AI Tools ‐ Google - spinningideas/resources GitHub Wiki
Below are the primary web-based AI tools offered by Google as of late 2025.
Description Google AI Studio is a browser-based prototyping environment for developers to experiment with Gemini models. It offers a low-barrier entry to prompt engineering, allowing users to test prompts, multimodal inputs (text, images, video), and structured outputs (JSON). It is the fastest way to start building with the Gemini API.
How to Get Started
- Go to aistudio.google.com.
- Sign in with your Google Account.
- Click "Create new" to start a prompt or chat interface.
- Test your prompts against various models (Gemini 2.0 Flash, Pro, etc.).
Cost
- Free Tier: Generous free tier for prototyping and testing.
- Pay-as-you-go: Pricing applies if you exceed free tier limits or move to production scaling (pricing follows Gemini API costs).
API Key Needed?
- Yes. You can generate an API key directly within the AI Studio interface to use the Gemini API in your external applications.
Extra Notable Features
- see Veo 3 : https://aistudio.google.com/apps/bundled/veo_studio?showPreview=true&showAssistant=true
- use image generation tool: https://deepmind.google/models/imagen/ - links to whisk or gemini image tool
Example Usages
- Non-Technical: Meal Planner. "Paste your dietary restrictions and ingredients into the prompt, then ask the model to generate a weekly meal plan with a shopping list sorted by grocery store aisle."
- Technical: Log Analyzer. "Paste raw server logs into the context window. Instruct the model to 'Identify the root cause of the 500 error sequence on timestamp X' and output the findings as a structured JSON object."
- Build with Gemini in AI Studio (Google Cloud Tech, ~10 min)
- Quickstart: AI Studio walkthrough (Google for Developers, ~8 min)
Description NotebookLM is an AI-powered research assistant that grounds its answers in your specific documents. You can upload PDFs, text files, or link Google Drive docs, and the model (Gemini) will answer questions, summarize, and synthesize information strictly from those sources, reducing hallucinations.
How to Get Started
- Go to notebooklm.google.com.
- Create a "New Notebook".
- Add sources (Upload PDF, Paste text, Drive link, website URL).
- Chat with your documents or generate summaries/audio overviews.
Cost
- Free: Currently free for personal use (with usage limits on source count/size).
- NotebookLM Plus: Included in Google One AI Premium subscription for higher limits.
API Key Needed?
- No. It is a SaaS application; no API key is needed to use the interface.
Example Usages
- Non-Technical: Study Buddy. "Upload a 40-page PDF insurance policy or textbook. Ask questions like 'What is the deductible for water damage?' or generate an 'Audio Overview' podcast to listen to the key points while commuting."
-
Technical: Legacy Docs Query. "Upload a set of 5 different deprecated API reference PDFs. Chat with the docs to find specific endpoint schemas (e.g., 'What is the payload format for
/user/update?') without manually grepping through files."
Description Google Flow is an AI filmmaking tool designed for creators to produce cinematic clips, scenes, and stories with visual consistency. It leverages Google’s advanced generative models, including Veo (for video), Imagen 4 (for images), and Gemini, allowing users to create high-quality video content from text prompts or existing frames.
How to Get Started
- Go to labs.google/fx/tools/flow.
- Click "Create with Flow" to enter the tool.
- Sign in with your eligible Google Account (requires specific subscription).
Cost
- Paid Subscription: Requires a Google AI Pro, AI Ultra, or AI Ultra for Business plan.
- Workspace: Select Workspace plans include monthly AI credits.
API Key Needed?
- No. It is a web-based creative tool; no API key is needed to use the interface.
Example Usages
- Non-Technical: Mood Board. "Describe 'A romantic sunset on a beach with soft focus' and use the editor to stitch together generated clips into a cohesive 30-second background loop for a wedding video."
- Technical: Game Asset Prototyping. "Rapidly generate placeholders for a game prototype. Create varied 'sci-fi corridor' video loops to test lighting engines and atmosphere before 3D artists commit to final models."
