Last week I went looking for a clean demo of Google’s newest AI content tools, and most of what I found was hype with no real workflow. Then I hit one video that actually builds something end to end. It comes from Grace Leung, a digital growth consultant who got hands-on with these tools at Google I/O and walked through them using a single demo snack brand called Healthy Crunch. What the creator does here is the twist: she doesn’t just list features, she starts with nothing but raw brand files and builds a full content engine using only Google Gemini. Blog posts, ad shots, product videos, social campaigns, even a website. All in one connected flow.
What’s new
The big shift the original poster highlights is how everything in Google’s AI now links together. NotebookLM lives right inside the Gemini app, so you get a project space that actually remembers your brand. The expert loads a “brand knowledge” notebook with hero products, customer insights, and product images, then treats it like a living brand brain that every other tool can pull from.
That one move changes the whole game. Instead of re-explaining your brand in every prompt, the context travels with you.
The twist that makes it click
Here’s the part that got me. Because the product images live in the notebook, Gemini generates visuals that are actually on brand, not generic stock-looking junk. The creator asks for a blog post, flips on canvas mode, then turns on image mode to add a feature banner and inline visuals that match the real product. Then she exports the whole thing straight to Google Docs for the team to comment and polish.
Same notebook, different output. She spins up a lead magnet PDF (a lunchbox planner for parents) and a one-page retail line sheet for the sales team. The mind behind it says it takes under 10 minutes each.
The mini-workflow
Here’s the repeatable loop the author lays out, simplified so you can copy it:
- 📌 Build a brand notebook in NotebookLM. Drop in products, insights, and real images.
- 📝 Generate text first. Ask Gemini for a content brief, then a full draft in canvas mode.
- 🎨 Add on-brand visuals with Nano Banana, Google’s image model, right in the same chat.
- 🎬 Make video with the Omni model or Google Flow for longer, edited clips.
- 📣 Scale social with Pomelli, which reads your website and spits out branded posts.
Nano Banana is the quiet star
The contributor spends real time here, and I get why. Nano Banana isn’t just for banners. She shows it doing product ad shots from a couple of images, website hero banners, and quick edits using a sketch tool to circle and remove objects.
Three uses stood out to me:
- A/B test angles: feed it a product image and ask for three text-to-image prompts with different moods and backgrounds, then generate the winner.
- Resizing creatives: turn one square ad into a vertical layout in about 10 seconds while keeping everything consistent. The person who shared it calls this a huge time saver, and I agree.
- Iteration without waste: keep refining the same shot instead of starting over.
GEMs kill the repeat-prompting pain
Everything above still means prompting by hand each time. The savvy professional’s fix is Gems, custom versions of Gemini you set up once. She builds a “Studio Ad Shot” Gem loaded with Nano Banana as the default tool, a style guide, and even product packaging dimensions so images come out realistic.
The smart bit: you can share a Gem with your team. Now everyone gets the same quality without rewriting prompts. Studio shots, lifestyle images, campaign copy, one Gem per repeatable task.
Video gets serious
This innovator then moves to Google’s Omni model for video. It takes almost any input and generates clips with strong physics and surprisingly solid character consistency. The same model stays the same person across scenes.
Her best tip here is about quota. Daily video limits are tight, so she plans a storyboard first using a Gem built on Google’s official Omni prompt guide. It generates scene-by-scene instructions and a storyboard image, so you adjust before burning a generation. Omni can even reference a 10-second clip of an existing Instagram reel to copy pacing and editing style.
For longer, stitched video, the creator switches to Google Flow. It wraps a real studio around the model: a timeline, reusable brand characters, and an agent mode that can generate three ad variations with different camera angles and ask for your approval first.
Pomelli puts social on autopilot
The last tool, Pomelli, reads your website, pulls your brand DNA, and builds on-brand social content in minutes. The author sets up the brand profile, checks the auto-collected product assets, then either uses pre-built campaign ideas or describes a campaign in the chat to get a creative brief and matching assets. It also does product photoshoots from your existing images and can even publish a simple test website from your brand sources.
Pro tips worth stealing
- Always let Gemini make the rough draft, then polish in Canva. Don’t start from a blank page.
- Plan video storyboards before generating to protect your daily quota.
- Build Gems for anything you do more than twice, then share them so the whole team stays consistent.
I think this is a real shift because the brand context finally follows you across every tool instead of living in your head. If you do any content or marketing work, the full walkthrough is worth your time. Watch the original video for the live demos and the exact prompts the creator uses. 🚀