Why every company now needs to think and operate like a lab team | Josh Woodward (Google Labs, Gemini, AI Studio)
Josh Woodward, head of Google Labs, Gemini app, and AI Studio, discusses how every company must now operate like a labs team to adapt to rapidly changing AI tech. He shares insights on generating ideas, identifying product-market fit, knowing when to quit, and the evolving skills and organizational structures needed for success in the AI era.
Deep Dive Analysis
20 Topic Outline
Why Every Company Needs a Labs Team
Where Great Ideas Truly Originate
The 'Almost Possible' Framework for Innovation
Identifying Genuine Product-Market Fit
Falling in Love with the Problem, Not the Solution
Signs It's Time to Kill an Idea
Overview of Google Labs and the Gemini App
Hypotheses for Next Consumer AI Breakthroughs
Google's Vision for a Personal AI Assistant
Overhyped and Underhyped Aspects of AI
Skills Rising in Value in the AI Era
Why Roles Aren't Blurring into One Universal Builder
Managing Work Intensity with Explosive Endurance Rhythms
Evolution of Planning in the AI Age
Labs Lifecycle Stages and Challenges
Tips for Setting Up an Internal Labs Team
Qualities of Ideal Labs Team Members
Common Mistakes in Labs Team Organizational Structure
The Future of Products: 10,000 or 5?
Improving Product Feedback Loops with AI
5 Key Concepts
Unlearning Rate
This refers to an individual's capacity to quickly learn new information or skills and then, just as rapidly, discard or walk away from previously held beliefs or methods based on new insights or data.
Explosive Endurance
Inspired by Roger Federer, this concept describes the ability to engage in periods of intense, high-effort work ('explosive') while simultaneously pacing oneself and maintaining well-being to sustain that effort over a long duration ('endurance').
Almost Possible Framework
A strategic approach where a team actively tracks and lists technical capabilities or user experiences that are not quite feasible yet, then quickly mobilizes to build when one of these items transitions into being technically possible.
Users First, Google Second, Product Third
A prioritization mantra used within Google Labs and Gemini, emphasizing that the ultimate goal is user success, followed by the broader company's success, with the specific product or team's success coming last.
Good Trouble
This refers to the beneficial disruption a labs team can create by not only developing new products but also by challenging and innovating internal company processes, job roles, and established ways of working.
9 Questions Answered
Great ideas rarely emerge from conventional design sprints or scheduled brainstorms; they often arise from individual curiosity, side projects, and constant alertness to new possibilities, especially when people are not actively trying to 'microwave' them.
Early signs of product-market fit are often qualitative, observed by watching users' eyes light up, seeing them lean in with interest, and identifying a 'nerve' hit with a pain point that makes them say, 'I want that.'
An idea should be killed when the team's passion starts to run out, when all conceivable avenues have been tried without success, and when the team itself, rather than just the leader, recognizes that it's simply not working.
Underhyped in AI is the discussion around principles and values—what kind of future we want to build. Overhyped are model benchmarks, as most users don't care about ELO ratings but rather how a product delivers real value.
Skills trending up in value include a high 'unlearning rate' (ability to quickly abandon old ideas), 'explosive endurance' (intense work balanced with pacing), and the ability to build and scale trust for effective collaboration with both people and agents.
Planning has shifted from multi-year roadmaps to rolling six-month windows, with a focus on achieving meaningful product milestones within 50 to 100 days, acknowledging that new possibilities emerge too quickly for long-term rigid plans.
Key tips include creating an independent space for 'weird things' to grow, clearly defining the lab's goal (e.g., graduating products vs. inventing new categories), and using the lab to challenge and improve company processes ('good trouble').
Ideal labs team members are intellectually curious, constantly building, obsessed with solving user problems (not their own ego), and energized by unstructured, unknown challenges where they need to figure out the way forward.
A common mistake is rushing to consolidate teams too quickly, rather than allowing for messiness and emergence, and hiring too many people too fast, which can lead to vanity metrics of growth without actual product-market fit.
18 Actionable Insights
1. Operate Like a Labs Team
Every company needs to adopt a ’labs team’ mindset, constantly experimenting with the latest AI models and technologies to discover new possibilities before competitors do.
2. Cultivate Idea Generation Spaces
Create an environment where ‘weird things can grow’ and ideas emerge organically from individual curiosity and side projects, rather than relying solely on structured design sprints or brainstorms.
