Why every company now needs to think and operate like a lab team | Josh Woodward (Google Labs, Gemini, AI Studio)

Episode Summary

Executive Summary: Josh Woodward, who leads Google Labs, Gemini, and AI Studio, argues that fast-moving AI makes experimentation essential for companies of every size. Successful labs need frontier-focused people, room for unconventional ideas, and connections to the broader business so experiments can become valuable products. He shares practical signals for assessing ideas and product-market fit, skills that matter in AI-era teams, and ways to sustain a fast pace without burning people out.

Main Topics: Building a frontier-oriented labs team: Companies should have people dedicated to trying new models and technologies, while giving them an environment to build and a path to connect discoveries with the larger organization. Labs need enough independence to explore, but should not become isolated research groups. Where ideas come from and how to pursue them: Strong ideas rarely emerge from scheduled design sprints. They often grow from small groups’ side projects, curiosity, and hands-on experimentation. Teams should track what is “almost possible,” form views about the future, and pair technical breakthroughs with real user problems. Recognizing product-market fit and stopping weak ideas: Early product signals can be qualitative: whether people lean in and their eyes light up when shown a prototype. Teams should fall in love with the problem rather than a specific solution, expect pivots, and create a culture where project teams can candidly recommend stopping or changing direction. Skills, collaboration, and team composition: Woodward values rapid unlearning, intellectual curiosity, sustainable intensity, and the ability to build trust and collaborate with both people and AI agents. Roles are becoming more flexible, but functional specialties still matter; builders should use new tools to strengthen, not abandon, their core expertise. Planning and sustaining an experimental pace: Labs plan on rolling horizons and use short milestones to move ideas forward. Leaders can name intense work periods and deliberately make room afterward for exploration and recovery; those less structured periods may seed the next wave of products. AI product opportunities and industry priorities: Woodward sees consumer opportunities in entertainment, messaging, and problems involving time, money, and shared experiences. He argues that product principles and the future companies want to create deserve more discussion, while model benchmarks receive too much attention relative to user value.

Key Arguments: Rapidly changing AI capabilities make experimentation a strategic necessity: companies need to discover what new models enable before competitors or startups do. A labs team needs both frontier access and organizational bridges. Woodward warns that corporate labs often lose effectiveness after a few years or fail to commercialize ideas when they are too detached from the business. Good ideas are cultivated through continual exposure to new capabilities and user behavior, not reliably generated in scheduled brainstorming sessions. A list of “almost possible” experiences can help teams recognize when a technical shift creates a new opportunity. Early product-market-fit signals are often human and qualitative. Prototype reactions can reveal whether a product resonates before conventional metrics are meaningful, while durable success requires addressing a real problem people value. Teams should be willing to kill or pivot ideas when effort and passion are fading and user response is weak. Leaders should make it safe for the people closest to the work to say so. AI lowers the cost of execution and may enable smaller, more fluid teams, but it does not make every specialty interchangeable. Strong teams combine distinct expertise with collaboration across functions and agents. Planning should accommodate rapid change: teams can maintain a roughly six-month horizon while using shorter milestones and making room for experimentation after intense delivery periods. AI companies should prioritize the products and futures they want to create over benchmark competition alone; capability measures matter chiefly when translated into meaningful user value.

Data Points: Woodward's tenure at Google: Over 16 years - The host’s introduction describes his experience at Google. Gemini's user scale: Over a billion users - Woodward and the host refer to Gemini as a product operating at global scale. Typical lifespan of an ineffective corporate lab: About 3 to 4 years - Woodward says large-company labs often fizzle out or struggle to commercialize their work after this period. Future predictions tracked by the team: 82 predictions - Woodward says the team maintains predictions about the future, while acknowledging most may be wrong. Expected accuracy of predictions: Nine out of ten, or perhaps 9.9 out of 10, wrong - Woodward emphasizes that predictions are a tool for developing a point of view, not claims of certainty. Typical time to a meaningful milestone: About 50 to 100 days - Woodward says the strongest labs teams can move from an idea to a meaningful milestone within this window. Labs planning horizon: About six months - The team uses a rolling planning horizon because model capabilities and opportunities change quickly. Typical size of a labs project team: Previously about 5 to 7 people; now about 2 to 4 - Woodward describes smaller teams collaborating across increasingly flexible roles. Experiments running concurrently: About 20 to 30 - Woodward describes the number of simultaneous experiments across labs at a given point. Lab project stages: Zero to one, one to ten, and ten to one hundred - The team uses these stages to distinguish early discovery from growth and scaling work.

Pivotal Quotes: "When you're showing people early prototypes, what I'm always doing in those is you're looking at people's eyes and like, that is the metric." — Josh Woodward: He describes a qualitative signal for whether an early prototype resonates with users. "The team usually knows before the leader knows." — Josh Woodward: He explains why leaders should make it safe for project teams to raise concerns and recommend ending an idea. "I think the devil's in the details, all those things you just said." — Josh Woodward: He agrees that all companies need to adopt some lab-like methods, while emphasizing the challenge is designing the right environment and culture.

Implications: Companies should create small, empowered teams that continuously test new capabilities, connect experiments to real user needs, and can candidly pivot or stop. The advantage will come not just from accessing AI, but from judgment, collaboration, and turning capability into products people value.