- Introducing Flow (Google DeepMind/YouTube, ~6 min)
- Create cinematic stories with Flow (Google Labs, ~12 min)
Description Stitch is an AI-powered design tool that enables users to create high-fidelity UI prototypes for web and mobile applications simply by describing them. It uses advanced models (like Gemini) to interpret design intent and generate visual layouts and components, accelerating the design-to-code workflow.
How to Get Started
- Go to stitch.withgoogle.com.
- Click "Try now" to access the dashboard.
- Sign in with your Google Account.
- Start a new "App" or "Web" project by typing a design prompt.
Cost
- Free (BETA): Currently in Google Labs beta and appears to be free for personal experiments.
API Key Needed?
- No. It is a browser-based design tool; no API key is needed.
Example Usages
- Non-Technical: Event Page Mockup. "Describe 'A clean, minimalist wedding RSVP page with a countdown timer and a photo gallery'. Interact with the generated UI to adjust colors by saying 'Make it warmer' to show a designer your vision."
- Technical: Client Demo Bootstrap. "Prompt 'A SaaS dashboard with a sidebar navigation, a main activity graph, and a user table'. Export the generated HTML/CSS/React code to jumpstart the frontend development process."
Description Gemini Canvas is a dedicated workspace within same Gemini interface designed for writing and coding projects. It provides a side-by-side view where you can generate content in a "canvas" (e.g., a document or code editor) while chatting with Gemini to refine, edit, or debug it. It is ideal for iterative creative work and building web apps.
How to Get Started
- Go to gemini.google.com/canvas.
- Or, open gemini.google.com and select "Canvas" from the mode selector (if available) or simply ask Gemini to "open a canvas".
- Start by asking Gemini to "write a blog post about..." or "create a python script to...".
Cost
- Free: Available to personal Google Accounts (uses standard models).
- Advanced: Included in the Gemini Advanced subscription (offering arguably better models and larger context windows for complex projects).
API Key Needed?
- No. It is a feature of the consumer Gemini web interface.
Example Usages
- Non-Technical: Blog Co-author. "Draft a marketing post in the canvas. Highlight a paragraph and ask Gemini to 'Rewrite this to be punchier' or 'Suggest 5 SEO-friendly titles', seeing the changes inline."
- Technical: Code Refactor. "Paste a legacy Python script. Highlight a large monolithic function and ask 'Refactor this into smaller, testable functions with type hints'. Verify and edit the code directly in the side-by-side view."
Description Gemini Gems are custom versions of Gemini that you can create to act as expert assistants on specific topics. You can give them unique instructions, knowledge, and personalities to handle repetitive tasks or provide specialized advice (e.g., a "Coding Mentor", "Creative Writing Coach", or "Gym Buddy").
How to Get Started
- Go to gemini.google.com.
- Look for "Gems" in the sidebar menu or Gem manager.
- Click "Create a Gem" or select a pre-made Gem to start chatting.
- Provide instructions to define your Gem's behavior and knowledge.
Cost
- Paid Subscription: Creation and use of Gems generally requires a Gemini Advanced subscription (part of Google One AI Premium) or a Google Workspace Business/Enterprise add-on.
API Key Needed?
- No. It is a feature of the consumer/business Gemini web interface.
Example Usages
- Non-Technical: Sous Chef Persona. "Create a Gem instructed to 'Always suggest substitutions for dairy allergies and convert units to metric'. Chat with it daily for personalized cooking advice."
-
Technical: Strict Code Reviewer. "Create a Gem primed with your team's style guide (e.g., 'Use snake_case, no
eval(), prefer async/await'). Paste code snippets to it for a pre-commit check."
Pomelli is an experimental AI marketing tool from Google Labs and Google DeepMind that analyzes your website to build a “Business DNA” (tone, colors, fonts, images) and then generates on-brand social media campaign ideas plus ready-to-use image/text assets. It targets small-to-medium businesses that need fast, consistent, multi-channel creative.
- Visit labs.google.com/pomelli (public beta; English).
- Sign in with your Google account.
- Enter your website URL; Pomelli builds your Business DNA automatically.
- Choose a suggested campaign idea or enter your own prompt.
