For most of my professional life, design followed a fairly clear and structured path.
You / I started by understanding the problem.
You / I researched.
You / I sketched ideas.
You / I turned those ideas into wireframes and mockups.
And eventually, the work was handed over to engineering so the product could be built.
It was a process that brought order to complexity — and for a long time it worked remarkably well.
Over the years I’ve worked in different disciplines where designing and building things meant thinking carefully about consequences before something became real. In some fields, once something is built, it stays there for a very long time. So planning, precision, and predictability mattered a great deal.
But the world of digital products has been changing rapidly — and with the arrival of AI, the pace of that change seems to have accelerated even further.
Recently I listened to a fascinating interview with Jenny Wen, who leads design for Claude at Anthropic, on Lenny’s Podcast. The episode is titled “The design process is dead. Here’s what’s replacing it.”
At first, these statements sound a bit intense. But the conversation behind them is thoughtful and surprisingly grounded in the realities that many product teams are already beginning to experience today.
Jenny talks about how AI is reshaping the way products are built, how design roles are evolving, and why the traditional sequence of ...
research → design → handoff → build is beginning to break down.
It’s an 80-minute conversation, and I highly recommend listening to the full interview. Youtube: Lenny's Podcast
Below are the ideas that stood out the most to me — along with a few reflections from my own perspective.
My Key Takeaways & Learnings from the Interview
1. The Classical Design Process No Longer Fits
The traditional design workflow we’ve relied on for years is increasingly out of sync with modern product development.
The familiar sequence looked like this:
Research → Wireframes → Mockups → Handoff → Build
But AI has dramatically accelerated development cycles:
- Engineers can build features the same day
- Long design phases become impractical
- Iteration cycles are much shorter
Design therefore needs to become faster, more collaborative, and more iterative.
2. Design Splits Into Two New Working Modes
Jenny describes two emerging roles for modern designers.
Execution Partner
Designers work directly with engineers during development.
Typical activities include:
- improving UI directly in code
- implementing small changes themselves
- optimizing user flows while features are being built
Design happens inside the development process, not before it.
Short-Horizon Vision
Designers also define direction for the next 3–6 months.
This work includes:
- experimental prototypes
- exploratory ideas
- conceptual directions
Instead of traditional long-term roadmaps, the focus is on near-term exploration.
3. Determinism Is Also Dead
One of the most interesting ideas in the interview is that AI products are not deterministic.
Deterministic systems behave like this: Input → predictable output eg. 2 + 2 → always 4
Traditional software works largely this way.
AI products do not.
With AI systems ...
- the same prompt can produce different responses
- behavior is not fully predictable
- many situations only emerge in real-world usage
Because of this, AI products cannot be completely designed or mocked in advance.
They must be:
built → tested → observed → understood
This part honestly feels a bit unsettling to me. I come from a background where things were supposed to be predictable and controllable. In my family of tax consultants, 2 + 2 always equaled 4.
I also come from a craft background — my first apprenticeship was as an upholsterer — and later I studied and worked in architecture and urban planning.
In those disciplines, the idea of:
“Let’s build it first, test it, and only then understand it in real use”
can feel slightly terrifying. Am I the only one who wouldn’t think the same way in this context? 🤔 As it reminds me of some of the biggest architectural mistakes in history, where buildings or urban concepts caused major social problems only after they were realized. But there is one crucial difference:
An AI application is not a city, a neighborhood, or a concrete structure built to last for decades.
Software, Bits and Bytes can evolve much faster 😉- but it can have effects on society as well.
Still — the shift in thinking is profound.
4. Designers Must Work Much Closer With Engineers
This is also the part that makes the new reality feel less worrying. Design and engineering must become much more tightly integrated.
In the past the workflow looked like this:
Designer → Mockups → Handoff → Engineering
Today it increasingly looks like this:
Designer + Engineer → Build together
Designers are now:
- closer to the code
- involved during development
- part of continuous iteration
5. Mockups Are Not Dead — But There Are Fewer of Them
Mockups are still important, but they are a smaller part of the job. Roughly speaking:
In the past
- 60–70% mockups
Today
- 30–40% mockups
- 30–40% collaboration with engineers
- more prototyping and iteration
The center of gravity in design work is shifting.
6. Early Release Beats Perfect Planning
Shipping early has become more important than perfect upfront planning.
Many AI teams release features as:
“Research previews.”
The goal is to:
-
gather feedback early
-
observe real user behavior
-
iterate quickly
Trust is built through continuous improvement, not through perfection at the first release.
7. Job Profiles Are Merging
Design roles are becoming hybrid roles. Jenny Wen explains that designers have to adapt to this and expand their work to engineering and management.
In a way, this reminds me of earlier eras in architecture - as I mentioned I was and worked as architeckt und t´won planer.
Historically, the master builder was many things at once:
- planner
- designer
- engineer
- craftsperson
- project manager
Modern designers increasingly resemble this model.
The designers of today and tomorrow must increase to a mix of:
- Designer
- Product manager
- Engineer
Important new skills include:
- prototyping
- AI tools
- technical understanding
- product thinking
8. AI as Idea Generator — Not Decision Maker
AI can generate design ideas.
It can already:
- produce design variations
- suggest UI patterns
- generate interface options
But it cannot make product decisions.
Humans still decide:
- Which idea actually makes sense
- Which fits the product strategy
- Which should be built
And in my opinion this is essential — because humans bring ethical, social, and human judgment into product decisions.
9. Great Designers Recognize “Illegible Ideas”
A fascinating concept Jenny mentions is the difference between legible and illegible ideas.
Legible ideas
- immediately understandable
- familiar
- easy to explain
Illegible ideas
- initially confusing
- difficult to explain
- sometimes feel strange at first
But these illegible ideas often contain real innovation.
Jenny compares strong design leaders to venture capitalists for ideas — they recognize potential in concepts that others might dismiss too early.
10. Career Advice: Build Real Things
For designers — especially early in their careers — the most valuable thing is not perfect case studies.
What matters more is:
- building real products
- publishing prototypes
- shipping things
- gathering feedback
A real project can be far more valuable than a polished portfolio presentation.
The Essence of the Interview
If I had to summarize the entire conversation in one shift in thinking, it would be this:
Before
think → design → handoff → build
Today
build → test → learn → improve
Design is moving away from pure planning toward active product creation.
And my Personal Closing Thoughts
For me, this interview was both exciting and slightly unsettling.
It challenges many assumptions about how design work has traditionally been structured. At the same time, it opens the door to a more dynamic and collaborative way of building products.
What resonates most with me is the idea that designers are moving closer to the actual making of things again. Less distance, fewer layers, more direct impact.
In some ways, it feels like a return to an older model — where thinking, designing, and building were not separated into rigid roles, but part of a continuous craft.
And perhaps that’s the real shift:
Design is becoming less about delivering artifacts and more about actively shaping evolving systems.
If you’re interested in the future of design in the AI era, the full interview is definitely worth your time.
Comments
Post a Comment