At Perplexity, the design system lives in the codebase, not Figma. Designers contribute directly to the frontend, creating a single source of truth that eliminates drift between design files and production code, forcing a highly practical and collaborative process.

To create a cohesive product across multiple teams, GitHub uses a framework that forces alignment upfront. By ensuring all teams first deeply understand the problem and collectively identify solutions, the final execution is naturally integrated, preventing a disjointed experience that mirrors the org structure.

AI's productivity gains mean that on a lean, early-stage team, there is little room for purely specialized roles. According to founder Drew Wilson, every team member, including designers, must be able to contribute directly to the codebase. The traditional "design artifact" workflow is too slow.

In an AI-driven world, product teams should operate like a busy shipyard: seemingly chaotic but underpinned by high skill and careful communication. This cross-functional pod (PM, Eng, Design, Research, Data, Marketing) collaborates constantly, breaking down traditional processes like standups.

The traditional, linear handoff from product spec to design to code is collapsing. Roles and stages are blurring, with interactive prototypes replacing static documents and the design file itself becoming the central place for the entire team to align and collaborate.

The true power of UX research is aligning the entire product team with a common understanding of the user. This shared language prevents working at cross-purposes and building a disjointed product that users can feel.

Technical tools are secondary to building a successful design system. The primary barrier is a lack of shared vision. Success requires designers to think about engineering constraints and engineers to understand UX intent, creating an empathetic, symbiotic relationship that underpins the entire system.

AI tools are collapsing the traditional moats around design, engineering, and product. As PMs and engineers gain design capabilities, designers must reciprocate by learning to code and, more importantly, taking on strategic business responsibilities to maintain their value and influence.

The traditional "assembly line" model of product development (PM -> Design -> Eng) fails with AI. Instead, teams must operate like a "jazz band," where roles are fluid, members "riff" off each other's work, and territorialism is a failure mode. PMs might code and designers might write specs.