SCTY
Back

January 25, 2026

Me-Search, Re-Search, Pre-Search

The design process is under pressure.

But the cracks were showing before AI arrived.

In February 2023, Rebecca Ackermann published a retrospective in MIT Technology Review asking what went wrong with design thinking. The answer: it became theater. Post-it sessions that generated excitement but not products. Consultants who, in the words of one nonprofit director who worked with IDEO.org, "were not contextualized in the problem at all... out of touch with reality."

Jake Knapp, who ran design thinking workshops at Google, put it bluntly: "Execution has always been the sticky wicket for design thinking." The brainstorms happened. The artifacts got made. The products didn't ship.

By 2023, IDEO was hemorrhaging staff. The methodology that promised to democratize design had, critics argued, reinforced existing inequities instead. "Innovation theater" became the term for checking boxes without changing anything.

Then AI arrived. And it didn't cause the collapse—it accelerated it.

Tools that took weeks to learn now take hours. Prototypes that took days now take minutes. The PM can vibe-code a working demo before you've finished writing the problem statement. The process that was already struggling to produce results now can't even justify its own timeline.

Two voices are offering different answers. One says: skip the process, trust your intuition. The other says: don't skip discovery, ground everything in customer understanding.

Both are right. Neither is complete.

This piece offers a way to think about where creative work actually comes from, and how that's changing. Not to declare a winner, but to help you see the options.

The Three Modes

Me-Search: Trust Your Intuition

Jenny Wen is a design lead at Anthropic and former director of design at Figma, where she took FigJam from zero to launch. At Hatch Conference, she made the case that the traditional design process has become theater.

Personas. Journey maps. Problem statements. Brainstorms. Wireframes. She's watched teams spend weeks on artifacts nobody uses while the actual work happens elsewhere.

"The user doesn't give a shit about the process artifacts you made or whether you made the perfect user journey. They care about the end experience."

Her alternative: trust your intuition. Not guessing. Intuition as "the ability to make reasoned judgments quickly," a shortcut built through sustained exposure to your users, your product, your domain.

She builds hers by reading feedback constantly (Twitter, Reddit, internal boards), attending research sessions, watching old recordings, checking dashboards. The goal: building an internal model of users so complete that you can move fast without second-guessing.

Me-search means the source of truth is inside you. But only because you've done the work to internalize what's outside.

When it works: You have genuine domain depth. Speed matters more than certainty. You're iterating on details within a known space. Your intuition has been tested and you know your hit rate.

The risk: Your intuition is trained on the past. If everyone's intuition is shaped by the same apps, the same design systems, the same internet, "trusting yourself" might mean reproducing what's familiar.

Re-Search: Trust Your Customers

Teresa Torres has spent 14 years as a product discovery coach. Her book Continuous Discovery Habits shaped how a generation of product teams think about building.

Her argument: it doesn't matter how good your evals are, or your intuition, if you picked the wrong customer problem to solve. AI makes this worse, not better. FOMO and hype pressure teams to ship AI features without understanding whether anyone needs them.

"The things that give me nightmares most when working with teams building AI products is when the product says 'ask me anything.'"

That phrase signals you haven't scoped the problem. You don't know what use cases you're covering or how you'll handle what falls outside.

Her method: weekly customer touchpoints. Small research activities. Mapping the opportunity space. Assumption testing. The scientific method applied to product.

One key insight: don't ask customers what they think or what they want. Those questions produce unreliable answers. Ask them to tell you about specific past experiences. "Tell me about the last time you..." This invokes memory, engages slower thinking, produces signal you can trust.

Re-search means the source of truth is outside you. In observed behavior. In evidence. In the gap between what you expected and what happened.

When it works: You need to validate that a problem exists. You're entering a new domain. The cost of building wrong is high. You need to convince others with evidence.

The risk: Customers can only articulate needs within what they already know. They'll ask for a faster horse. Discovery anchors you to the present when AI might require a leap.

Pre-Search: Trust the Technology to Show You What It Can Do

This position is harder to attribute to a single voice. It shows up in the work of Bret Victor, Andy Matuschak, the speculative design tradition. And, implicitly, in Wen's own Claude Artifacts example.

The argument: neither your intuition nor your customers can see what's newly available when technology is changing this fast. Both are trained on what exists. AI changes what can exist.

"We didn't know it was a problem worth solving until we saw the solution."

That's Wen describing how Claude Artifacts emerged at Anthropic. A researcher built a prototype. Not because user research suggested it. Not because intuition said it was right. Because the model could do something new and they wanted to see what that meant.

