From a user behavior point of view, it's going to be important to start raising some awareness of the potential pitfalls of over-reliance on this technology. Applying generative AI to PDF documents promises to save a lot of time while ensuring essential information gets conveyed. But AI can only work with the data in front of it. Yes, in a perfect world, where well-structured reports tell the whole story with no bias or omissions, having an AI-generated summary is a clear time-saver. But I often find that what a document or transcript doesn't say is often as important as what it includes, while others, whether by accident or design, misdirect the reader away from the crucial issue. Having AI to hand is going to invaluable, but we still can't afford to let our guard down — human judgement and context will become even more important for evaluating the reliability of information that AI serves up to us.

Meanwhile, Adobe no doubt is looking for this announcement to give it a significant leg-up in the digital teamwork space. As part of this announcement, it is emphasizing that last year saw a 75% increase in the volume of documents shared using Acrobat, out of the mind-bending figure of 400 billion PDFs opened using the tool. Now it's hoping that its roadmap for the AI Assistant will lead to a further surge in collaboration on the platform along with more use of Acrobat to create, edit and search content. I would still rate the Document Cloud as a distant laggard among those competing for leadership within what diginomica calls the Collaborative Canvas of enterprise digital teamwork. But as I wrote recently, the advent of generative AI has brought big changes to the competitive landscape here, opening up new opportunities for others to gain ground. It will be interesting to see what more Adobe plans to make of this window of opportunity over the next year or two.

Loading
A grey colored placeholder image