Narayen outlines Adobe's multi-phase roadmap for document AI
Shantanu Narayen · How Adobe is leveraging AI (with CEO & Chair Shantanu Narayen) | Masters of Scale · Feb 28, 2025 · at 18:24
Adobe CEO Shantanu Narayen details the product progression of generative AI across Acrobat and enterprise document systems during a discussion with Reid Hoffman.
0:00 / 0:20exact quote · 21.0s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“PDF, we just recently introduced AI assistant for PDF or Acrobat. You can have a conversational interface now with a PDF. The next step clearly is how do you do it across all of the documents, whether they be on your desktop or whether they be in the knowledge management system in the enterprise. And then the third step on that is how do you do it for vertical documents?”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Shantanu Narayen
Narayen: Adobe trains AI models only on fully licensed data
“If we create models, one of the things that we have to do is create a model where every piece of data that went into the creation of that model, we can testify to the fact that, you know, we have license to do that.”
Narayen: AI will potentially replace people who don't use AI
“We said, people who use AI, this is an augmentation tool and it will potentially replace people who don't use AI.”
Narayen: AI will create far more jobs than it disrupts
“I think the reality with all technology, all automation is, you know, I, it introduces way more jobs, and it disrupts some, and I don't think AI will be you know, that different, and a lot of the, you know, automation or menial things that people do, it'll for…”
Narayen: Only 4 to 5 entities have capital to build frontier LLMs
“The large language models, I'm sure you'd agree there'd be four or five people who have the capital to create those large language models, and, you know, there'll be maybe one open source, there'll be three or four proprietary”