The Lab-First Mandate: Why Every Company Needs to Operate Like a Startup

In this conversation, Josh Woodward, head of Google Labs and Gemini, outlines a shift in organizational strategy: the lab-first model. As AI capabilities evolve quickly, the traditional barrier between experimental R&D and core product development is disappearing. Woodward argues that companies failing to integrate frontier testing into their DNA will be disrupted by smaller, more agile teams. The consequence of this shift is that product development is no longer a linear planning process but a high-frequency feedback loop. For leaders, this requires a re-evaluation of how they allocate talent and define success. Those who master the art of explosive endurance, the ability to sprint while pacing for the long term, will create a competitive advantage, while those who rely on legacy planning cycles will find themselves obsolete.

The Illusion of the Design Sprint

Conventional wisdom suggests that innovation is a scheduled event, like a design sprint or a brainstorming session. Woodward’s experience running Google’s largest labs team reveals the opposite: the best ideas are rarely manufactured in conference rooms. They emerge from the frontier, where engineers and product managers are constantly playing with the latest models, often during off-hours or through side projects.

> They very rarely at least in my experience have come from design sprints or times where you set aside on a calendar, oh wait this is we are going to go come up with our next idea. They usually come when people are swimming in the hallway maybe on a week and often over a break.
>
> -- Josh Woodward

The implication here is that innovation is a byproduct of exposure, not process. By forcing teams into rigid, calendar-driven ideation, companies filter out the organic, curiosity-driven breakthroughs that occur when people are simply living in the future. The competitive advantage goes to the team that has the most almost possible ideas in their pipeline, ready to be triggered the moment the underlying technology matures.

Why Your Fast Solution is a Downstream Trap

Woodward notes that the most effective teams maintain a list of things that are almost possible, technologies that are just on the edge of viability. The trap most organizations fall into is trying to force these ideas into immediate, polished products.

This creates a systemic failure: teams fall in love with the solution rather than the problem. When the solution hits a wall, the team lacks the flexibility to pivot. Woodward suggests that product-market fit is not a data-driven science in the early stages; it is an intuitive signal. If you are not looking at the user's reaction, specifically if their eyes light up, you are likely chasing the wrong metric.

> I think when people fall in love with an actual product solution first, it is gonna at least to my experience lead to a bad time.
>
> -- Josh Woodward

The effect of premature commitment is a bloated, ineffective lab. If a lab is nested under an existing business unit, it often becomes a gimmick factory that fails to commercialize. True labs require independence because they are not just building products; they are often creating good trouble by challenging the company internal processes, such as hiring ladders or cross-functional collaboration norms.

The Rise of the Unlearning Rate

As the execution cost of building software trends toward zero, the value of traditional expertise is shifting. Woodward identifies a critical skill: the unlearning rate. In a system where the underlying tech changes quarterly, the ability to discard a previously successful approach is more valuable than the ability to master a single tool.

This creates a feedback loop where the most successful builders are those who can jam across functions. The traditional PM-Designer-Engineer silos are blurring. Woodward predicts that the most valuable teams will be small, two to three people, who act as super-talented musicians capable of switching roles depending on the product needs. This requires a high degree of trust. In the AI era, trust is the primary unit of scale. If your team cannot build trust rapidly, they cannot iterate at the speed the frontier demands.

Key Action Items

• Audit your ideation process: Stop relying on scheduled brainstorms. Instead, create a frontier team whose sole job is to subscribe to the latest tools, play with them, and maintain a list of almost possible features. (Immediate)
• Implement the 100-day milestone: Move away from multi-year roadmaps. Force teams to take an idea from zero to a meaningful milestone in 50 to 100 days. This creates the necessary pressure to kill failing ideas early. (Over the next quarter)
• Normalize unlearning: Explicitly reward team members who identify when a project is failing and pivot. Celebrate the killing of a feature as a win for efficiency, not a loss for the team. (Ongoing)
• Adopt seasonal intensity: Do not expect explosive endurance 365 days a year. Explicitly define sprint seasons, such as around a major model release, and hack seasons, such as post-launch, to prevent burnout. (12-18 months)
• Build a Good Trouble culture: Use your lab team to pilot new internal processes, like new job ladders or cross-functional communication styles, before rolling them out to the rest of the company. (12-18 months)