Rather than relying on formal knowledge sharing, Alphabet's X embeds central teams (like legal, finance, prototyping) that float between projects. These individuals become natural vectors, carrying insights, best practices, and innovative ideas from one project to another, fostering organic knowledge transfer.

To balance AI hype with reality, leaders should create two distinct teams. One focuses on generating measurable ROI this quarter using current AI capabilities. A separate "tiger team" incubates high-risk, experimental projects that operate at startup speed to prevent long-term disruption.

We gain 20 IQ points advising others but lose 20 advising ourselves. 'Deep sparring'—collaborative problem-solving with trusted peers—leverages this effect. A few hours of this per quarter provides outside perspective that can break through personal biases more effectively than weeks of isolated work.

While traditionally creating cultural friction, separate innovation teams are now more viable thanks to AI. The ability to go from idea to prototype extremely fast and leanly allows a small team to explore the "next frontier" without derailing the core product org, provided clear handoff rules exist.

To innovate quickly without being bogged down by technical debt, portfolio companies should ring-fence new AI development. By outsourcing it and treating it as a separate "skunk works" project, the core tech team can focus on existing systems while the new initiative succeeds or fails on its own merits.

Afeyan advises against making breakthrough innovation everyone's responsibility, as it's unsustainable and disruptive to daily jobs. Instead, companies should create a separate group with different motivations, composition, and rewards, focused solely on discontinuous leaps.

To ensure rigorous vetting of ideas, create an environment of friendly competition between teams. This structure naturally motivates each group to find flaws in the other's thinking, a process that might be socially awkward in a purely collaborative setting. The result is a more robust, error-checked outcome.

Stripe's Experimental Projects Team discovered that embedding its members directly within existing product and infrastructure teams leads to higher success rates. These "embedded projects" are more likely to reach escape velocity and be successfully adopted by the business, contrasting with the common model of an isolated R&D or innovation lab.

To ensure effectiveness, new government tech talent shouldn't be scattered individually across agencies. Instead, they must be deployed as self-contained teams focused on specific projects. This strategy prevents them from being absorbed and neutralized by existing bureaucracy, allowing them to maintain momentum and achieve their objectives.