Google's Mixboard gets Nano Banana Pro and presentation tools

Turning a messy moodboard into a polished slide deck is usually where creative momentum goes to die. You spend hours collecting scraps of inspiration, only to spend another three hours manually copying them into a rigid presentation template.

Google Labs is trying to eliminate that friction with a fresh round of updates to Mixboard, its AI-powered canvas experiment.

Turn chaotic canvases into presentations

Mixboard can now automatically convert your main board into a cohesive slide deck using a model called Nano Banana Pro.

Instead of just exporting raw images, the system reads the full context of your canvas along with your inputs, such as target format, key focus, and aesthetic preference. It then generates high-fidelity visuals and structured text to build out the deck.

Editing also got more tactile. You can now doodle directly over images or slide layouts to mark areas you want to tweak, write a quick prompt, and watch the model adjust that specific spot. Keep in mind that Google noted generation limits may fluctuate depending on real-time server traffic.

Bigger canvases, direct selfies, and PDFs

Input options are getting much broader beyond standard web images. Mixboard added a built-in selfie camera, making it easier to snap a photo of yourself for practical projects like wardrobe or fashion planning.

File support now includes PDFs, so research papers and multi-page documents can sit right alongside images.

To keep everything organized, projects now support multiple boards. Rather than crowding everything onto a single giant surface, you can split your workflow into a brainstorming board, a scrap pile for random assets, and a polished final board.

Why it matters

We think bridging the gap between open-ended brainstorming and linear presentation is one of the smartest moves Google Labs could make here.

Tools like Miro and Canva have been racing toward similar automated handoffs. The friction in creative work rarely lies in coming up with raw ideas; it lies in translating those chaotic references into something a client or team can actually digest. Mixboard attacks that exact bottleneck.

The doodle-to-edit feature is particularly clever. Pure text prompting is notoriously bad at spatial adjustments, so giving users a pen tool keeps them in control without forcing them into a complex image editor.

That said, it clearly remains an experimental sandbox. Google's caveat about fluctuating generation limits shows that compute resources are still tightly managed. It is well worth playing with if you frequently build visual concepts, but it feels like a preview of what Google Slides might become rather than a standalone enterprise tool just yet.