- Review/edit the generated creatives (text + images) and download the assets for posting.
- Public beta with no published pricing yet (experiment phase). Expect usage limits; final pricing TBD.
- Not currently. Pomelli is a web app experiment; no public API advertised yet.
- Business DNA: auto-learns brand voice, colors, fonts, imagery from your site to keep outputs on-brand.
- Campaign ideation: generates campaign angles/ideas you can refine with custom prompts.
- Asset generation: produces multi-asset sets (social posts, captions, images) you can edit inline, then download.
- Early-access scope: currently available in the US, Canada, Australia, and New Zealand (English).
- Non-Technical: Local café enters its website, Pomelli builds brand DNA, suggests a “Fall latte launch” campaign with posts and visuals; owner tweaks copy and downloads assets for Instagram and Facebook.
- Technical/Team: A startup marketer seeds Pomelli with the product site, asks for a “Q1 feature spotlight” campaign, gets multiple on-brand variants, edits text tone to match product positioning, then exports assets for scheduled social posts.
Opal is a Google Labs experimental tool that lets you build and share AI "mini-apps" by chaining prompts, Gemini model calls, and tools using natural language plus a visual editor—no code required. It’s aimed at rapid prototyping and sharing workflows that turn plain-language descriptions into functional AI apps.
- Go to opal.google/landing (public beta, currently US-only per launch notes).
- Sign in with your Google account.
- Start from a gallery template or create a new Opal from scratch.
- Describe the workflow or edit steps in the visual editor (prompts, tool calls, inputs/outputs).
- Test, iterate, and then share the Opal for others to use with their own Google accounts.
- Experimental/public beta; no public pricing published. Expect usage limits while in Labs.
- No API key required for the Opal builder itself; access is via Google account in the web experience. (No public API advertised yet.)
- Visual + natural language editor: build and modify steps either via UI or conversational edits.
- Workflow chaining: connect multiple prompt/model/tool steps into a multi-stage app.
- Gallery templates & remixing: start from prebuilt Opals and customize.
- Sharing: publish an Opal so others can run it immediately with their Google account (hosting handled by Opal).
- Console/details view: inspect inputs/outputs, thinking steps, and tool calls for debugging.
- Non-Technical: Describe “plan my weekly meals from a grocery list and output a shopping checklist,” tweak steps in the visual editor, then share with family.
- Technical/Team: Prototype a lead-qualifier app that ingests form submissions, scores leads with a Gemini call, enriches via a tool step, and outputs a Google Sheet—built and iterated via natural language edits.
- Introducing Opal: build AI mini-apps (Google Developers, ~8 min)
- Opal tutorial: workflows with natural language (Google Labs, ~12 min)
Description Vertex AI is Google Cloud's unified, enterprise-grade machine learning platform. It allows users to train, deploy, and manage ML models and AI applications. It includes Vertex AI Studio for generative AI, AutoML for training without code, and model garden for accessing open-source and first-party models. It is designed for scalability, security, and compliance.
How to Get Started
- Go to the Google Cloud Console.
- Select or create a Google Cloud Project.
- Enable the "Vertex AI API".
- Navigate to "Vertex AI" in the menu to access the dashboard and Studio.
Cost
- Free Trial: New Google Cloud users usually get $300 in credits.
- Pay-as-you-go: Charged based on compute hours (training/prediction), storage, and token usage for generative models.
- Enterprise: Custom pricing for large scale commitments.
API Key Needed?
- Yes. Access implies using Google Cloud credentials (Service Accounts/OAuth) or API keys associated with the project.
- Intro to Vertex AI (Google Cloud Tech, ~14 min)
- Vertex AI quickstart for generative AI (Google Cloud Tech, ~12 min)
Example Usages
- Non-Technical: Sentiment Dashboard. "A Marketing Analyst uploads a CSV of 10,000 survey responses to a Vertex App which uses a pre-trained model to categorize feedback as 'Positive', 'Negative', or 'Neutral' without writing code."
- Technical: Custom Model Fine-tuning. "An ML Engineer sets up a pipeline to fine-tune Gemini 1.5 Pro on a dataset of internal support tickets. The goal is to improve the accuracy of auto-responses for a specific product line."