Bret Victor calls this "inventing on principle." You fight for a vision of what technology should mean for human capability by building things that embody that vision.

Andy Matuschak observes that the most interesting work in tools for thought isn't driven by user research or craft alone: "Ted Nelson had a dream. Alan Kay measured progress in 'Sistine-Chapel-Ceilings/lifetime.'"

Speculative design takes this further. The artifact's job isn't to solve a known problem or express existing taste. It's to reveal a possibility. To provoke.

Pre-search means the source of truth is in what the technology can now do. You explore that space, build provocations, and watch what resonates.

When it works: The tools are changing faster than anyone can track. You don't know what's available yet. You need to expand your own imagination, or your team's, or your customers'.

The risk: Untethered futurism. Beautiful provocations that never ship, never create value, never connect to what anyone needs. A different kind of innovation theater.

What's Still True

Some things haven't changed.

Craft still matters. AI raises the floor. You have to be better than the floor. The apps that feel considered (Linear, Notion Calendar, that camera app with the tactile knobs) win because details compound. Caring is visible.

Customers still matter. People use products to do things they care about. If you're not solving a real problem for a real person, nothing else saves you.

Intuition still matters. Fast, grounded judgment is valuable. The best operators make decisions quickly because they've internalized their domain. This doesn't get automated away.

The scientific method still works. Hypothesis, test, revise. Whether you call it discovery or evals or assumption testing, the pattern holds: have beliefs, expose them to reality, update.

What's Changed

Speed changed the economics of process. When building was expensive, heavy upfront process made sense. Measure twice, cut once. But when building gets cheap (when a PM can prototype in an afternoon, when AI can generate fifty variations before lunch) the ROI of process shifts. Less valuable: artifacts produced before building. More valuable: feedback loops run while building.

The tools are outpacing mental models. Customers don't know what AI can do. Neither do most designers. The technology moves faster than intuition or research can track. User research tells you what problems exist in the current frame. AI changes the frame.

The floor rose. Anyone can generate a passable UI, a decent logo, a serviceable blog post. Good for access. Bad for differentiation. What separates good from passable isn't output quality anymore. It's judgment: knowing what to make, what to cut, what matters.

Homogenization became a real risk. Millions of people using the same tools with similar prompts. Outputs converge. The Midjourney aesthetic. Purple gradients and Inter font. The premium version of the same sameness. This applies to human judgment too. If everyone's intuition is trained on the same internet, trusting yourself might mean reproducing what's common.

The Way Forward

There's no single answer. The right mode depends on context.

Use me-search when you have genuine domain depth, speed matters, you're iterating within a known space, and your intuition has been calibrated through feedback.

Practical move: Build your internal model deliberately. Read feedback constantly. Attend research sessions even when they're not about your feature. Track the gap between what you expected and what happened. That gap is your calibration.

Use re-search when you're unsure the problem is real, entering a new domain, facing high cost of building wrong, or needing evidence to convince others.

Practical move: One customer conversation per week. Ask about specific past experiences, not opinions. Write down what you expect before you ship, then measure. Let the gap between expectation and reality sharpen your intuition over time.

Use pre-search when the tools are moving faster than you can track, you don't know what's available, or you need to expand what's imaginable.

Practical move: Build provocations. Not prototypes (solutions to known problems) but artifacts that explore what the technology can do. Show them to people, not to validate, but to expand what everyone can see. Notice what resonates.

The Integration

The most effective operators move between modes.

Pre-search to see what the technology opens up. Me-search to move fast on details. Re-search to validate that it matters.

Or: re-search to understand the problem space. Pre-search to explore how new tools might address it. Me-search to execute with craft.

The sequence matters less than the awareness. Know which mode you're in. Know what it's good for. Know when to switch.

Where This Leaves Us

AI broke the old playbook. The design process that worked for the last decade assumed humans were the bottleneck. That's no longer true. The tools assumed a slower rate of change. That's gone too.

Nobody has the new playbook yet. The intuition-trusters are running on pattern-matching trained before the shift. The customer-trusters are gathering evidence about a world that's already different. The capability-explorers are building futures that may never connect to present needs.

Three responses to the same disruption:

  • Go inward, trust your judgment, move fast.
  • Go outward, trust evidence, stay grounded.
  • Go forward, trust the technology, expand what's imaginable.

Each breaks down somewhere. Each works somewhere else. The skill now is sensing which mode fits the moment, and switching before you're stuck.

Sources