I Tried Google Labs Whisk and Flow And I Haven't Stopped Since

I just wanted to throw a retro CRT monitor onto a flower-covered hill and see what happened. What came back genuinely stunned me.

I'd glanced at Google Labs before. Clicked around, shrugged, moved on. It felt like a science fair interesting in theory, not obviously useful. Then someone in a Discord I'm in dropped a screenshot that made me stop scrolling. A lush green hill. Monarch butterflies. Stacked vintage monitors. Wildflowers everywhere. Clear blue sky.

"Did you paint that?" I asked.

"Nope. Whisk. Twenty minutes."

That was all I needed to hear.

What Is Google Labs, Actually?

Google Labs is Google's public-facing experiment zone a place where product ideas get tested before they become full features (or quietly disappear). Think of it as a sandbox where Google's best models get creative interfaces bolted onto them, then released to see what people do.

The tools there aren't polished consumer products. They're deliberate experiments. But because they're built on Google's actual infrastructure Gemini, Veo 2, Imagen 3 they're not toy demos either. You're playing with the real thing.

Right now, the two that have completely taken over my creative workflow are Whisk and Flow.

Whisk: Image Generation That Thinks Like a Mood Board

Most AI image tools ask you to describe what you want. Type a prompt, iterate, get frustrated, type a better prompt. Repeat.

Whisk flips this. Instead of describing everything from scratch, you bring reference images as building blocks:

• A subject image (what you want to be in the scene)

• A scene image (the environment or setting)

• A style image (the visual aesthetic or mood)

Whisk takes these three inputs, understands the essence of each, and generates something new that combines them not a collage, but a synthesized original.

The result feels less like prompt engineering and more like art direction. You're making creative choices, not writing instructions and hoping.

What I Actually Made

I spent about an hour with Whisk and came out with a full set of visuals for a concept I'd been noodling on something I was calling "Proximity." Here's roughly what I generated:

• Rolling green hills with oversized cartoon daisies under a soft blue sky

• Monarch butterflies scattered across a clear blue sky with fluffy clouds below

• Two stacked vintage CRT monitors on a wildflower-covered hill, butterflies circling

• A giant green globe completely covered in colourful wildflowers

• A single retro monitor on a sunlit hill, rays cutting across the sky

Every single one came out coherent. Same colour temperature. Same dreamy, slightly surreal quality. Same world.

That aesthetic consistency across images, without any elaborate system prompting is what I haven't been able to get out of other tools without a lot of effort. Whisk just... understood the vibe.

Here's the prompt I used to get the stacked monitors shot:

Two stacked vintage beige CRT monitors on top of a lush round green 
hill covered in colorful wildflowers pink, blue, white. Monarch 
butterflies flying around and perched on the monitors. Bright blue 
sky with large fluffy white clouds. Low angle looking up, cinematic, 
photorealistic, soft dreamy lighting.

Flow: AI Video That Doesn't Feel Like a Glitch

Flow is Google's AI video generation tool, built on Veo 2. The premise is simple: take an image or a text prompt, and generate a short video clip from it.

I know what you're thinking. AI video is still pretty rough. The morphing, the uncanny motion, the backgrounds that melt when someone walks past them.

Flow is different meaningfully different.

I took some of my Whisk images and ran them through Flow. The butterfly animations in particular were genuinely impressive. They moved like real butterflies the wing physics, the slight irregularity, the way they drifted on air. The clouds shifted at the right pace. The whole scene felt like it had weight and atmosphere, not like a dream dissolving in real time.

The clips are short that's still a limitation. But within that window, you're getting something you could actually use: a hero section loop, a social media intro, an animated splash screen.

For anyone making websites, video content, or brand assets this changes what "rapid prototyping" means.

The Part That Actually Surprised Me

I expected impressive outputs. I didn't expect to feel like the tools understood what I was going for.

When I was building the Proximity visuals, I wasn't following a formula I was making intuitive choices. An old monitor feels nostalgic but also slightly alienating. A hill full of wildflowers feels alive and optimistic. Butterflies feel like things finding their own path. Put them together and you get something that says technology and nature can coexist without anyone writing that in a brief.

Whisk picked up on that energy and ran with it. That's not something you can fully explain technically. It's just good output.

Why Google Labs Specifically?

You might ask: isn't this just Midjourney or Runway with a different logo?

Not quite. A few things make Google Labs different:

1. It's genuinely free to explore. No premium tier constantly in your face, no watermarks on every output, no 10-image daily limit on the interesting features. You can actually play.

2. The models are world-class. Imagen 3 for images, Veo 2 for video these are the same models powering Google's commercial products. You're not getting a lite version.

3. The interfaces are built for creative exploration. Whisk's three-input system, Flow's image-to-video pipeline these aren't just API wrappers. Someone thought carefully about how creative people work and designed accordingly.

4. It moves fast. The tools in Labs update quickly. What I used last month is already better this month.

Who Should Go Try This Right Now

• Web designers building hero sections, landing pages, or visual identities

• Brand designers who need to explore visual directions quickly before committing

• Content creators who want cinematic social video without a full production setup

• Developers who want stunning assets for their side projects without a design budget

• Honestly, anyone who's ever thought "I have this visual idea but I can't execute it"

My Honest Take

I've been making things on the internet for a while. I've seen a lot of tools come through that promised to change everything and mostly didn't.

Whisk and Flow aren't perfect. The video clips are still short. Whisk sometimes interprets your references in unexpected ways (usually interestingly, sometimes frustratingly). There are limits.

But the ceiling of what you can produce in an afternoon with no design background, no budget, just a decent sense of what looks good has moved significantly. That matters.

Go to labs.google, spend an hour, bring a loose idea. I promise you'll come out with something that surprises you.

All images in this post were generated using Google Whisk. Videos generated with Google Flow. Total time: one afternoon.