Oct 11, 2026 · Lenny's Podcast: Product | Growth | Career · 5 min read

Building Internal Labs for AI Innovation

Josh Woodward of Google Labs outlines the strategic framework for corporate innovation teams. Learn how to identify 'almost possible' technologies, measure product-market fit through user engagement, and structure labs to avoid the common failure modes of large enterprises.

Technology Business Strategy Entrepreneurship

The takeaways

Track Almost Possible Technologies. Maintain a dynamic list of capabilities that are nearly viable to identify phase shifts. When a technology crosses the threshold from theoretical to practical, immediately allocate a small team to explore its commercial potential.
Measure Fit via User Eyes. Abandon early-stage metrics like D7 retention in favor of qualitative observation. Watch for pupil dilation and leaning in during prototype demos to gauge genuine user excitement and pain point resolution.
Prioritize Unlearning Rate in Hiring. Evaluate candidates on how quickly they can discard outdated methods in favor of new AI-driven workflows. This adaptability is more critical than static technical knowledge in rapidly evolving AI markets.
Isolate Labs from Core Units. Ensure innovation teams have structural independence to prevent bureaucratic stifling. Labs must report to senior leadership with the autonomy to pursue 'weird' ideas without immediate commercial pressure.
Kill Ideas When Passion Fades. Trust team members to signal when a project is failing before leadership does. If the team's enthusiasm wanes despite exhaustive testing, pivot immediately rather than forcing a launch.
Define Clear Lab Success Metrics. Clarify whether the lab's goal is to graduate projects to existing products or invent new categories. This distinction dictates the team's structure, hiring profile, and performance evaluation criteria.

The Strategic Imperative for Corporate Labs

In an era where AI capabilities shift weekly, traditional R&D models are obsolete. Josh Woodward, Head of Google Labs, argues that every company must operate with a 'labs mindset.' This requires a dedicated frontier team whose sole purpose is to test the latest models and identify emerging opportunities before competitors do. The core challenge is not just technical exploration but organizational design: how to maintain the agility of a startup within the scale of a large enterprise.

Identifying and Validating Opportunities

Woodward emphasizes that high-impact ideas rarely emerge from structured design sprints. Instead, they arise from continuous observation of the 'frontier.' Teams must maintain a list of 'almost possible' technologies and monitor for phase shifts where a capability becomes viable. Validation in the early stages is less about data and more about human reaction. The primary metric for product-market fit is the user's eye: if they lean in and their eyes light up, the product has hit a nerve. This qualitative signal is more reliable than early retention metrics, which often lag behind genuine user delight.

Organizational Structure and Talent

A common failure mode for corporate labs is nesting them under existing business units, which leads to bureaucratic friction and eventual stagnation. Successful labs require structural independence, often reporting directly to the CEO, to allow 'weird things to grow.' Talent selection has shifted from static skills to dynamic attributes. The 'unlearning rate'—how quickly a person can discard old methods—is now a key hiring criterion. Additionally, 'explosive endurance' is vital; teams must sustain high-intensity work over long periods without burning out. Roles are blurring, but specialists remain essential. The best teams combine deep functional expertise with the ability to collaborate across disciplines and with AI agents.

Execution and Longevity

Planning horizons have compressed to six months, with meaningful milestones expected within 100 days. Leaders must create a rhythm of work, alternating between intense sprints and periods of exploratory hacking. This balance prevents burnout and fosters innovation. Ultimately, the goal of a lab is not to build a product for its own sake, but to make users successful. By focusing on user value and maintaining a culture of 'good trouble' that challenges existing processes, companies can sustain innovation long-term. The future of enterprise AI lies in these agile, independent teams that can bridge the gap between frontier research and commercial reality.

Key insights

Corporate labs fail when nested under existing business units due to bureaucratic constraints. Structural independence is required to allow experimentation and failure without immediate commercial pressure.

Organizational Design →

Impact: Prevents the stagnation of innovation teams and ensures that high-risk, high-reward AI experiments can proceed to maturity.
Early product-market fit is best identified through qualitative user reactions, specifically eye engagement and leaning in, rather than quantitative retention metrics.

Product Strategy →

Impact: Allows for faster iteration cycles by providing immediate, intuitive feedback on whether a prototype solves a genuine user pain point.
The 'unlearning rate' is a critical hiring metric in the AI era, measuring how quickly employees can discard outdated workflows in favor of new AI capabilities.

Talent Management →

Impact: Ensures the workforce remains agile and capable of leveraging the latest AI tools, preventing skill obsolescence and operational inefficiency.
Ideas rarely come from scheduled design sprints; they emerge from continuous observation of the 'frontier' and tracking 'almost possible' technologies.

Innovation Process →

Impact: Shifts R&D focus from forced brainstorming to proactive monitoring of technological phase shifts, increasing the hit rate of viable product concepts.
Teams usually know when a project is failing before leadership does, signaled by a loss of passion and enthusiasm despite exhaustive testing.

Leadership →

Impact: Empowering teams to kill failing projects early saves resources and allows for rapid pivoting to more promising opportunities.

Action items

Create a dedicated 'frontier team' with the explicit mandate to test the latest AI models and report on emerging capabilities to the rest of the organization.

Impact: Keeps the company at the technological edge without distracting core product teams from their primary objectives.
Maintain a living document of 'almost possible' technologies and assign owners to monitor for the moment these capabilities cross into viability.

Impact: Enables proactive product development by identifying market opportunities before competitors recognize them.
Revise hiring criteria to prioritize 'unlearning rate' and 'explosive endurance' over static technical skills, assessing candidates on their adaptability and sustainability.

Impact: Builds a resilient workforce capable of thriving in high-velocity AI environments and adapting to rapid technological changes.
Structure innovation labs with direct reporting lines to senior leadership, ensuring they have the autonomy to pursue unconventional ideas without bureaucratic interference.

Impact: Protects the lab from the 'fizzle out' effect common in corporate R&D, allowing for genuine experimentation and breakthroughs.
Implement a culture of 'good trouble' where labs are encouraged to challenge existing company processes and pilot new ways of working, such as builder job ladders.

Impact: Drives organizational evolution by using the lab as a testing ground for structural and cultural changes that can be scaled company-wide.

Quotes

“You can't microwave good ideas.”

“When you're showing people early prototypes, you're looking at people's eyes. That is the metric.”

“The team usually knows before the leader knows.”