Key Moments
Why every company now needs to think and operate like a lab team | Josh Woodward (VP Google Labs)
Lenny's PodcastKey Moments
Google Labs VP Josh Woodward argues that all companies must operate like labs to stay competitive, but traditional structures often kill innovation within 3-4 years, requiring special autonomy for labs teams.
Key Insights
The most successful ideas rarely come from scheduled 'design sprints' or brainstorming sessions; they emerge organically during unstructured time or from passionate side projects.
Companies should actively monitor the 'almost possible' frontier, creating a "phase transition" when a new capability becomes feasible, and then quickly form a team to capitalize on it.
The true measure of early product-market fit is observing users' eyes light up and their physical engagement with a prototype, not just traditional metrics.
A key skill in the AI era is the 'unlearning rate' – how quickly individuals can abandon old knowledge and adapt to new information.
Labs teams, even successful ones, often need special autonomy and cannot be simply absorbed into existing business units, as they risk becoming inefficient and fading within 3-4 years.
Companies should cultivate a culture where teams feel empowered to signal when an idea isn't working, with the team often recognizing it before leadership does.
The imperative for companies to function as labs
Josh Woodward, VP of Google Labs, advocates for every company to adopt a lab-like operational model due to the rapid pace of technological change, particularly in AI. This shift is crucial for staying ahead of competitors and emerging startups. Companies must become adept at experimenting with the latest models, identifying new possibilities before others, and fostering an environment where unconventional ideas can flourish. However, he notes that traditional corporate structures often hinder innovation, with dedicated lab teams typically fading or becoming ineffective within three to four years if not granted special autonomy. This suggests that the 'how' of integrating lab-like functions into larger organizations is as critical as the 'why'.
Unconventional sources of groundbreaking ideas
Contrary to popular belief, Woodward highlights that the most impactful ideas rarely originate from structured 'design sprints' or dedicated brainstorming sessions. Instead, they often emerge serendipitously during moments of reflection, like swimming, walking in hallways, or even during breaks. Many of Google Labs' successful projects, such as NotebookLM (now Gemini Notebook), Project Genie, and Flow, began with small, intensely curious, and driven individuals who couldn't stop working on something they were passionate about. This underscores the importance of creating space for organic innovation and being constantly vigilant for nascent ideas, rather than trying to 'manufacture' them on demand. The key is to foster a culture where continuous exploration and passion projects are not only tolerated but encouraged.
Identifying and acting on 'almost possible' opportunities
A core strategy within Google Labs involves actively tracking what Woodward calls the 'almost possible' – capabilities that are on the cusp of becoming technologically feasible. This proactive approach involves maintaining a list of such near-term possibilities, viewing their emergence as a 'phase transition' akin to a change of state in matter. When a new possibility crosses this threshold, it signals an opportune moment to form a team and rapidly pursue it. This method is crucial for identifying disruptive opportunities before competitors or startups do. Woodward suggests that for startups, this translates to understanding user problems deeply and combining that with technological feasibility to create something users will pay for. This systematic monitoring of technological frontiers allows for strategic bets on emerging capabilities.
Gauging true product-market fit beyond metrics
Woodward emphasizes that in the early stages of product development, the most reliable indicator of product-market fit is not traditional metrics like daily active users or retention rates. Instead, he advocates for observing users' genuine reactions to early prototypes. He looks for a 'spark' in their eyes, an inclination to lean forward, and enthusiastic engagement when demonstrating the product. This visceral, human-centric feedback is considered the true measure. Furthermore, he advises teams to fall in love with the problem, not the solution, as products often require significant iterations. Focusing on the problem allows for greater flexibility and resilience through multiple pivots, avoiding the trap of becoming overly attached to an initial, potentially flawed, solution.
The critical role of 'unlearning rate' and adaptive skills
In the current AI-driven landscape, Woodward identifies 'unlearning rate' as a significantly increasing valuable skill. This refers to an individual's capacity to quickly learn a new concept or technology and then, just as rapidly, discard it when it's no longer optimal or relevant. This agility is paramount in a field where paradigms shift constantly. Beyond unlearning, he also values 'explosive endurance' – the ability to work intensely and productively for sustained periods without burnout, akin to an athlete managing their energy over a long competition. Collaboration, especially the ability to build trust quickly with both humans and AI agents, is also crucial, as teams are becoming smaller and more fluid.
