Netflix's product org is divided into swim lanes (streaming, DVD, personalization, advertising, original content, etc.), each run as a small pod: roughly one product leader, 2-3 designers, and 5-50 engineers. Each pod owns exactly one proxy metric and decides its own tactics for moving it.
Reviewed on a Quarterly Product Strategy Review cadence; each pod's metric maps onto the "M" in SMT Model (Strategy, Metric, Tactics) and typically supports one leg of GEM Model (Growth, Engagement, Monetization) or DHM Model (Delight, Hard-to-Copy, Margin-Enhancing). Complements Organizational Duality: Scale Structure vs. Innovation Structure and Highly Aligned, Loosely Coupled.
Adam Mosseri describes Instagram replacing its canonical big-company product team — roughly a 'baker's dozen' of specialists (PM, several engineers, designer, data scientist, researcher, etc.) — with small pods of ~6-7 people. The team is built around a generalist 'product staff' role: an evolution of the PM that uses AI tools to absorb work previously routed to dedicated specialists — running funnel/waterfall analyses, mocking up design proposals, shaping research questions — directly. Senior specialists (e.g., a pricing data scientist) are added only when a pod hits a need generalist tooling can't cover. Mosseri frames this as evidence of a broader 'age of the generalist' (see Systems Thinkers Over Specialists) enabled by AI collapsing the cost of design, analysis, and research work that used to require dedicated headcount.
Adam Mosseri describes Instagram/Meta's old 'canonical' product team as roughly a baker's dozen: 2-3 Android engineers, 2-3 iOS engineers, 2-3 server engineers, plus a generalist PM, designer, data scientist, and researcher. The newer 'pod' model often shrinks this to just 4 engineers plus a single 'product staff' generalist — with no dedicated designer, data scientist, researcher, or content designer, those functions absorbed by AI tooling and the generalist role. See waterfall-analysis for an example of data-science work a generalist can now do directly.