From Playground to Proving Ground: The Evolution of Google Labs

A deep dive into the history of Google Labs.

Post methodology: Gemini Pro 2.5 Deep Research with prompts: Please prepare a report on the history of google labs from the beginning to the present, including specific projects and products; can you enhance the report with references to and quoates from this recent interview with the current head of google labs, josh woodward; can you summarize the report into a substack post aimed at AI founders?; Claude 3.7: please summarize [Deep Research report] into a substack post aimed at AI founders; please rewrite this text based on [supplied] outline. Claude and Gemini outputs were merged, lightly edited and reformatted for the Substack editor.

[Updated post Google I/O 2025]

Google Labs, Google’s sandbox for AI experimentation, was thrust into the AI spotlight last year when its NotebookLM product went viral. Josh Woodward, head of Google Labs, appeared on Training Data to talk about present and future Google Labs experiments, and where AI is heading generally.

Google Labs has served as a fascinating barometer for Google's approach to innovation over the past two decades. Its evolution mirrors Google's transformation from a scrappy startup to a mature tech giant carefully balancing experimentation with strategic focus.

The Original Google Labs (2002-2011): The Innovation Playground

The original Google Labs launched in 2002 as a public-facing "playground" where users could interact with early prototypes and provide direct feedback. This embodied Google's early culture of open experimentation and its famous "20% time" policy, which allowed engineers to dedicate one day per week to side projects.

This era saw the incubation of features and products that would become core Google offerings:

• Gmail features like "Send & Archive" and the invaluable "Undo Send"

• Google Maps components and iterations

• Google News aggregator

• Google Reader (a beloved RSS reader)

• Various experimental tools like Similar Images, News Timeline, and Public Data Explorer

The original Labs represented Google's commitment to bottom-up innovation and the "launch early, iterate often" philosophy that defined its early culture. It served as both a practical testing ground and a powerful symbol of Google's experimental ethos.

The 2011 Shutdown: "More Wood Behind Fewer Arrows"

When Larry Page returned as CEO in 2011, he initiated a strategic overhaul aimed at increasing focus and efficiency. This led to the shutdown of Google Labs in July 2011, marking a significant pivot in Google's approach to innovation.

Page's mantra became "more wood behind fewer arrows," signaling a shift from broad experimentation to concentrated resource allocation on high-priority areas. This strategic realignment prioritized core products (Search, Ads) and emerging platforms (YouTube, Android, Google+).

The closure sparked controversy, with critics arguing it betrayed Google's innovative spirit. It also cast doubt on the future of the celebrated "20% time" policy, as Labs had served as the most visible showcase for projects born from this initiative.

Post-2011: Evolution of Structured Experimentation

After the public Labs shutdown, Google's approach to experimentation evolved through more structured channels:

Area 120 (2016-Present)

In 2016, Google launched Area 120, an in-house incubator where employees could develop entrepreneurial ideas with clear paths toward integration or spin-off. Notable graduates include Tables (to Google Cloud), AdLingo (to Workspace), and Touring Bird (to Google Travel).

Unlike the open playground of the original Labs, Area 120 represented a more formalized approach with greater strategic oversight and clearer business objectives.

The Internal "Labs" Revival (2021)

In 2021, Google created an internal division also named "Google Labs," led by Clay Bavor. This consolidated forward-looking initiatives including Area 120, AR/VR projects, and Project Starline (holographic video conferencing). This was a strategic organizational move to elevate long-term technological bets, rather than a return to the public playground model.

Google Labs Today: The AI Frontier (2023-Present)

The latest chapter began in 2023 with the return of Google Labs as a public-facing entity, now explicitly focused on AI experiments. Led by Josh Woodward, today's Labs serves as "the home for AI experiments at Google," providing a platform for users to discover and provide feedback on Google's latest AI products.

Key experiments and their recent advancements showcased at Google I/O 2025 include:

• NotebookLM: This AI-powered research assistant, which allows users to upload source materials and ask questions or generate summaries based on those documents, has expanded its capabilities. It is now available on Android and iOS, offering features like offline access and "Audio Overviews" which generate podcast-style briefings from notebook content. Recent updates also include the ability to create Mind Maps for visual summarization and it's now powered by Gemini 2.5.

• Project Mariner: Now evolving into "Agent Mode," this research prototype built on Gemini 2.0 focuses on automating tasks within a browser, such as purchasing tickets or booking reservations. Its capabilities are being brought to AI Mode in Search (available in Labs) and to developers via the Gemini API. Project Mariner is available in the US to Google AI Ultra subscribers.