Knowing when to pivot or kill an idea
Woodward stresses that the decision to end an idea often originates within the team itself, sometimes before leadership recognizes it. Signs include a diminishing team enthusiasm, numerous failed experiments, and a general lack of traction. He advocates for creating an environment where team members feel safe to voice concerns and suggest abandoning projects. A prime example was a feature within Gemini where initial data showed a lack of enthusiasm, prompting the product manager to recommend against its launch, a decision leadership fully supported. This highlights the importance of psychological safety and trusting the team's ground-level insights.
The evolving structure of Google Labs and AI's dominance
Google Labs, now focused entirely on AI, operates across different scales, from 'zero-to-one' (new ideas) to 'one-to-ten' and 'ten-to-hundred' (scaling products). Despite the vastly different scopes, Woodward believes the core talent pool is similar – individuals passionate about AI and building. The guiding principle for all teams, including Labs and Gemini, is 'users first, Google second, our product third,' ensuring alignment with broader company goals. While most projects are AI-centric, Labs also explores hardware, like the 'Google Beam' holographic projector, which can be seen as AI-enabling hardware.
AI's potential to redefine consumer products and Google's data strategy
Woodward sees AI as a major catalyst for innovation in the consumer space, countering the notion that the market is saturated. He highlights opportunities in immersive entertainment, the evolution of messaging and chat, and addressing fundamental human needs like time scarcity and financial well-being. Regarding Google's data advantage, he acknowledges user feedback about their vast data and the desire for it to be leveraged more effectively. While Gemini's integration with Google services is progressing, the focus is shifting towards a single, unified command interface rather than distinct modes. He hints at upcoming developments in 'personal AI' that aim to be proactive and powerful for users, emphasizing that Google is listening to this feedback.
Underappreciated principles and overrated metrics in AI
Woodward believes the AI field under-discusses principles and values, arguing that products embody a company's ethos. He feels the industry overemphasizes model performance metrics, like Elo ratings, which are often detached from real-world user value. The focus, he contends, should be on building products and features that genuinely help people, rather than solely on speed and abstract capabilities. He also notes that while AI can assist in product management, the core human element of understanding user needs and visions remains critical.
Adapting to a fast-changing tech landscape and work rhythms
Woodward discusses the challenge of building products when the underlying technology is constantly shifting. His advice, echoing others, is to have dedicated 'frontier' teams to explore new tech and report back, protecting the core teams. He emphasizes the importance of establishing work rhythms, such as 'seasons' of intense development followed by periods of exploration. Naming these phases, like 'explosive endurance,' helps manage expectations and team energy. This approach acknowledges the dual nature of modern work: marathon efforts combined with sprint-like bursts of activity.
Structuring for innovation: Autonomy, purpose, and challenging the status quo
For establishing an effective internal lab, Woodward offers three key pieces of advice: 1. Create dedicated space for unconventional ideas, ensuring autonomy and avoiding direct integration into existing business units. 2. Foster a 'user-first, Google second' mentality, with the lab's success tied to user and company success, not just its own. Labs should aim to innovate for existing products or create new categories. 3. Utilize labs not just for product development but to challenge the company's internal operations and processes, creating 'good problems' that drive systemic improvements, such as advocating for new career ladders for builders. This multi-dimensional view of labs is crucial for long-term impact.
Hiring for passion and adaptability in labs
When hiring for labs, Woodward prioritizes individuals who are intrinsically driven builders with intellectual curiosity and a passion for solving user problems, rather than focusing on personal glory. He looks for candidates who energize those around them and possess a high 'unlearning rate.' The ability to demonstrate a project built from scratch or handle ambiguous, unstructured problems is a strong indicator. He also notes that the profile of successful individuals can shift slightly between different stages of lab product development (zero-to-one vs. one-to-hundred).
Organizational pitfalls and the dangers of rapid scaling
Common organizational mistakes in labs include rushing to consolidate teams or over-organize, which can stifle emergent ideas. Woodward advises embracing a degree of managed chaos. A more critical error is hiring too many people too quickly; while seemingly impressive, rapid scaling without clear product-market fit often leads to inefficiency and failure, as exemplified by a past project with 30 engineers that lacked market alignment.
