Knowledge to Imagery: How AI is Transforming Information into Visual Artifacts

At the end of 2025 we can safely say that there is no single AI solution that does it all. If you want to get the most out of AI, you end up using multiple tools. That’s mainly because the underlying models have specific strengths and every few months a new capability comes online.

One of the tools I encourage people to use is Notebook LM by Google. The underlying model is the new Gemini 3 which has leapt ahead of the others (for now) and its new capability is accurate image generation and editing. This is allowing for new ways of working with information, especially condensing large amounts of text information into images - illustrations, infographics, ‘photos’, diagrams, maps, and slides. eg “Generate a kid’s treasure map based on my career path - use my linkedin profile pdf”. Might not sound like much but this is a pretty profound new capability that again will change how we work.

A New Modality: Curating Knowledge into Visual Decks

By way of example, “I” created this slide deck as a resource for founders in the Hub community. A more honest telling would be that I “curated” this deck. I have a notebook in Notebook LM which is loaded with all of the articles, posts and podcast interviews that capture the ideas and wisdom of one of Silicon Valley’s most sought after PR/Comms/Branding gurus, Lulu Cheng Meservey.

Think of a notebook as a knowledge base. Notebook LM lets you add pdf’s, text snippets and youtube videos to those notebooks. These days long form podcast interviews with experts like Lulu are a gold mine of helpful advice and insights. Notebook LM ingests those messy text transcripts from the podcast videos on youtube and is able to make sense of them.

It is easy to add/remove those sources to keep your knowledge base up to date. And over the last few months the Gemini team have added some cool new ways to transform those knowledge sources into useful artifacts: first it was “Audio Overview”, an automated 2-person podcast where the hosts would discuss the subject matter. And since then they’ve introduced other artifacts: Video Overview, Mind Map, Reports, Flashcards, Quiz, and Slide deck.

The Power of Gemini 3’s Image Generation

Google has started to wire the new Gemini 3 model into these artifacts and its upgraded image generation capabilities is making a difference. In this deck, you’ll see illustrations, subtle design elements and diagrams that are mostly spot on. You’ll also see a few errors in the text (because the text is generated as an image). I was able to do this on their very generous free tier.

Whereas previously I recommended that a founder watch a bunch of podcasts, read articles etc (which few would have the time to do), now I can provide this deck as a starter/cheat sheet and let them dig in further from there. Compressing a huge amount of knowledge like that into a 20 slide deck of the key insights used to be a huge task and I would have struggled to create something like this myself.

It feels like a new modality: knowledge to imagery.

Congrats to the Notebook LM and Gemini teams and apologies to Lulu.