I enjoy tweaking website layouts and creatives to get every element looking just right; however, sometimes I just want to get the job done quickly. So when I tried Google’s Stitch 2.0, an AI-powered design tool that can turn prompts and reference images into interfaces, I was keen to see how efficiently it could make my vision a reality.

Recreating a landing page from a screenshot

Stitch gets the big picture right

For this experiment, I used an Andes Adventure brochure that I created in Figma as a test project and gave Stitch three tasks: recreate a landing page from a screenshot, rebuild an informative inner page, and generate an original landing page from scratch. Each task took about five minutes and gave me enough information to see how Stitch 2.0 measured up.

My first task was to recreate my existing Andes Adventure hero page, complete with a sunset photograph of Machu Picchu, the brand logo, the headline “Your next great journey starts here,” and contact details along the bottom. Stitch produced an attractive starting point in under a minute; however, the results looked more like a travel website than a generic wireframe.

From there, getting closer to my reference image took several prompts. Some versions duplicated the logo or headline, while others struggled with the logo’s size and positioning. Finally, after several attempts, Stitch completely changed the image composition. While this looked very attractive, it had strayed far from my original request.

So while Stitch was excellent at producing aesthetically pleasing results, it hadn’t recreated the design accurately. This is great news if you’re working from scratch—less so if you are trying to create something specific based on existing material.

Rebuilding an informative inner page

Good structure, stubborn details

So experiment 1, while very interesting, had mixed results. My next task was to recreate an informational page titled “Signature Experiences,” featuring a heading, introductory copy, a featured Inca trek, three experience cards, and a row of brand values at the bottom.

The first results were encouraging, with a coherent layout, decent images, and most of the content retained. However, problems emerged when I started requesting specific details. I wanted the main heading to fit on two lines, and the missing brand-values row restored. Stitch produced a very similar version without fixing those details.

So again, Stitch worked well under its own initiative, but issues arose when I started asking for detailed amendments. A tool that creates a good first draft but ignores precise feedback can leave you spending time repeatedly correcting the same problems.

Still, to give credit where due, Stitch gave me a structured page to work from rather than forcing me to build every section manually, and again, the results were attractive and functional.

Generating an original design from scratch

This is where Stitch shines

For my final test, I decided to remove reference images from the equation entirely and let Stitch work under its own steam. I asked it to create a premium Andes Adventure landing page, with a hero section, calls to action, experience cards, reasons to travel with the company, and a final booking prompt.

The results were the strongest so far, producing a more ambitious design, with a cohesive aesthetic, attractive imagery, and multiple sections that felt like parts of the same website. It also generated alternative sections and supporting content that gave me more ideas to play with.

This removes much of the legwork for anyone starting a project from scratch, providing a ready-made, dynamic creative complete with editable color swatches, fonts, and design elements. With Stitch, instead of staring at a blank canvas, I can describe what I want and quickly get something tangible to evaluate and adapt to my vision and requirements.

Stitch versus Figma: which is better?

Two tools with different strengths

Contrary to my first impression, Stitch gave me dynamic elements, editable color swatches and font choices, and multiple export options, including generated code. I quickly realized it is far more than a tool for producing flat screenshots or mockups, and I can see it becoming a regular tool for producing design features or even full creatives.

Figma, meanwhile, is a tried-and-tested design environment that supports detailed editing, reusable components, collaboration, and prototyping. I like that I can refine designs, easily maintain consistency across different tasks, and share results with team members at any stage of the process.

For now, I’ll keep using Stitch as a starting point for exploring ideas quickly, but as I get more familiar with it, it will likely become a more permanent part of my workflow. Figma remains the stronger choice when a project relies on detailed control and collaboration, although it takes much longer to achieve similar results.

Stitch is a serious design companion, but it’s not a replacement—yet

Google Stitch made it easy to turn a brief into an attractive interface, experiment with colors, fonts, and layouts, and generate a website concept in minutes with minimal fuss. However, it fell short when asked to refine that concept, and adding or removing elements produced wildly varying results.

Yet AI-generated design is here to stay, and I believe tools like Figma and Affinity should pay attention to Stitch, as it makes the design process more accessible and, if incorporated into those platforms, would simplify and improve the workflow. In fact, Adobe Firefly already demonstrates how AI can be integrated into creative tools, and I can see more programs following suit in the not-too-distant future.