The future of Google products: Fragmentation vs. consolidation
Woodward remains undecided on whether Google will offer 10,000 niche products or a few consolidated ones in five years. He suggests the very concept of 'products' might evolve, with personalized experiences potentially driven by models and facilitated by specialized lab teams building agents. He acknowledges the success of consolidated apps in certain contexts but notes their complexity can hinder user behavior adoption. This uncertainty reflects the dynamic nature of AI's impact on how we interact with technology.
Unique rewards and fostering a culture of recognition
Google Labs employs creative, quirky awards to recognize specific contributions. Examples include the 'TPU Harvester' (for optimizing compute resources), a 'Golden Band-Aid' (for fixing minor bugs), and the 'LLM Whisperer' (for eliciting surprising insights from models), often accompanied by whimsical prizes like a small wind chime or large ears. These awards aim to codify value in unconventional ways, celebrating efficiency, meticulousness, and breakthrough innovation. They also host a 'secret society' meeting where leaders express appreciation for team members, fostering a positive and supportive culture.
Leveraging AI for feedback and collaborative insights
Woodward sees a significant opportunity in applying AI to product feedback loops, enhancing how companies gather and interpret user insights. He questions whether AI can augment human product managers by creating a collaborative environment where insights and ideas flow more naturally, potentially replicating the dynamic seen in conversations with inspiring individuals.
Mentioned in This Episode
●Products
●Software & Apps
●Companies
●Organizations
●People Referenced
AI Model Performance vs. Real-World Value
Data extracted from this episode
| Metric | Focus in AI Industry | Real-World User Impact |
|---|---|---|
| Model Performance Metrics (e.g., Elo ratings, speed, capabilities) | Heavily emphasized, often leading to industry-wide focus | Often a poor indicator of actual user value or product success |
| User-Centric Value Proposition | Less emphasized, but crucial for product success | Determines if a product solves a real problem and is desirable |
Common Questions
Because the pace of technological change, especially in AI, is so rapid. Companies need to be adept at experimenting with new models, identifying emerging opportunities before competitors, and fostering an environment where innovation can thrive.
Topics
Mentioned in this video
The research and development division at Google, focused on experimenting with and scaling new AI products and technologies.
Mentioned as a company using WorkOS, showcasing the platform's role in enabling enterprise-ready features for SaaS products.
A product developed by Google, discussed in the context of its large user base and its integration with other Google services.
A Google product that is an example of an idea originating from Google Labs and growing into a more substantial project.
A Google product that allows for blending multiple images and animating them, showcasing creative control previously unavailable.
A platform that helps engineering leaders measure and evaluate the impact of AI adoption across the product lifecycle, from developer productivity to cost management.
A software company that uses WorkOS, highlighting the platform's utility for enterprise-focused SaaS products.
Mentioned as a company that successfully uses WorkOS, underscoring the platform's value for B2B SaaS development.
A component of Gemini that allows users to connect various data sources like Gmail and Google Calendar, showcasing the potential of personal AI agents.
A platform that provides essential enterprise features like SSO, SCIM, and RBAC through APIs, enabling B2B SaaS companies to build faster and scale.
Mentioned as a successful company that utilizes WorkOS, indicating its presence in the AI development space.
A project mentioned as one of the examples of Google products that had successful moments, alongside others like Notebook LM and Flow.
Mentioned as one of the successful companies that use WorkOS, highlighting its role in the AI landscape.
One of the successful companies that leverage WorkOS, indicating the platform's broad applicability in the B2B SaaS market.
Used as an analogy for WorkOS, comparing WorkOS to Stripe but for enterprise features, indicating its significance in its market.
A company that uses DX to measure the impact of AI on developer productivity, highlighting DX's value in assessing AI adoption.
A former guest on the podcast and colleague who proposed a question about Google's future product landscape, indicating a collegial relationship and shared interests.
A speaker at the 'Lenny's Friends Summit' who presented insights on building products in a rapidly changing technological landscape, advocating for dedicated 'frontier' teams.
A tennis player whose concept of 'explosive endurance' is used as a metaphor for sustained high-performance work and self-regulation.
More from Lenny's Podcast
View all 58 summaries
38 minOpenAI’s Head of ChatGPT: We’re entering a new era of AI (again) | Tibo Sottiaux
29 minWhy Claude can’t be your PM (yet) | Anthropic CPO Panel
30 minStop planning for 2027: how OpenAI builds product 90 days at a time | Tara Sesha and Nan Yu (OpenAI)
22 minRoles aren't converging—they're expanding | Tamar Yehoshua (Atlassian CPO)
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free