Tips for Prompting Nano Banana, From Me and the WSJ

In partnership with

Find out why 100K+ engineers read The Code twice a week

Staying behind on tech trends can be a career killer.

But let’s face it, no one has hours to spare every week trying to stay updated.

That’s why over 100,000 engineers at companies like Google, Meta, and Apple read The Code twice a week.

Here’s why it works:

• No fluff, just signal – Learn the most important tech news delivered in just two short emails.

• Supercharge your skills – Get access to top research papers and resources that give you an edge in the industry.

• See the future first – Discover what’s next before it hits the mainstream, so you can lead, not follow.

Join 100,000+ engineers who read The Code to stay ahead of the curve.

Tips for Prompting Nano Banana, From Me, Google and the WSJ

I’ve been talking about Nano Banana nonstop since it launched. Google’s image AI model, now embedded within Gemini, is much better than any other model at editing images.

That’s amazing, because it means you can create totally new, realistic images without image editing skills. It’s concerning, because it allows people to potentially alter history or create realistic deepfakes very easily.

It’s also in its infancy, and constantly evolving.

Yesterday, the Wall Street Journal published a piece they created in collaboration with Google (part of their Paid Program, where companies can pay to commission articles that are then produced by WSJ’s journalists and to their Editorial standards) about best practices for prompting Nano Banana.

And guess who they turned to for expert advice—and some cool example photos?

Here’s the piece.

Straight from Google

Simultaneously, I’m at the Digital Media Licensing Association’s annual conference in Atlanta, where I’ve been speaking with—Google!

Specifically, I sat down with someone from Google’s Images team and used Nano Banana together.

A big takeaway from the conversation was that Google itself is still figuring out exactly what the system can do, and where it falls flat. My Google contact seemed just as delighted as I was by Nano Banana’s successes, and just as frustrated when the system’s output fell flat.

Nano Banana feels a bit like Google’s version of ChatGPT—a tool that they released without too much preparation, which happened to take off—unexpectedly—among consumers and normal people.

Now, Google is working to figure out how best to develop the model. My Google contact showed me how users can report when Nano Banana produces bad results (press Thumbs Down on the response and fill out the form), and confirmed that people actually read those notes.

So try it out, and flag stuff that’s broken!

OpenAI Finally Launches a Browser

They thought about doing it. They hinted about doing it. Now, they’ve done it!

OpenAI has finally released an AI-powered browser of its own. Here’s a breakdown of what it includes: