Lessons from Google Labs leader Josh Woodward on how companies can experiment with AI, find strong ideas, and turn prototypes into products.
The main idea
In an era of rapidly advancing AI, every company needs to operate like an experimental laboratory. Old approaches no longer work.
Where the best ideas come from
- Almost never from design sprints or brainstorming sessions.
- They emerge from small, obsessed groups of people—in the shower, over a weekend, or during a hallway conversation.
- You cannot microwave a good idea.
- A leader’s job is to create an environment where such ideas can emerge and to recognize them at the right moment.
The “Almost Possible” framework
- Maintain a living list of things that are not quite possible yet.
- As soon as technology crosses the required threshold, launch an experiment.
- This is critical in a world where the frontier moves every three to four months.
How to recognize early product-market fit
- At the zero-to-one stage, watch people rather than dashboards.
- The strongest signal is physical: their eyes widen, they lean toward the screen, and they cannot look away.
- Early on, emotion matters more than metrics.
How to relate to ideas
- Fall in love with the problem, not the solution.
- Most ideas need three to five significant pivots.
- Knowing when to kill a project is one of the essential skills of a Labs team.
Skills that have sharply increased in value
-
Unlearning rate—how quickly you can abandon outdated knowledge.
-
Burst endurance: the ability to work in short, intense sprints.
-
The ability to build trust.
-
A conventional growth mindset is no longer enough.
How to structure a Labs team
- A zero-to-one team should have two to four people; this work used to require five to seven.
- Structural independence is essential.
- Ideally, the team reports directly to the CEO.
- Without independence, a lab almost always fades within three to four years.
The hardest part: transferring work into the company
- The most difficult point is the handoff from Labs to the core business.
- A team can build excellent prototypes and still lose everything during the transfer.
- Historically, almost all corporate labs either fade after three to four years or fail to commercialize their results.
Why the transition fails
- The core business sees new products as “toys.”
- Labs and the core organization have different metrics and cultures.
- A lab placed inside a business unit is eventually constrained by that unit’s current priorities.
- Successful products crowd out new experiments.
What to do
- Define the Labs mandate in advance: improve existing products or create new categories.
- Employ people who can translate between the two worlds.
- Make the handoff deliberate, changing the team, metrics, and processes as needed.
Conclusion
The winners will be companies that learn to experiment quickly at the frontier, honestly kill weak ideas, and successfully transfer strong ones into the core organization. Everyone else will be catching up.