Adobe introduced a major update to Acrobat’s AI assistant in February. This update included AI-generated summaries and a chatbot that can answer questions about the content of large documents. Starting tomorrow, Adobe will add new capabilities to this chatbot, giving it the ability to analyze and query multiple documents.

Adobe adds multi-document analysis and rendering capabilities to Acrobat’s AI assistant

Adobe is adding an AI-powered image creator to its Acrobat software. This feature, powered by the Firefly model, will allow users to remove backgrounds or make minor changes within existing PDF documents.

Adobe says these AI models are “business-safe” and will not compromise user privacy or create images that would negatively impact a brand’s image.

The most interesting part of this update is the addition of multi-document AI analysis to Acrobat. Previously, the assistant only focused on individual documents and provided users with answers based on the sources within the documents.

Adobe will expand this feature to allow AI analysis on a group of documents. Users will be able to drag and drop non-PDF file types, such as Word and PowerPoint documents, into the chatbot interface. From there, they can ask questions in the usual way or ask the AI to identify trends across multiple documents.

While other chatbots, such as Gemini and ChatGPT, allow users to upload documents for analysis, Adobe thinks users of its document editing software will see the value of an integrated solution.

Acrobat’s offering is also cheaper. The AI assistant costs $5 per month, less than the $20 you’d have to pay for ChatGPT Plus or Gemini Advanced. Adobe will also offer the feature for free for all users from June 18-28, 2024.

Confidence in Adobe’s AI capabilities has fluctuated in recent weeks. The company faced user backlash after admitting it used a small sample of Midjourney images to train its Firefly model.

It later backtracked on some terms of service clauses that potentially granted it the right to analyze user content. The company clarified that it would not access user content to train AI models and would prohibit third-party providers from using that data.

With this update, Adobe continues its stance on protecting user privacy. Documents will be uploaded to the cloud, but the company will not use the data to train its models. Similar to Apple’s ChatGPT integration, third-party LLM providers will also be prohibited from using the data.