• Veo: Google's state-of-the-art video generation model. The latest version, Veo 3, offers improved realism, prompt adherence, creative control, 4K output, and can natively generate audio, including dialogue and sound effects. Veo is being integrated into tools like Flow, an AI filmmaking tool designed for Veo that evolved from Google Labs' VideoFX experiment. Flow allows users to create and edit AI video clips, and access to Veo 3 with audio generation is available through the premium Google AI Ultra plan. Veo is also accessible via Vertex AI.

• Whisk: A generative media experiment from labs.google/fx (which seems to be a new branding or sub-section of Google Labs for creative AI tools). Whisk focuses on fast visual ideation by allowing users to combine visual elements (uploaded images or text prompts) to create new images, powered by Gemini for visual understanding and Imagen 4 for generation. It aims to simplify image creation without requiring deep prompting knowledge. Whisk Animate further allows users to transform these generated images into short videos using Veo 2.

• Search Generative Experience (SGE) / AI Mode in Search: SGE continues to evolve, with AI Mode in Search becoming more prominent and even graduating some features to the core Search experience. This includes bringing agentic capabilities from Project Mariner for tasks like event ticket purchasing and restaurant reservations into AI Mode in Labs. AI Mode is also integrating deep research capabilities and live capabilities from Project Astra.

• Imagen 4: Google's newest image generation model, offering enhanced photorealism, detail, and improved spelling and typography in images. It is available in the Gemini app and powers tools like Whisk.

• Workspace AI tools: Features like "Help me write" in Docs and Gmail continue to be part of Labs' exploration of applied AI.

Unlike the original Labs' broad scope, the current iteration is narrowly focused on artificial intelligence, reflecting AI's position as Google's central strategic priority. Access is more controlled, with waitlists and trusted tester programs managing rollouts.

The Strategic Evolution

Google Labs has consistently served as a strategic lever, used differently across eras:

1. 2002-2011: A public showcase for bottom-up innovation and 20% time projects

2. 2011-2016: A period of focus and integration following the shutdown

3. 2016-2022: More structured experimentation through Area 120 and internal teams

4. 2023-Present: A targeted platform for AI experimentation and rapid iteration

This evolution reflects Google's journey from prioritizing freewheeling experimentation to balancing innovation with strategic focus as a mature tech giant. Today's AI-focused Labs represents Google's all-in bet on artificial intelligence, serving as a critical proving ground for integrating AI into its core offerings while exploring novel applications.

The Labs concept continues to evolve as Google navigates the future, demonstrating how innovation structures must adapt to changing technological landscapes, competitive pressures, and organizational needs.

Lessons for AI Founders from Google Labs & Josh Woodward

So, what can AI founders take away from Google's evolving approach and Woodward's insights?

1. Speed is Everything (Especially Now): The current Labs team prides itself on moving incredibly fast. "We really pride ourselves in trying to be really fast moving as a culture. So we’ll go from an idea to end users’ hands, 50 to 100 days," says Woodward. In an "AI platform shift moment," this rapid iteration cycle is critical.

2. Focus on the Application Layer: Groundbreaking models are essential, but Woodward stresses that value comes from applying AI to solve real problems. Ask: How can AI enhance user workflows, creativity, or productivity in tangible ways?

3. Build for the Future Models: Don't just build for today's AI capabilities. Strategically "make sure your product is aligned to the models getting smarter, cheaper, faster". Anticipate the trajectory of AI development.

4. Big Things Start Small: Even at Google scale, Labs celebrates early traction. "We get really excited if we get, like, 10,000 weekly active users," Woodward notes. Focus on solving specific user pain points first; scale comes later.

5. Embrace Co-Creation: The Labs team has "one foot in the outside world... co-creating with startups and others, but also one foot inside Google DeepMind". Founders should actively seek feedback, partnerships, and insights from both users and the broader research community.

6. Build the Right Team & Culture: Labs attracts a mix of Google veterans and ex-founders, seeking creative "underdogs" with "hustle". Crucially, they "normalize things like failure," recognizing that if everything works, "you’re not swinging big enough". Building a resilient, fast-moving team that isn't afraid to fail is key.

Different Innovation Models Exist: It's worth noting Labs isn't Google's only approach. X (the Moonshot Factory) tackles radical, long-term bets (like Waymo). Labs focuses on nearer-term AI applications. Recognize what kind of innovation you're pursuing and structure accordingly.

Watch the full episode:

Related Episode: Google NotebookLM’s Raiza Martin and Jason Spielman on the Potential for Source-Grounded AI