Table of Contents
- Why these updates matter
- 1. Create studio-ready product photo shoots from home with Pomelli
- 2. Gemini inside Chrome: Meet Gemini 3.1 Pro (agentic capabilities)
- 3. Prompt quality matters: craft better prompts for images and beyond
- 4. NotebookLM gets real editing: revise slides on the fly
- 5. Stitch: export to Figma and intelligent design ideation (ID8)
- Practical tips to get the best results across these tools
- Ethics, legal, and production considerations
- Suggested assets, meta info, and SEO details for publishing this article
- Where to dive deeper
- Final thoughts
- FAQ
- Call to action
Why these updates matter
Google just rolled out a set of Labs updates that dramatically level up what you can do with Gemini and the rest of Google’s AI toolkit. These are not incremental tweaks. They open up workflows that previously required a small team or specialized tools: product photography from your couch, agent-like automation inside Chrome, fast slide revisions inside NotebookLM, and design problem-solving that spits out usable Figma-ready ideas.
If you sell products, build web apps, make presentations, or just want a smarter browser assistant, these features will save time, money, and friction. Below I break each update down, show how to use it, give practical use cases, and share prompt tips so you get reliable, professional outputs.
1. Create studio-ready product photo shoots from home with Pomelli
The new Pomelli feature in Google Labs is a sleeper hit. Instead of booking a photographer, model, and studio, you can upload a product image, tell Pomelli your brand DNA, pick a template or theme, and produce dozens of polished lifestyle and product shots automatically.
How it works
- Upload a clean product image or choose to generate one from a prompt.
- Fill your brand DNA: logo, tagline, product types, tone, and preferred styles.
- Pick templates and optional holiday / thematic filters (Easter, summer, minimal, urban, etc.).
- Request a photo shoot and download a batch of edits: models, scenes, angles, and contextual props.
Outputs can include models wearing the product, lifestyle shots, and different backgrounds. You can download, tweak, and reuse them across ads, social posts, and store galleries.
Practical use cases
- Ecommerce sellers who need fast hero images and lifestyle shots for listings and ads.
- Small brands that can’t afford professional shoots every season.
- Ad agencies or creators who want to A/B test dozens of creative variations quickly.
Limitations and tips
- Image realism varies by product complexity. Clean, high-contrast product photos work best.
- Be explicit about brand constraints—colors, logos, and text placement—to keep outputs on-brand.
- Watch for potential model likeness or trademark issues when distributing generated imagery commercially; verify licensing and usage terms.
2. Gemini inside Chrome: Meet Gemini 3.1 Pro (agentic capabilities)
Gemini in Chrome just gained a major upgrade: the option to use Gemini 3.1 Pro as your model. This is not just a faster chatbot. It brings agentic capabilities that automate tasks inside your browser and reason over multiple web sources, calendars, and maps.
What Gemini 3.1 Pro does better
- Advanced reasoning for complex, multi-step tasks (it can plan routes, build itineraries, and check calendars).
- Improved math and coding support for developer workflows.
- Agentic behavior: it can interact with browser elements, open tabs, fetch data, and generate actionable plans.
Example: planning a day trip
Ask Gemini 3.1 Pro to plan a day trip from your current location. It can:
- Detect your location and propose nearby options (towns, parks, attractions).
- Create a timed itinerary: morning hike, lunch recommendation, afternoon sights.
- Plot driving routes and approximate durations, flag caveats, and include confidence levels for each suggestion.
That same agentic workflow is useful beyond travel: lead generation, email triage, research summarization, and any repetitive browser task that would benefit from programmatic thought and execution.
How to use it effectively
- Choose Gemini 3.1 Pro in the Chrome model selector when you need multi-step workflows or automation inside the browser.
- Provide context: calendar access, map preferences, and any constraints (budget, time, mobility).
- Verify results—agentic models are powerful but should be double-checked before actioning sensitive things like purchases or calendar invites.
3. Prompt quality matters: craft better prompts for images and beyond
One of the most underrated influences on your AI results is how you write prompts. A basic prompt might get you passable outputs, but a detailed, optimized prompt will produce results that look like they were created by a pro.
Before and after: a Yellowstone image example
A simple prompt like “sunset over Lamar Valley, Yellowstone” might yield a decent image but miss photorealism and atmospheric nuance. An optimized prompt that includes camera type, lens, lighting, mood, composition, and subject details delivers far richer outputs: realistic fog on the river, proper sun position, accurate wildlife placement, and more natural color grading.
Use a prompt optimizer
Tools like MyPromptBuddy (I use it regularly) help convert short prompts into comprehensive, model-friendly inputs. They produce optimized prompts suitable for Gemini, ChatGPT, Claude, and other models. The difference in image quality and consistency is dramatic.
Prompting checklist for photorealistic images
- Specify camera and lens: “full-frame DSLR, 50mm, f/1.8” to influence depth of field.
- Define lighting and time of day: “golden hour, warm backlight, soft shadows.”
- Describe composition: “wide-angle landscape, subject centered in lower third, leading lines.”
- Mood and color grading: “cinematic, high contrast, muted greens.”
- Small observables: “fog over river, grazing bison, ripples in foreground.”
4. NotebookLM gets real editing: revise slides on the fly
NotebookLM’s slide creation was already handy for producing decks quickly. The missing piece was straightforward in-tool editing. The new revise functionality lets you request tweaks to an existing slide deck—change colors, shorten text, or alter tone—and NotebookLM generates an updated deck that only applies the requested changes.
Why this is useful
- No need to restart the creation process when you want a different visual style or more concise copy.
- You can iterate quickly: reduce slide text for a presenter view or expand for a detailed handout.
