New logo. No designer.

I needed a new logo for my Substack publication.

My old one was a stick figure I grabbed from a free icon website.

It worked fine when I was just starting.

But now I'm building out a bit more of a brand around the concept of “Unscaled” and I wanted something that felt a bit more…professionally made.

Something I could use across my quiz funnel, my newsletter, my coaching program.

Something that didn't look like I grabbed it in 30 seconds from a design library.

Luckily, Google just released Nano Banana pro, and I threw this project at it.

It only took 20 minutes or so to get a decent logo.

Probably won’t be winning any design awards, but it’s good enough for me for right now.

I didn't pay a designer and didn't spend weeks going back and forth on revisions.

I just… made it.

Here's how.

Step 1: Created a design brief in ChatGPT

I've been building the Unscaled brand concept inside ChatGPT.

So I asked it to create a design brief. The kind of thing you'd hand to a designer.

It gave me clear direction: tone, visual style, colors, what to avoid.

Step 2: Took the brief to Gemini (Nano Banana 3)

I pasted the brief into Gemini, turned on Create Images and said:
"Based on this, give me three specific logo design ideas."

It came back with three distinct concepts.

Then I said:
"Create concept 1."
"Create concept 2."
"Create concept 3."

And it generated each one.

Step 3: Asked Gemini to assess its own work

This part was cool.

I asked: "Do any of these feel like they capture the Unscaled brand?"

It gave me an honest critique of each concept.

Interestingly, the one it thought was strongest was the same one I liked best.

Step 4: Iterated from there

The first version felt too geometric. Too corporate.

I wanted something softer. More hand-drawn.

So I asked it to redo the concept in a looser, more organic style.

It delivered.

Then I kept refining:

What's wild is that it remembered context between iterations.

In the past, if you asked it to change one small thing, it often would regenerate the entire image and you'd lose what you liked.

Now? Gemini kept the design intact and only adjusted what I asked for.

When I got stuck (like when it kept putting the wordmark on top of the logomark instead of below it), I just downloaded the best version, edited it in Canva, re-uploaded it as a reference, and kept going.

But the process was overall smooth.

At various stages, I even asked Gemini to review the design like a designer would. For example, I asked it, “is the proportion of the word to the logomark in this image good”

It replied with things like:
"Based on general design principles for logo proportions, the balance in this image is excellent. Here's a breakdown of why…"

And then give me a detailed analysis backing it up.

Pretty cool.

Eventually, I decided to go wtih a simple ‘u’ as the logomark and got this:

Step 5: Created variations

Once I had the main logo, I adapted it for:

Here's what improved in Nano Banana Pro vs the previous version:

1. "Thinking" architecture
Built on Gemini 3 Pro, it reasons through your prompt before generating. That means better coherence for complex requests.

2. Professional-grade controls
You can define camera angles, lighting, depth of field—stuff photographers and designers care about.

3. Reference consistency
You can upload up to 14 reference images to maintain style across generations. Huge for branding.

4. Text that actually works
It can render legible, correctly spelled text inside the image. No more garbled AI nonsense.

5. Search grounding
Because it's connected to Google Search, you can ask it to create infographics with real, up-to-date data—and it will pull accurate info instead of hallucinating numbers.

6. Native 4K resolution
No need for an external upscaler. The output is print-ready.

AI image generators are getting better at an astounding pace.

Still not perfect.

There were definitely moments where I had to work around limitations or jump into Canva to fix something manually.

But the gap between "AI-generated" and "professional-looking" is closing fast.

These tools are evolving constantly.

Something that sucked in January might be game-changing by June.

Gemini feels like a lost, forgotten LLM compared to ChatGPT and Claude.

But it shouldn't be.

Especially now.

If you haven't tried it in a while, it's worth another look.

And if you've been putting off a project because you thought you needed to hire someone or learn a new skill, try the AI version first.

You might be surprised at what's possible now.

👉Try Nano Banana

👉 How to Make & Edit Images in Nano Banana PRO (Full Guide)

Thanks for reading!

Nathan Rodgers

👋 Say hello on Substack and LinkedIn

Were you forwarded this email? Click here to subscribe.