Devoured - April 23, 2026
Rethinking the shape of design teams in an AI world (11 minute read)

Rethinking the shape of design teams in an AI world (11 minute read)

Design aiteams Read original

AI is forcing design teams to rethink organizational structure as individual productivity surges but traditional hierarchies create a "permission tax" more expensive than building the features themselves.

What: An analysis of how AI-powered tools are disrupting design team structures, career progression, and organizational hierarchies. The article proposes a "dual transformation" model that balances fast-moving experimental teams with stable core teams to prevent hollowing out the workforce while maintaining innovation speed.
Why it matters: While framed around design, this addresses a broader problem facing all technical organizations: when AI lets one person do a team's work in hours, traditional approval processes and rigid role divisions become the primary bottleneck rather than a safety feature, forcing a fundamental rethink of how teams operate.
Takeaway: Organizations should evaluate whether they need a dual-track structure with experimental "atom" teams for innovation and stable "tomato" teams for core work, connected by a capability link that allows proven ideas to scale without killing velocity.
Deep dive
  • The debate started when Jenny Wen (Anthropic) declared the rigid design process dead, but Nielsen Norman Group countered that expert designers have simply internalized and compressed the process rather than abandoned it (the "Master Effect")
  • The real crisis is for junior designers who lack the foundation to skip steps safely—as AI automates entry-level work, companies may bypass juniors entirely and hire only seniors, eventually depleting the senior pool when there's no pipeline to replace them
  • AI creates a "paradox of autonomy" where individuals can build in hours what once took teams weeks, but corporate structures built for risk mitigation prevent people from moving that fast
  • The "permission tax" of hierarchical approvals has become more expensive than the actual cost of building features, fundamentally flipping the economics of organizations
  • Roles are blurring into "Full-Stack Everything" hybrids where designers write backend prompts, engineers generate UI with AI, and PMs build their own dashboards—the most valuable workers are "product-trio" types who bridge all three disciplines
  • Three team structure models are emerging: Pyramid (traditional hierarchy optimized for stability), Tomato (self-contained squads with defined roles), and Atom (temporary swarms that form around tasks with completely blurred roles)
  • Pure Atom structures risk losing "good friction"—the healthy tension between product, engineering, and design perspectives that prevents building what's easiest to code rather than what's best for users
  • The proposed solution is Scott Anthony's "Dual Transformation" framework: Transformation A maintains the stable core business (Pyramid/Tomato) with intentional friction for safety, while Transformation B operates as fast Atomic swarms exploring new paradigms
  • A "Capability Link" bridges the two transformations, providing a structured pathway to scale proven innovations from the experimental teams back into the core business without killing their velocity
  • This model allows juniors to develop foundational skills in the stable Transformation A environment while staying exposed to cutting-edge practices from Transformation B teams, preserving both innovation speed and workforce development
  • The framework prevents the structural mismatch where forcing innovative teams to adopt heavy stakeholder reviews turns "good friction" into toxic bureaucracy that smothers discovery
Decoder
  • Vibecoding: Rapid prototyping approach where designers use AI to go from concept to functional code by describing the desired outcome rather than following traditional step-by-step design processes
  • Permission tax: The organizational overhead cost of hierarchical approvals and sign-offs, which in the AI era can exceed the actual cost of building features
  • Full-Stack Everything: Hybrid roles where individuals use AI to bridge traditional discipline boundaries (designers coding, engineers doing UI design, PMs doing data analysis)
  • Pyramid/Tomato/Atom models: Three team structures—Pyramid is traditional hierarchy, Tomato is the squad model with self-contained teams, Atom is temporary task-based swarms with fluid roles
  • Dual Transformation: Framework that runs two parallel organizational modes—one maintaining the stable core business (Transformation A) and one exploring new paradigms (Transformation B)
  • Capability Link: The bridge between stable and experimental teams that evaluates which innovations are mature enough to integrate into core operations
  • Good friction: Productive tension between different roles (PM/designer/engineer) that prevents groupthink and ensures user needs aren't sacrificed for implementation convenience
  • Master Effect: Phenomenon where experts have done formal processes so many times they execute them intuitively in seconds rather than explicitly following steps
Original article

The claim that Jenny Wen “broke” the design process has sparked debate, but critics like Sarah Gibbons argue the process isn't gone—experienced designers have simply internalized and compressed it, while skipping it entirely risks weakening junior talent. The bigger concern is that AI-driven speed could hollow out the workforce (fewer juniors becoming seniors) and strain traditional organizations, pushing a shift toward flatter, hybrid “full-stack” roles and new team models—balancing fast, autonomous “atom” structures for innovation with stable “tomato” systems to preserve rigor, skills development, and long-term resilience.