The design process is dead. Here’s what’s replacing it. | Jenny Wen (head of design at Claude)

Jenny Win, a leader in AI design, declares that the traditional design process, once treated as gospel, is "basically dead." The old model of extensive research, divergent thinking, and iterative convergence is no longer viable. The rapid advancement of AI-powered engineering means designers must fundamentally change their approach, as engineers can now quickly generate and ship features.

Jenny outlines two evolving categories of design work in the AI era. The first focuses on supporting implementation and execution, where designers guide engineers as they rapidly spin up features. The second involves creating a strategic vision, a crucial but now much shorter-term endeavor, typically spanning three to six months. The days of crafting elaborate, multi-year design visions are over, replaced by the need for quick, adaptable direction.

The relentless speed of engineering, empowered by AI, is the primary force driving changes in design. Engineers can now ship code at an unprecedented rate, making traditional design bottlenecks unacceptable. Designers are increasingly involved in the 'last-mile' aspects of development, implementing polish and prototyping directly in code. This means letting go of the old 'gatekeeper' mentality and embracing direct, hands-on contribution.

This fundamental shift in the design process is being felt broadly across the industry, extending beyond just AI-focused companies. Many designers acknowledge the necessity of abandoning outdated methods. However, some who have invested their entire careers in the traditional process express strong resistance, fearing the erosion of established design practices and the expertise they've cultivated.

The central question facing product development is whether extensive upfront research and meticulous mocks or rapid iteration and shipping leads to the best outcomes. Jenny argues that with non-deterministic AI models, rapid execution and testing with real data are often superior. It's nearly impossible to mock up all potential states, making direct user interaction with working prototypes essential for discovering true use cases and improving the product.

In Jenny's day-to-day at Anthropic, a significant portion of time is spent simply 'catching up' on the torrent of information and new projects. Her role often involves connecting disparate ideas, guiding engineers, and ensuring cohesive experiences as products are built. This requires not just design skill, but a high degree of curiosity and the ability to synthesize vast amounts of rapidly changing information, acting as a crucial central node in a dynamic ecosystem.

Jenny describes how her role now includes allocating time for traditional design vision, but also for 'jamming' directly with engineers to give feedback and consult. Crucially, she also dedicates part of her day to writing code herself, polishing and implementing features. This blend of vision, collaboration, and direct coding represents a significant evolution, allowing designers to influence the product at a deeper, more granular level than ever before.

While traditional design activities like user research, prototyping, and mocking still exist, their proportion within the overall design process has drastically changed. Previously, 60-70% of a designer's time might have been spent on mocking and prototyping; now, Jenny estimates it's closer to 30-40%. More time is allocated to direct collaboration with engineers, working in code, and navigating the evolving technological landscape.

Jenny details her current AI design stack, which heavily features Anthropic's Claude family of tools. She uses Claude chat for general queries, Claude co-work for managing longer-running tasks, and Claude Code, often integrated with VS Code, for front-end tweaks. She also leverages Claude Code remotely via mobile and Slack, highlighting the flexibility and power of interacting with AI directly to rapidly iterate and implement design changes, such as tweaking an icon with a simple command.

Even with the shift towards code, Figma continues to play a vital role in Jenny's workflow. She stresses its unparalleled ability to facilitate broad exploration, allowing designers to brainstorm 8-10 different options for a problem. Unlike the linear nature of coding tools, Figma provides a canvas for divergent thinking and refining fine visual and interaction details, ensuring that creativity and multi-faceted problem-solving remain central to the design process.

Jenny shares her journey of moving from an Individual Contributor (IC) role to management, and then back to an IC role at Anthropic. She emphasizes that her time as an IC, directly engaging with the evolving design process, equipped her with invaluable hard skills and empathy. This experience is now crucial for understanding and guiding teams in a landscape where traditional management approaches may no longer suffice, and even engineers are struggling to keep pace.

In an era of continuous delivery, maintaining craft and trust hinges on transparent, rapid iteration. Jenny explains that launching features as 'research previews' and openly committing to frequent improvements helps manage user expectations. The key is to genuinely listen to feedback and visibly incorporate it, demonstrating that the product is continuously evolving. This approach fosters trust and engagement, even when releasing early-stage products.

While AI is rapidly becoming proficient in taste, judgment, and design, Jenny asserts that the human brain's enduring value lies in deciding 'what actually gets built' and 'what truly matters.' The most challenging aspects of software development aren't technical execution but human disagreements and decision-making about product direction. AI can provide data and suggestions, but it cannot solve these fundamental human disputes or assume ultimate accountability for choices.

Jenny predicts that AI interfaces will evolve into a combination of traditional graphical user interfaces (UIs) and direct chat/terminal interactions. While chatbots offer infinite flexibility and have become an unexpected gateway to AI, UIs still excel for specific, tactile tasks. She foresees UIs becoming increasingly AI-generated, simplifying their creation, while direct conversational interfaces continue to be a powerful, rich medium for human-computer interaction, enabling communication across vast intelligence levels.

Jenny's experience moving between IC and management roles has taught her the critical need for managers to stay hands-on. She believes that good managers in the AI era must periodically return to IC work to truly understand the evolving design process. This direct engagement fosters empathy and equips them with the relevant skills to lead teams effectively, rather than relying solely on outdated management frameworks. This constant re-engagement ensures leaders remain aligned with the rapid pace of change.

Jenny seeks three distinct designer archetypes: 'Strong Generalists' with deep expertise in a few core skills who can flex across different areas; 'Deep Specialists' who are top 10% in a niche like visual design or engineering; and 'Craft Newgrads'—humble, quick-learning early-career individuals eager to build and adapt without rigid preconceptions. She also stresses that while coding isn't mandatory, familiarity with coding tools is increasingly essential for all designers.

Jenny challenges the notion of strictly avoiding 'low-leverage' tasks. She argues that seemingly small acts, like meticulously testing the product, personally fixing a bug, or even writing a thoughtful anniversary card for a teammate, can be incredibly high-leverage. These actions demonstrate deep care, build trust, and set a powerful example for the team, creating a strong culture where everyone feels valued and standards are implicitly understood. This blend of hands-on engagement and genuine connection defines effective leadership in her view.

Jenny encourages playful 'roasting' among team members, not as a means of disrespect, but as a sign of deep psychological safety and comfort. Similar to how friends playfully tease each other, a team that feels secure enough to poke fun knows they trust and respect one another. This dynamic indicates that people aren't afraid to speak up, even challenging leaders, because they know they won't be penalized. It's a barometer for a truly open and high-trust environment.

Jenny embraces Evan Tana's 'legibility framework' which categorizes ideas and founders as either 'legible' (easily understood, perhaps less novel) or 'illegible' (frontier, complex, not yet clear). For designers, the key is to spot illegible ideas – those with underlying energy and potential, even if not fully articulated – and transform them into something understandable through storytelling, UX, or form factor. This approach allows designers to act like internal VCs, identifying and nurturing high-potential, unconventional concepts within their organizations.

Jenny recommends 'The Power Broker' by Robert Caro, a monumental biography that emphasizes long-arc thinking and understanding how things get done, despite its length. She also loves 'Insomniac City' by Bill Hayes, a poignant memoir about love and mortality. For TV, she highlights 'The Bear' season two, praising its depiction of highly competent individuals excelling at their craft. Her favorite product is Retro, a photo-sharing app focused on weekly memories, while her life motto is the deceptively simple, 'It is what it is,' embracing life's inevitable challenges with acceptance.