What Can Specific Google Labs Experiments Like Dreambeans or Google Flow Do?
Google Labs hosts a rotating set of AI experiments, each aimed at a different kind of task. Dreambeans delivers daily personalized story collections, Google Flow is an AI creative studio for bringing stories to life, Google Flow Music is an AI music studio for creating full songs, and Pomelli is an AI marketing tool for building on-brand content. Each one is listed on labs.google with a "Try It Now" or "Learn More" entry point, so you can go straight to the experiment that matches what you want to make.
Creative and media experiments
These are the tools for producing stories, visuals, and music.
Dreambeans
Dreambeans provides personalized collections of stories each day, covering the things that matter most to you. It's positioned as "daily inspiration, brewed just for you," so the intended use is a recurring, low-effort way to get reading material tailored to your interests rather than a one-off search.
Google Flow
Google Flow is described as an AI creative studio that helps you bring your stories to life. If you have a narrative idea and want to turn it into something visual, this is the experiment to open first.
Google Flow Music
Google Flow Music is an AI music studio for creating full songs. It sits alongside Google Flow as the audio counterpart to the visual studio.
Stitch
Stitch is an AI design canvas that transforms natural language into high-fidelity UI you can iterate on and collaborate on. Use it when the output you need is an interface rather than a story or a song.
Marketing and business experiments
Pomelli
Pomelli is an AI-powered marketing tool designed to build scalable, on-brand content so you can connect with your audience faster. The emphasis is on brand consistency at volume, which makes it a fit for teams producing recurring marketing material rather than one-off creative pieces.
Research, coding, and learning experiments
Beyond creative and marketing work, the Labs page lists tools for research, development, and skill-building.
| Experiment | What it does |
|---|---|
| Literature Insights | Finds papers, structures data tables, and creates artifacts like reports and slide decks |
| Hypothesis Generation | Multi-agent research tool that simulates the scientific method to identify knowledge gaps and generate novel hypotheses |
| Computational Discovery | Agentic research engine that generates and scores code variations to help discover models and accelerate iteration |
| Jules | Asynchronous coding agent that automates tasks like bug fixes, writing tests, and adding features |
| Opal | Builds, edits, and shares AI mini-apps using natural language |
| Stax | AI evaluation toolkit that tests models and prompts |
| Mixboard | AI-powered concepting board for exploring, expanding, and refining ideas |
| Learn Your Way | Turns content into a dynamic, tailored learning experience |
| Vantage | Develops and measures skills like collaboration, creativity, and critical thinking through GenAI-simulated teamwork |
| AI Edge Eloquent | Transforms raw, unpolished speech into well-crafted, easy-to-understand text |
| CC | An experimental group agent for families and households to help manage the logistics of running a busy home |
| Play with Putty | A research experiment in collaborative vibe coding that lets teams build tools and websites together in real time |
How to pick the right one
Match the experiment to the output you want:
- Daily reading tailored to you → Dreambeans
- Visual storytelling → Google Flow
- Full songs → Google Flow Music
- On-brand marketing content → Pomelli
- UI design from a text prompt → Stitch
- Idea exploration → Mixboard
- Research and hypothesis work → Literature Insights, Hypothesis Generation, or Computational Discovery
- Coding help → Jules or Play with Putty
- Model and prompt testing → Stax
- Learning and skills → Learn Your Way or Vantage
- Turning speech into clean text → AI Edge Eloquent
- Household logistics → CC
Things to know before you start
Availability is not uniform across the list. Some experiments show a "Try It Now" button, while others show "Learn More," and Play with Putty offers a "Join Waitlist" option. That means access can differ by experiment — a "Learn More" entry may not lead directly into the tool, and a waitlist means you may need to sign up before you can use it. Check the individual entry on labs.google for the current access path rather than assuming every experiment is immediately open.
Google Labs also frames the collection as a place to "discover our latest experiments and help shape the future of AI technology," so the lineup is expected to change over time. If an experiment you read about here isn't listed when you visit, it may have graduated, been renamed, or been retired.