3. Track ‘Almost Possible’ Ideas
Maintain a list of technical capabilities or experiences that are ‘almost possible’ and act quickly to explore them when they become feasible, recognizing these as phase shifts.
4. Fall in Love with the Problem
Focus on deeply understanding and solving user problems, rather than becoming attached to a specific product solution, as most successful ideas require multiple pivots.
5. Observe User Eye Reactions for PMF
When testing early prototypes, look for non-verbal cues like users’ eyes lighting up or leaning in, as these are critical early signals of genuine product-market fit.
6. Empower Teams to Kill Ideas
Foster a culture where team members feel safe and empowered to honestly communicate when a project isn’t working, as they often know before leadership does.
7. Prioritize ‘Unlearning Rate’
Seek out and cultivate individuals who can quickly learn new information and then readily abandon previous assumptions or approaches when new data suggests a different path.
8. Develop ‘Explosive Endurance’
Encourage a work rhythm that allows for periods of intense, ’explosive’ effort, balanced with intentional recovery and pacing to ensure long-term sustainability and prevent burnout.
9. Name and Declare Work Modes
As a leader, clearly communicate the current operational mode (e.g., ’explosive time’ for sprints, ‘hack time’ for exploration) to set expectations and manage team energy effectively.
10. Plan with Rolling Windows and Milestones
Adopt a flexible planning approach with rolling six-month roadmaps and focus on achieving meaningful milestones within 50 to 100 days, acknowledging rapid technological change.
11. Set Up an Independent Labs Team
For larger companies, establish a labs team with sufficient independence, often reporting at a distance from existing business units, to allow for experimentation with ‘weird things’.
12. Define Lab’s Success Metrics Clearly
Be explicit about the lab’s primary goal: whether it’s to graduate innovations into existing products or to invent entirely new product lines and categories.
13. Use Labs for ‘Good Trouble’
Leverage labs teams not just for product development but also to challenge and innovate internal company processes, job roles, and ways of working.
14. Hire Obsessive Builders
Recruit individuals for labs who are intellectually curious, constantly building, obsessed with solving user problems (not ego), and add positive energy to the team.
15. Avoid Rapid Team Expansion
Resist the temptation to quickly double or triple team size, as vanity metrics of growth can distract from the crucial focus on achieving product-market fit.
16. Actively Seek Product Feedback
Continuously seek and aggregate feedback on products and experiences, exploring how AI can enhance the collection and insight generation from these feedback loops.
17. Reward Efficiency and Innovation
Implement unique recognition programs (e.g., ‘TPU Harvester’ for efficiency, ‘Golden Band-Aid’ for bug fixes, ‘LLM Whisperer’ for model breakthroughs) to celebrate specific valuable contributions.
18. Give Direct Leadership Appreciation
Surprise valuable team members with ‘secret society’ meetings where leaders take turns expressing genuine appreciation for their contributions, fostering a strong sense of value.
9 Key Quotes
You can't microwave good ideas.
Josh Woodward
What's their unlearning rate? How fast can they learn something and then walk away from it?
Josh Woodward
When you're showing people early prototypes, you're looking at people's eyes. That is the metric.
Josh Woodward
The people who get excited about things, usually we try to kind of think about what's the problem we're falling in love with, not the product.
Josh Woodward
The team usually knows before the leader knows.
Josh Woodward
I still feel like sometimes as an industry, it can kind of over over rotate to that and kind of lose sight of like, why is this useful to someone?
Josh Woodward
Never waste an event or like a date.
Josh Woodward
You've got to try to create a space where weird things can grow.
Josh Woodward
Labs sometimes creates good trouble at Google.
Josh Woodward
1 Protocols
Google Labs Recognition Awards
Josh Woodward- TPU Harvester: Awarded to individuals who optimize and reclaim TPUs for efficiency. The award is a miniature rake and a cash prize.
- Golden Band-Aid: Given to team members who fix 'paper cuts' or small, annoying product issues. The award is a golden Band-Aid to display on their desk.
- LLM Whisperer: Presented to people who can push the model to its frontier in surprising ways. The award includes a giant set of ears and a nice mechanical keyboard.
- Secret Society Thank You: An unannounced meeting is scheduled on a team member's calendar with multiple leaders. Upon arrival, leaders take turns expressing genuine appreciation for their contributions to the team.