- It reduces back-and-forth between content generation and slide formatting tools.
How to use revise
- Open your generated deck inside NotebookLM.
- Click More and select Revise.
- Type the change you want, such as “Make titles bright green and remove bullet points” or “Shorten each slide to one sentence for a presenter deck.”
- Review the updated deck and export or present.
The feature is free and available to everyone with access to NotebookLM in Labs.
5. Stitch: export to Figma and intelligent design ideation (ID8)
Stitch continues to evolve into a practical design assistant. Two updates stand out: direct export into Figma and a new ideation model called ID8. Together these changes make design problem-solving faster and more actionable.
ID8: bring a problem, get multiple design solutions
ID8 allows you to upload a screenshot or mockup (for web dashboards, mobile apps, or landing pages) and ask for design solutions to a specific problem like “make this dashboard less cluttered.” The model returns multiple design approaches—minimalist list, bento grid, soft card visuals—plus mockups you can iterate on.
Exporting to Figma
Once you like a proposed redesign, Stitch now offers export options directly into Figma. That shortens the gap between ideation and high-fidelity execution.
Workflow example
- Upload your current dashboard screenshot into Stitch.
- Describe the problem: “Dashboard looks cluttered, users get overwhelmed.”
- Ask ID8 to propose three redesign styles and mockups.
- Choose a mockup, tweak the content, and export to Figma for handoff or further design work.
Stitch’s output is useful for product managers, UX designers, and founders who need fast iterations without a full design sprint.
Practical tips to get the best results across these tools
- Provide context and constraints—brand colors, tone, allowed assets, and usage intent reduce noisy outputs.
- Iterate in small steps—ask for one change at a time when using agentic models or revise functions to avoid unintended global edits.
- Use optimized prompts—tools like MyPromptBuddy or a prompt checklist produce higher-quality images, copy, and designs.
- Keep human review—validate generated imagery for model likeness, trademarked content, and factual accuracy before publishing.
- Export early to production tools—use Stitch to Figma or Pomelli downloads to speed up downstream workflows.
Ethics, legal, and production considerations
These new features accelerate creative production, but they do not remove responsibilities. Generated models and people can unintentionally resemble real individuals. Logos, brand marks, and copyrighted art require care when used in commercial contexts.
Before pushing generated assets live, confirm licensing terms from Google Labs and any model-specific usage rules. If you use images with human subjects, check for model-release policies relevant to your region and platform.
Suggested assets, meta info, and SEO details for publishing this article
Suggested images: screenshots of Pomelli product shoots, a Gemini 3.1 Pro itinerary example, before-and-after image prompts (basic vs optimized), NotebookLM revise UI, and Stitch ID8 mockups. Use descriptive alt text like “Pomelli generated lifestyle product shot of shorts on a model” and “Gemini 3.1 Pro planned day trip itinerary with route map.”
Meta description (150-160 characters): Google Gemini Labs adds Pomelli product shoots, Gemini 3.1 Pro agents, NotebookLM slide revisions, and Stitch ID8 with Figma export—game-changing AI updates.
Tags and categories to consider: Google Gemini, AI tools, product photography, prompt engineering, NotebookLM, Stitch, Figma export, AI design, AI in Chrome.
Where to dive deeper
- Try out prompt optimizers to learn how small additions to prompts change outputs dramatically.
- Test Pomelli with a single SKU to build a creative library before scaling to your full catalog.
- Start using Gemini 3.1 Pro inside Chrome for one automated task—calendar checks or lead warmups—to get comfortable with agentic behavior.
Final thoughts
These Google Labs updates together represent a shift from AI as an assistant to AI as an operational partner. Whether you are a creator, founder, marketer, or designer, the barrier to producing high-quality visuals, automating browser workflows, and iterating on designs has dropped substantially.
Use the tools responsibly, invest a little time in better prompts, and you’ll see outputs that feel like they came from a small, highly skilled team.
FAQ
Can I use Pomelli-generated images for e-commerce product listings?
Generally yes, but check Google Labs’ licensing and any model-specific terms. Also verify the generated images do not infringe on trademarks or include recognizable likenesses that require releases. When in doubt, adjust outputs to remove potential conflicts or consult legal counsel.
What makes Gemini 3.1 Pro different from the regular Gemini models?
Gemini 3.1 Pro adds agentic capabilities and stronger reasoning for multi-step tasks. It handles browser automation tasks, workflows that require reading multiple sources, and offers improved coding and math abilities compared to standard models.
Does NotebookLM’s revise feature change the entire deck?
No. Revise applies only the edits you request and generates an updated deck reflecting those changes. It avoids wholesale rewrites unless you ask for a broad transformation.
How do I export Stitch designs into Figma?
After using ID8 to generate mockups, select the desired mockup and choose the Figma export option. The export will create editable layers and components in Figma for further refinement.
Which prompt elements matter most for photorealistic image generation?
Camera and lens, lighting and time of day, composition, mood and color grading, and small scene details (fog, reflections, subject actions) are the most impactful. Provide specific constraints and context for consistent results.
Are these features free to use?
Many Labs features are available for free, but access tiers and limits can apply. Gemini 3.1 Pro and certain export options might be gated by account level or usage quotas. Check Google Labs and product pages for the latest access details.
Call to action
Try one feature today: generate a single Pomelli photo shoot for a SKU, or ask Gemini 3.1 Pro to automate a small browser task. Keep prompts clear, iterate quickly, and share what you create. If you want help setting up prompt templates or automating a specific workflow, I’m available to help.
This article was created from the video Google Gemini’s New Features Are GAME-CHANGERS 👀 (Google Labs Update) with the help of AI.