Description Google Colaboratory (Colab) is a hosted Jupyter Notebook environment that runs in the browser. It allows users to write and execute Python code, making it ideal for data science, machine learning, and education. It provides free access to computing resources including GPUs and TPUs.
How to Get Started
- Go to colab.research.google.com.
- Sign in with your Google Account.
- Click "New Notebook" to start writing Python code.
Cost
- Free: Basic access to standard GPUs/TPUs (with usage limits/interruptions).
- Colab Pro ($9.99/mo): Priority access to faster GPUs and longer runtimes.
- Colab Pro+ ($49.99/mo): Background execution and even higher priority.
- Colab Enterprise: Pay-as-you-go integration with GCP.
API Key Needed?
- No. Not for the tool itself. However, if you call external APIs (like Gemini) from within Colab, you will need those keys.
Example Usages
- Non-Technical: One-Click Visualization. "Opening a shared link to a 'Spotify Data Analyzer'. Clicking 'Run All' to execute pre-written Python cells that visualize your personal listening history, without needing to understand the code."
- Technical: Model Experimentation. "Using the free T4 GPU runtime to experiment with different PyTorch neural network architectures. Visualizing loss curves in real-time before committing to a costly training run on a cluster."
Description Firebase Studio is an AI-assisted workspace for full-stack, multiplatform app development in the cloud. It provides a web-based IDE (familiar to VS Code users) with built-in simulators (Android/iOS) and generative AI assistance for coding.
How to Get Started
- Go to firebase.google.com/studio.
- Sign in with your Google Account.
- Choose a template (Next.js, Flutter, etc.) or import a repo to create a workspace.
Cost
- Free: Currently in public beta/preview and free to use.
- Future: Likely to have tiered pricing similar to other IDE cloud services.
API Key Needed?
- No. The environment itself is an IDE. (You would use API keys for services you build within it).
Example Usages
- Non-Technical: Prototype Review. "A Product Owner accesses the cloud workspace to click through the 'Login' and 'Onboarding' flow of a live, work-in-progress application to validate UX decisions."
- Technical: Full-Stack Acceleration. "A Developer asks the integrated AI: 'Create a User Profile feature'. The AI generates the Next.js frontend component AND provisions the necessary Firestore database schema and security rules in the cloud."
Description Project Jules (jules.google) is an AI-powered asynchronous coding agent. Unlike chat interfaces, Jules lives where the code lives—integrating directly with GitHub. It explores repositories, plans changes, and executes complex tasks (like migrations, bug fixes, or feature additions) by creating pull requests with detailed implementation notes.
How to Get Started
- Go to jules.google.
- Sign in with your Google Account.
- Install the Jules GitHub App and select the repositories you want it to access.
- Describe a task in the Jules dashboard (e.g., "Update the dependency version for React and fix breaking changes").
Cost
- Free (Preview): Currently free during the preview period.
- Future: Expected to have tiered pricing (e.g., Pro, Ultra) based on usage intensity.
API Key Needed?
- No. It uses a GitHub App installation for access.
Example Usages
- Non-Technical: Content Updates. "A Marketing Manager assigns a task: 'Update the copyright year to 2026 and change the support email address on all footer files'. Jules finds every instance and creates a PR for review."
-
Technical: Design System Migration. "Assigning a complex chore: 'Find all buttons using inline styles and replace them with our new
<Button>component from the design system package'. Jules handles the tedious refactoring across dozens of files."
Veo is Google’s flagship text-to-video generation model (latest: Veo 3) delivering high-fidelity, cinematic video with stronger temporal consistency and controllable camera motion. It’s available via Google AI Studio and is being integrated into creator experiences like YouTube (Dream Screen/VideoFX), often paired with Imagen 3/4 for reference frames.
- Go to aistudio.google.com and create a new video project (model: Veo) if available in your region/account.
- Sign in with your Google Account and accept the experimental terms.
- Provide a text prompt (optionally add reference images/frames) and generate preview clips.
