Table of Contents
The traditional design process, long treated as gospel by the industry, is undergoing a radical collapse. As AI accelerates engineering cycles to breakneck speeds, the luxury of lengthy discovery phases and high-fidelity mockups is disappearing. Jenny Wen, Head of Design for Claude at Anthropic and former Director of Design at Figma, argues that designers can no longer afford to be the bottlenecks of production. Instead of clinging to rigid methodologies, the next generation of design leaders is embracing a more fluid, execution-heavy approach that prioritizes speed and direct collaboration with engineering agents.
Key Takeaways
- The "Double Diamond" is dead: Traditional research and convergence cycles are too slow for the AI era; designers must move from being gatekeepers to facilitators of execution.
- Shift in time allocation: Prototyping and mocking have dropped from 70% of a designer's day to roughly 30%, replaced by "jamming" with engineers and direct implementation in code.
- Vision is shrinking: Five-year design decks are being replaced by 3-to-6-month directional prototypes that provide a North Star for rapidly iterating teams.
- Hiring for "Blocks" and "T-Shapes": The industry is shifting toward strong generalists who can flex into PM and engineering roles, and deep specialists who offer high-craft differentiation.
- Trust through speed: Quality is maintained not through perfectionism, but through rapid iteration and showing users that their feedback is heard and implemented in real-time.
The Collapse of the Traditional Design Process
For decades, designers have been taught a stable, linear process: research, discovery, divergence, and convergence. This methodology provided a sense of security and professional rigor. However, Wen suggests that this framework has become a liability in an environment where engineers can spin up multiple AI agents to build features in hours rather than weeks.
The rise of non-deterministic AI models means that traditional mocks are often obsolete before they are finished. You cannot easily mock up every possible state of an AI interaction, nor can you theorize how a user will prompt a model. Real learning happens through usage, not speculation.
This design process that designers have been taught, we sort of treat it as gospel. That's basically dead.
This shift doesn't mean design is less important; it means the timing of design has changed. Designers are moving away from being the first step in a waterfall and toward becoming "last-mile" specialists who provide polish, taste, and cohesion to live code.
The New Designer-Engineer Relationship: Letting Them Cook
One of the most significant changes at frontier companies like Anthropic is the blurring of lines between roles. When engineering velocity increases exponentially, designers must learn to "let the engineers cook." Rather than blocking a release to ensure every pixel matches a Figma file, designers are acting as consultants who help engineers stay on track while they ship.
From Mocks to Pairing
Previously, a designer might spend 60% to 70% of their time in prototyping tools. Today, that number has plummeted. The new pie chart of design work is split between high-level direction, pair-programming with engineers, and "last mile" implementation. This requires designers to develop a technical vocabulary that allows them to nudge code directly rather than throwing static assets over a wall.
Building Trust Through Speed
There is a persistent fear that moving faster degrades craft. Wen argues the opposite: trust is built through the responsiveness of the team. When a user reports a flaw and sees a fix live the next day, it creates a bond of accountability that a "perfect" but slow product cannot match. This "research preview" mindset allows teams to ship products that are good enough to provide value while iterating toward excellence in public.
You as a designer actually like do not have the time to make these beautiful mocks anymore.
The Evolution of Design Vision
Long-term "North Star" visions used to be 5-to-10-year outlooks presented in polished decks. In the AI space, the technology changes so fast that a two-year vision is effectively science fiction. Vision work has shifted to a much tighter 3-to-6-month horizon.
The goal of modern design vision is not to predict the distant future, but to point the team in a direction that makes their current, frantic execution more efficient. If seven different engineering agents are spinning up seven different features, the designer’s job is to ensure they are all moving toward a unified cause. This is often achieved through functional prototypes rather than static presentations.
The AI-Native Design Stack
While Figma remains a core tool for exploring micro-interactions and visual styles, the rest of the designer's toolkit is becoming increasingly technical. Designers are now spending significant time in IDEs like VS Code, using AI tools like Claude Code to tweak CSS, adjust layout classes, and implement polish directly into the repository.
Figma’s Persistent Role
Despite the move toward code, Figma remains essential for non-linear exploration. Coding is inherently linear; you invest in one direction and iterate. Figma allows a designer to throw ten different ideas against a canvas simultaneously to curate the best path forward. It is the playground for "micro-directions"—testing typography, styles, and layout before committing to the structure of code.
The Chat vs. UI Debate
There is frequent debate over whether chat interfaces are merely a temporary stopgap. While interactive widgets and tactile UIs are being integrated into AI platforms, the "chatbot" paradigm offers a level of infinite flexibility that traditional baked-in UIs cannot match. The future likely involves UIs that are generated on-the-fly by models based on the user's specific intent, rather than static interfaces hand-coded by humans.
Hiring the Future: Three New Archetypes
As the role changes, the criteria for "hirable" designers are also evolving. Wen identifies three specific archetypes that are becoming increasingly valuable in high-velocity AI environments.
1. Strong Generalists (The "Block" Shape)
Unlike the traditional T-shaped designer who has one deep skill and a broad surface, Wen looks for "block-shaped" designers. these are individuals who are in the 80th percentile of design, PM, and engineering skills. They can flex into whichever role the team needs most at any given moment.
2. Deep Specialists
Because AI can now handle the "average" design tasks, there is a premium on the top 10% of specialists. These are the designers with world-class visual taste, icon designers, or highly technical motion designers who provide the differentiation that AI cannot yet replicate.
3. The Craft New Grad
There is a unique advantage to hiring designers who are "unburdened" by the legacy processes of the last decade. These "craft new grads" are quick learners who treat building as a hobby, often participating in "build-and-show" communities like Socratica. They lack the "baked-in" rituals that can slow down senior talent in a fast-changing environment.
Someone has to decide what is actually going to get built and what actually matters.
The Human Element: Accountability and Taste
As AI agents begin to review their own code and suggest their own product ideas, the question remains: where do humans stay valuable? The answer lies in accountability and judgment. While AI can provide the data and even suggest the "best" path, a human must still sign off on the decision and take responsibility for the outcome.
Designers are moving from being "makers" to "curators" and "deciders." They are the ones who must bridge the gap between what the technology can do and what is actually meaningful for the user. By letting go of the old "gospel" of the design process, they gain the freedom to focus on the only thing that ultimately matters: shipping products that solve real problems in record time.
){kind=link}