A Hatch Conference talk by Jenny Wen (Design Lead at Anthropic, formerly Director of Design at Figma) on why “trust the process” no longer fits the AI era.


Source: https://www.youtube.com/watch?v=4u94juYwLLM

Summary:
TL;DR - Why can't designers just “trust the process” anymore?

  • The traditional design process (research → persona → journey → problem → solution) isn't wrong, but it no longer makes sense to apply it rigidly in today's reality.
  • AI accelerates prototyping so PMs, engineers, and designers can all build solutions quickly, while smaller teams are expected to do more.
  • The biggest problem is worshipping process artifacts (personas, journey maps…) rather than focusing on the final experience. Users don't care whether you followed the right process; they care whether the product is good.
  • Great work often doesn't follow a standard process; instead, it comes from:
    • starting with a solution when technology opens new possibilities
    • iterating deeply on details, craft, and quality
    • using intuition developed through feedback, research, and data
    • removing or remixing process steps to fit the context
    • designing to create delight, not just to “solve a problem”
  • Additional views noted in the recap:
    • Process is still useful for juniors, new teams, and high-risk problems (such as training wheels)
    • Intuition works only with a strong understanding of the problem space
      Conclusion:
      No design process fits every project. In the AI era, what we need to trust is no longer the process, but the designer's judgment, taste, and responsibility for output quality.

Why can't designers “trust the process” as before?

For years, designers were taught that simply “trust the process” would lead to good products.

The familiar process usually looks like this: Research → Persona → Journey Map → Problem Statement → Ideate → Wireframe → Test → Polish

In her Hatch Conference talk, Jenny Wen (Design Lead at Anthropic, formerly Director of Design at Figma) argues that

this process no longer reflects the reality of modern product work. Not because the design process is “wrong,” but because the context has changed so quickly.

A new context: Why old processes are starting to break

1. AI changes speed and roles

Today:

  • PMs and engineers can build prototypes very quickly, sometimes faster than writing a problem statement.
  • Designers no longer just “draw screens”; they are becoming more full-stack: prototyping, testing, and even implementing.

When speed becomes essential, a linear, sequential process begins to become a constraint.

2. Doing more with fewer resources

Layoffs and smaller teams, yet:

  • output expectations don't fall
  • individual responsibility increases

In this context, steps that involve:

  • lots of coordination
  • lots of rituals
  • but no direct value for the experience

…are increasingly questioned.

3. When everyone can make something, craft becomes the difference

AI has raised the baseline of output: with a few prompts, anyone can create an “okay” UI.

As a result, a designer's value no longer lies in producing artifacts, but in:

  • Taste - choosing what is worth making
  • Craft - attention to detail
  • Quality - the overall feel of the experience

Many beloved products aren't loved for extensive research, but because they:

  • feel smooth
  • respond quickly
  • have personality
  • feel satisfying to use and look at

4. Every project is a different problem

No two projects are alike:

  • different stakeholders
  • different technical constraints
  • different timelines
  • different levels of ambiguity

As a result, a fixed checklist rarely produces great work in every context.

Users don't care whether you made a journey map.

They care only whether the final experience is good.

How is great work actually created?

Jenny identifies 5 common ways good products are created, even though they without following the standard design process.

1. Start with the solution, not the problem

Sometimes, only after seeing a solution prototypedoes a team realize:

“This is what we should pursue.”

For example, Claude Artifacts began with a researcher's prototype.

Designers saw its potential, iterated, and launched it - and the product changed how people see AI:

not just as “writing code,” but as creating something immediately interactive.

2. Be ruthless about detail - iterate on quality over time

With FigJam, after launch, the team didn't rush to add features. Instead, they refined:

  • snapping
  • alignment
  • selection border
  • colors
  • typography
  • toolbar interaction

This is a continuous quality loop, rather than a Double Diamond where “done means done.”

3. Work with intuition (intuition isn't guessing)

Theo Jenny, intuition isn't merely a feeling.

It is the ability to:

  • make quick decisions
  • based on a deep understanding of users and the product

Intuition is built from:

  • real feedback
  • past research
  • data, metrics
  • psychological principles and cognitive biases

The goal is to develop an internal model of users, so that:

  • you don't need research for every small decision
  • you can still make sound decisions quickly

4. Skip steps, combine steps, or design your own process

Process should be treated as toolbox, rather than rigid rules.

For example:

  • A 5-day Design Sprint → shortened to 3 days, focused on prototyping
  • Amazon-style “working backwards” → at Figma, a tweet or landing-page headline

What matters isn't following the process perfectly, but creating shared understanding and good decisions.

5. Design simply to make people smile

Many FigJam features - cursor chat, emotes, stamps - didn't come from a problem statement.

The team simply:

  • built something to try
  • let people use it
  • saw them smile, enjoy it, and like it

Despite many remaining bugs, the team knew:

“Something about this feels right - it's worth pursuing.”

Some perspectives from comments

Some comments noted from the video add the following perspectives. They are not a representative survey of the design community.

1. Process isn't dead - the issue is how it's used

Many emphasize that:

  • process is still needed for juniors, new teams, and complex problems
  • the problem is mechanical application, treating process as the goal rather than a means

Process should be training wheels, not an absolute truth.

2. Artifact ≠ Impact

A common feeling:

“We ship artifacts, not outcomes.”

Designers are judged by:

  • the number of frameworks
  • the size of their decks

rather than the quality of the final experience.

3. Senior designers feel vindicated

Many experienced designers share that:

  • they've worked outside the process for a long time
  • but had to disguise intuition in the language of research

This video is seen as recognition of judgment and experience.

4. Legitimate concerns for junior designers

The big question:

“If they can't trust the process, what can juniors rely on?”

One perspective noted:

  • process is still needed to learn how to think
  • but it must be taught clearly that: process is temporary scaffolding

5. AI amplifies the differences

AI doesn't replace designers, but it exposes the gap between:

  • people with taste, craft, and judgment
  • and people who only follow checklists

The increasingly important skill isn't what you can make, but:

knowing what you shouldn't make

Conclusion

  • No design process is right for every project
  • A designer's value isn't in repeating a process, but in the ability to:
    • choose the right tools and steps for the problem
    • adapt the process to the context
    • take responsibility for output quality (strategy + craft)

We're living in a time that is frightening but also exciting.

To survive and grow, designers need to do something harder:

Not just trust the process - but trust ourselves again.