- Refine with style/camera controls; download or export to supported surfaces (e.g., Shorts/VideoFX) as access rolls out.
- Experimental; pricing not broadly published yet. Expect usage quotas while in Labs/preview.
- Not for the AI Studio UI. If/when API access is offered, it would use a Gemini-style API key.
- Higher-res, longer clips with improved temporal consistency and camera trajectories.
- Reference-to-video support for style/subject consistency; pairs with Imagen frames.
- Emerging native audio support (Veo 3) for synchronized sound.
- Creator-focused integrations (YouTube Dream Screen/VideoFX) announced for rollout.
- Non-Technical: "Create a 10s cinematic drone shot over misty mountains at sunrise, slow pan right."
- Technical/Team: Generate B-roll variants for a product launch video, using reference frames to keep branding consistent across scenes.
Imagen 3 and Imagen 4 are Google’s latest high-fidelity text-to-image models, focused on photorealism, fine detail, and text rendering improvements. They’re exposed via AI Studio and Labs experiences like ImageFX, and often pair with Veo/Flow to provide reference frames for video workflows.
- Go to aistudio.google.com and choose the Imagen image generation app (or use ImageFX if available in your region).
- Sign in with your Google Account and accept experimental terms.
- Enter a text prompt; optionally provide image references or style guidance.
- Generate, refine with safety/variations, and download/export images.
- Experimental; no broad public pricing yet in Labs/UI. Expect quota-based usage while in preview.
- Not for the web UI (AI Studio/ImageFX). If/when API endpoints are exposed, they would likely use Gemini-style API keys.
- High realism with better text rendering and typography than prior Imagen versions.
- Style and reference guidance for consistency across sets of images.
- Safety filters and grounded generations aligned with Google’s policies.
- Integrations as reference frames for Veo/Flow video pipelines.
- Non-Technical: "Generate a cozy reading nook with warm lighting, plants, and mid-century furniture for a mood board."
- Technical/Team: Create a consistent set of product hero shots (multiple angles/backgrounds) to feed into a Veo storyboard for launch B-roll.
- Imagen 3: next-gen image generation (Google DeepMind, ~6 min)
- Design with Imagen in AI Studio/ImageFX (Google, ~8 min)
Description Mixboard is an experimental, AI-powered "concepting board" from Google Labs designed for visual brainstorming and idea development. It provides an open canvas where users can generate, organize, and refine ideas using a combination of images and text. Powered by Gemini models and a specialized image editing model called Nano Banana, it allows users to transform abstract concepts into visual mood boards with natural language commands.
How to Get Started
- Go to mixboard.google.com or labs.google/mixboard.
- Sign in with your Google Account.
- Click "New project" and enter a text prompt to start a board from scratch, or select one of the pre-populated "Example" boards to explore existing concepts.
- Currently, the tool is available as a public beta in the U.S.
Cost
- Free: As an experimental project in Google Labs, Mixboard is currently free to use during its beta phase.
API Key Needed?
- No. Mixboard is a standalone web-based experimental tool. There is currently no public API for developers to integrate Mixboard's specific "canvas" functionality into external applications, although it uses Google's underlying AI models.
Extra Notable Features
- Nano Banana Model: A specialized model for fine-grained image editing and combining visuals using natural language instructions.
- Multimodal Canvas: The ability to generate text descriptions or project ideas based on the context of images already placed on the board.
- One-Click Iteration: Features like "Regenerate" and "More like this" allow users to quickly explore variations of a specific visual style.
- Community Integration: A dedicated Google Labs Discord channel is available for users to share creations and provide feedback directly to the developers.
Example Usages
- Non-Technical: Event Planning. "Create a 'Cozy Fall Living Room Party' board. Prompt the AI for rustic decor images, then use natural language to 'add more pumpkins' or 'change the lighting to warm candlelight' to visualize the event's atmosphere."
- Technical/Creative: Product Design Prototyping. "Generate a board for 'Memphis Dinnerware Inspiration.' Use the Nano Banana model to combine specific geometric patterns with 3D ceramic mockups, then use the 'more like this' feature to rapidly iterate on a series of cohesive product concepts for a design pitch."