Google Labs is in the middle of a quiet productivity revolution. Rather than announce a single monolithic AI product, Google is shipping four small, focused beta tools that each solve one layer of the creation pipeline — and they are all free to try right now.
They map to distinct stages of building something from scratch:
| Stage | Tool | What it does |
|---|---|---|
| Design | Stitch | Turn a prompt into UI designs and frontend code |
| Marketing | Pomelli | Read your site, build brand DNA, generate on-brand content |
| Automation | Opal | No-code AI mini-apps from natural language |
| Visuals | Whisk | Blend subject + scene + style into new images |
They are experiments, not finished products. But together they sketch what Google’s AI-assisted workflow could become.
Stitch: From Prompt to UI in Seconds
Stitch is Google’s AI-native design canvas. You write a text prompt describing an app or interface, and Stitch generates a visual design plus the HTML/CSS code behind it — in minutes.
What it does well:
- Generate mobile and web UIs from a single text prompt
- Export to Figma with proper Auto Layouts and named layers
- Generate multiple design variants for rapid exploration
- Export clean HTML/CSS (including Tailwind) that developers can use as a starting point
- Recent MCP server integration lets AI coding agents (like Antigravity) connect two-way with Stitch designs
What it does not do:
- Stitch does not generate backend logic or database schemas
- Designs are starting points, not final products — you still need design refinement and development
- It is in beta, so features and availability change
The most compelling use case is rapid prototyping: describe an idea, get a visual mockup and exportable code, iterate quickly. The MCP integration is the interesting technical development — it creates a feedback loop where your coding agent can read Stitch designs and make direct layout edits.
Where Stitch fits: Early-stage design exploration and developer handoff. If you are sketching app ideas and need something clickable fast, Stitch cuts the design-to-code gap significantly.
Pomelli: Your Brand DNA, Automated
Pomelli does something different. Instead of generating from a blank prompt, it reads your existing website and builds a Business DNA profile — capturing your tone of voice, fonts, imagery, and color palette. Then it generates on-brand social posts, ads, and campaign assets that actually match your existing look and feel.
What makes Pomelli different:
- Brand extraction happens first — no guessing at colors or tone
- Generates social posts, ad copy, and visual assets consistent with your brand
- Has a Photoshoot feature that turns flat product photos into studio-style imagery
- Free during beta with several hundred image and video generations
Pomelli is aimed at small businesses that cannot afford an agency. If you have a website, it can produce a consistent stream of marketing content without you writing a single brief.
Where Pomelli fits: Small business marketing and rapid content creation. If you are spending hours crafting Instagram posts that match your brand, Pomelli can produce first-pass drafts in seconds.
Opal: No-Code AI Mini-Apps
Opal is Google’s answer to no-code workflow builders, but with AI baked in from the start. You describe an app in natural language, and Opal builds a multi-step workflow — chaining prompts, model calls, and tools together — that runs as a shareable mini-app.
Key capabilities:
- Write natural language descriptions; Opal builds the workflow visually
- Chain together steps: user input → AI generation → tool calls → output
- Built-in tools include web search, maps, weather, and more (via the
@prefix) - Apps are shareable immediately — no hosting or deployment needed
- Runs on a Gemini 3 Flash agent
Opal is closest to what you might use n8n or Zapier for, but designed for AI-native tasks from the ground up. The visual editor lets you see and tweak every step, and the natural language editor lets you iterate by simply describing changes.
Where Opal fits: Prototyping AI-powered workflows and mini-apps. If you have a repetitive task that involves searching the web, processing text with an LLM, and producing an output, Opal can turn that into a shareable app without writing code.
Whisk: Image Blending, Not Prompting
Whisk takes a visual-first approach to image generation. Instead of writing prompts, you drag in three images — a subject, a scene, and a style — and Whisk blends them into something new using Gemini and Imagen 3.
How it works:
- Subject — a person, pet, object, or character
- Scene — the environment or background
- Style — artistic treatment (Sticker, Plushie, Capsule Toy, Enamel Pin, Chocolate Box, Card)
- Optional text can refine the result
This removes the barrier of prompt engineering entirely. A designer can drop in a character sketch and a watercolor reference and get a styled illustration in seconds. The preset styles feel genuinely fresh — not just “realistic” or “anime,” but specific aesthetics like enamel pins or capsule toys.
Where Whisk fits: Rapid visual exploration for designers, content creators, and marketers who want to iterate on aesthetics fast. It is not a replacement for professional illustration, but it is a powerful brainstorming tool.
The Full Stack, Together
What excites me about these four tools is that they can chain together:
- Ideate with Stitch — get a visual UI concept and exported code
- Brand it with Pomelli — extract brand DNA and generate on-brand assets
- Automate with Opal — build workflows that combine web data, AI processing, and your Stitch-generated UI
- Visualize with Whisk — create promotional images and social assets from your app’s actual design
None of them requires a paid plan. None of them is production-critical — they are all experiments that Google can sunset. But for creators, startups, and small teams, the combination is powerful: you can go from idea to a branded, automated, visually rich mini-product using only free Google tools.
Honest Limits
These are beta experiments, and Google has a track record of sunsetting Labs projects. The tools also require a Google account and are available in English only, for users 18+. Feature availability varies by country.
My recommendation: start with the one layer that matches your biggest gap. Designers should try Stitch. Marketers with an existing site should try Pomelli. Builders of workflows should try Opal. And if you just want to play with AI image generation, Whisk is the most intuitive entry point.
All four are free to try at stitch.withgoogle.com, labs.google.com/pomelli, opal.withgoogle.com, and labs.google/fx/tools/whisk.
Related Resources:
