I Used to Love H.E.R.
By David Thomson, Founder Emeritus, Momentum Design Lab. Currently CEO at Inevitable AI.
This comes from building products through every major shift in the digital era. Founded a UX agency in the ashes of the dot‑com crash, scaled it into one of the top firms in the world, worked with Fortune 500 companies and high‑velocity startups, hired and led hundreds of designers, and delivered results in every market cycle. That company exited successfully.
For the past year, I've been unshackled from the old world and fully immersed in solo AI-native product development. I can tell you with absolute conviction… This cycle is different. And you might not be ready for it.
The Overdue Death of UX as We Knew It
Over the past few months, conversations with peers, agency owners, and executive design leaders keep ending in the same uncomfortable truth when no one else is listening: that the design that's dominated the industry for the past 15+ years is dead. The UX industry is in denial. RIP.
This isn't an attack on the work we've done. It's a reckoning with this new reality. The physics of our industry have shifted and survival requires understanding where we really stand. We built our careers on the old constraints. Now we're watching the industry's DNA get rewritten in real time. You don't adapt by protecting feelings, you do by facing the hard truth before it's too late.
How We Got Here
Design has been on the ropes for well over a decade.
Tech companies hoarded talent after the iPhone proved what design could do. Scarcity created a market for manufactured thin talent, pumping out narrow-skilled grads who could be hired, but not be led. Leadership roles filled with non-designers who couldn't actually lead designers. The entire industry got creatively diluted.
The UX community open-sourced our own craft with design systems and strategy templates, making it worse. We ended up diluting the creative essence, the craft, of design and strategy themselves and didn't think it was going to matter. Honestly, it was a great time to be an agency owner watching the industry sabotage itself. "We don't trust our team, we're looking for better execution from you." Thank you, and good night!
Here is something no agency owner will tell you… scarcity drives demand and also tanks your profits simultaneously because it's a vicious and delicate balance between breaking even, keeping your employees happy, and funding your kids' college. You gamble your survival daily.
The talent hoarding had been hampering our ability to hire, driving junior wages way above what agencies could afford to pay even senior talent. Then came the chef's kiss. We were doing the opposite of what the tech companies were doing by being careful and meticulous with our hiring and crafting internal talent pipelines, so when the correction started, we had the talent they didn't have and they needed us. Aside from Covid, 2016 to 2022 was mostly a great run due to this.
Then the Russia-Ukraine war gave companies cover to slash the product of all these colliding factors: low-productivity design teams that lacked the strategic mindset they needed at scale.
As fast as a Zoom meeting ending, the lights went out for everyone. Why would anyone trust UX after that? The layoffs sucked all the strategic air out of the room. Big Tech created the problem, then made design look disposable when they were done with it.
This may sound like I have contempt for UX. I don’t. I have contempt for what happened to it. This is my love letter to what doesn’t exist anymore.
And while all of this was happening, voices outside the community got control of the microphone. UX influencers on TikTok and Instagram with no UX background somehow became the loudest voices in the room. Algorithms rewarded the performance of expertise over the reality of it. The profession’s narrative was rewritten by people who had never actually done the work.
Inside the community, the conversation stalled in process theatre. Arguments over synthetic research, semantics, authenticity, and what is “epistemically grounded” drowned out forward motion. The debates became about who could put on the most convincing show. Meanwhile, AI systems listened, indexed, and reasoned right past us, even reasoning about themselves. Some of us didn’t notice. Many still haven’t and worse, don’t even understand the impact of what happened next.
The Day Everything Changed
On June 20, 2024, Anthropic dropped Claude Sonnet 3.5 with all of this encoded into it.
ALL. OF. IT.
It was a hurricane kick to design's head.
Design stopped being a bottleneck and became a chat box anyone could use. From that moment, it's been in a coma, still breathing but on life support. It just doesn't know it's dead yet.
They Canva'd the entire product development industry in one single release. This was the nexus of nearly every single modern development and design tool released this past year. And if any existed before, this turbocharged them into relevance.
It wasn't just Anthropic, they just opened the door. Every model provider since has followed. OpenAI, Google, Meta, Alibaba, xAI, plus Kimi, DeepSeek, GLM, and dozens you've never heard of. The capability is now ambient in the technology itself.
What's Dead and What's Not
Design as critical thinking about what to build? More important than ever.
Design as a standalone, gated discipline between research and engineering? Over. It's being collapsed into the AI execution loop and democratized. It's not just in one place, it's in all places at the same time. You can try to argue this all you want, but sitting in front of Cursor and creative directing a full-stack, living product seismically shifts things. We designed ourselves almost completely out of the loop, but not entirely.
Dimensionality isn’t dead either. Two years ago, I told my team straight: adapt fast, skill up on anything, differentiate yourself, or you probably won’t be here in a year. This may have sounded like a threat, but agencies can't afford to operate in a time vortex. It's not that anyone was going to become a bad designer, but what it takes to execute was moving faster than anyone could keep up, and every single extra dimension counts. Because a designer in an agency's job is to be better than every other agency's designers. Because you and every other agency are supposed to be better than in-house. And you better be better than the talent on the market. That's the math. If your math starts to not compute, it's because your designers aren't keeping up. Or your service model isn't. Or both. It's that simple.
The thing is, this no longer applies to just agencies. It’s everyone in every position. The playing field is opening up and the traditional lanes are up for grabs. It might be time to evaluate your math to see if it still computes or which lanes make sense.
I know what's next because I've lived it for the past year on my own. At the risk of turning myself into Cassandra, everything in this piece comes from being inside that shift, not watching from the sidelines on LinkedIn. Once you’ve seen what’s possible, you can’t unsee it.
The New Weekly Reality
I now have a near-weekly ritual. Last Tuesday, I built a complete SaaS application before lunch. Authentication, database, API. Not a prototype. A functioning product. Three hours. Before, that would have taken a team four to five months after six to eight weeks of research and design.
Here's what nobody talks about: I couldn't ship it. Like most weeks, it sits there before joining 21 others in a folder, collecting dust. But most don't make it past 3 hours, not because they don't work, but because I don't know if they should exist. Because they have no billing infrastructure. No admin tools. No compliance frameworks, no observability. Because I need a team to support them. Banks, insurance, employees, accounting, all the fun stuff.
You can build a product faster than you can meet about it. Create product at the speed of idea. That means infinite velocity in every wrong direction unless you have conviction and productization from the start.
The New Physics of Product Development
The old "fast, cheap, good, pick two" assumed human labor was the bottleneck. AI makes execution essentially free. Now the constraints are different:
Conviction: Know What to Build
When you can build in hours, the cost of being wrong hits opportunity, not budget. Every hour perfecting the wrong thing is an hour not finding the right one. Conviction must run in parallel with execution, validated continuously. Real research still matters: PhD-level rigor that measures cognitive impact, models market reactions, simulates consequences. The better the conviction loop, the more everything downstream benefits.
And before the design thinking defenders jump in, yes, empathy and testing still exist. They just got compressed into hour four against a real product and stay there forever. The process doesn't die, it becomes ambient.
Architecture: Make It Real
Your Claude or Cursor demo is still a toy. Where's your multi-tenancy? Your billing layer? Your admin panel? Your audit logs? Your GDPR compliance? Your SOC 2 readiness? Your pen testing? The productization gap is where ideas go to die now. Not in presentations or Figma files, but because they can't get over the hill to secure production release. Without these from day one, you'll rebuild everything later, if you get there at all.
Evolution: Outrun the Market
By the time you validate, the market moves. By the time you productize, competitors launch. By the time you find product-market fit, the market shifts again. The old loop was research, then build, then ship, then learn. The new loop is everything, simultaneously, constantly. Products either evolve in real-time or instantly fossilize. No version 2.0. Just continuous metamorphosis or death.
Design is Dead. Long Live Design.
The middle of the industry is hollowing itself out. What's left are four archetypes:
The Epistemologists: PhD-level researchers who architect belief systems, measure cognitive impact, and run longitudinal studies. They don't gate the process anymore. They operate in parallel with execution, feeding the loop with evidence that strengthens or challenges product direction. Their work becomes more valuable as everything else becomes cheap and fast. Over time, they'll merge with AI research.
The Orchestrators: Full-stack strategic designers inside agencies, consulting, or software companies. They thrive between vision and delivery, working directly with clients or stakeholders to translate strategic intent into executable reality. They navigate messy organizations, balance competing agendas, and keep products coherent through design, engineering, and release. They speak the language of leadership and execution, often in the same meeting. Before you finish a thought, they already have a deck in progress because they know exactly where you're headed. In an AI-native loop, they orchestrate the why, ensuring every iteration stays rooted in original purpose. Building an agency full of these people taught me one thing: they're extremely hard to find and even harder to keep.
The Shapeshifters: Operators who validate, design, code, market, and ship, cycling roles multiple times daily. Craft matches the Code even though they ditched Figma for code execution at least six months ago… because it was slowing them down. They can debug code, run just enough *nix commands to navigate with confidence, and are unafraid of npm and git. They design in the terminal and manipulate the UI directly in the browser console. They didn't wait around for the industry to rm -rf them because they saw the opportunity in front of them and charged ahead. They're using AI to do the work of five 10x designers and engineers while keeping the cohesion of one. In rare cases, a Shapeshifter is also an Orchestrator. They are the new 100x unicorns.
You don't team them up. They are the team.
The Survivors: The last holdouts in slow-moving sectors. Healthcare, Financial Services, and Government where regulation and bureaucracy buy time. Their superpower is institutional insulation which is a buffer from the AI velocity storm. Their trajectory should be stable…until it isn't. These roles are safe harbors, but they risk career obsolescence if they wait too long to adapt.
Not one of these? Here’s the opportunity and my advice… you need to find a way to join the Orchestrator or Survivor packs, and you might be eyeing Epistemologists but this isn’t the good idea you think it is. Or jump over to CX where this still matters, product management for some temporary shielding, strategy consulting where your elbows are out, way upstream in innovation where time isn't as compressed as product development, an agency whose brand anchors them in narrative control, downstream in AI implementation or slow digital transformation where in both cases you'll have a strategic seat at the table.
Three Crystal-Clear Realities
After diving deep into AI-native product development, three truths became undeniable:
Execution is near free. Direction is priceless. The real cost is maintaining conviction inside the loop, continuously validating why you're building, not just what.
Prototypes are everywhere. Products are scarce. The ability to go from demo to deployed platform, with all the boring necessities baked in from day one, is the new moat.
Static products are dead. Living systems win. If a product can't evolve faster than user needs, for agentic needs, it's already obsolete. Can people and agents have a dialogue with it? If not, maybe it's dead.
The Infrastructure Coming Fast
This isn't a theory. The tools are being built now, not by incumbents optimizing for the old world, but by those building through the transition. The questions have changed. Not "what does a user need for this screen?" but "what does this agent need from that agent?"
Infrastructure that validates ideas before they consume resources. That transforms prototypes into platforms without rebuilding. That makes products evolutionary by default. The gaps are being filled, the conviction problem is being solved, the productization path is being automated, and the evolution engine is being embedded.
Once this last-mile infrastructure becomes widely available, and it will soon, legacy roles and workflows will collapse. The entire pipeline of design, prototyping, engineering, QA, DevOps, SecOps, and validation will become a single continuous CI/CD pipeline loop. We're nearly there.
The reality extends beyond design. The entire talent stack is converging into the pipeline above. The lines between design, product management, and engineering have already blurred. Once last-mile infrastructure arrives, every role operates on the same playing field.
Proximic lanes are opening up. Your SecOps colleague might accidentally out design you using your own design system while solving a security problem. You might string together some agents in an afternoon that handle what the data science team has been building for months because your curiosity beat their rigor. Engineers will optimize product management workflows that seem untouchable. Product managers will architect the new product intentionally. Everyone becomes capable of everything and your name supersedes any meaningless title.
It's not chaos, it's democratization. The winner isn't who protects their turf but who adapts fastest to having no turf to protect. Who automates best. Who integrates most seamlessly. Who has the best critical thinking. Who inverts politics and makes it easy. Who floats instead of anchors. Shapeshifts between whats needed now.
This is the new new. You are part of the hive and there is no Queen. The work self-organizes around whoever can ship. This is happening already, this is the nextgen five person software company ramping into 30M ARR with no funding and no debt.
Will it be messy? Sure. But in the midst of all this transformation, sometimes the best response is to breathe, see where the patterns connect, and choose which dimensions to apply to you and where.
What To Do Now
Here's what most miss out of all of this… it looks like destruction, but it's also liberation. If you can swim in ambiguity, thrive between vision and execution, and orchestrate instead of just execute, this is the greatest opportunity in the history of our industry.
One person with conviction and the right tools can now build what once took entire companies years to do. Small teams can compete with enterprises. Experience plus AI becomes a multiplier that shatters every traditional ceiling.
But you have six to eight months to decide which side of this transition you're on before the market decides for you. After that, nostalgia about your process won't save you.
Don't Count Agencies Out
Agencies already have the talent filters, creative range, strategy, and cross-functional expertise many product companies let atrophy. Why would a top designer now join a product company as a node in the chain where design is a cost center instead of a strategic driver? We may see a massive reversal of where talent goes to live.
The smart agencies will plug into the why, build conviction, build their own solutions, deliver the products around it, and take them to market for the customer, owning the entire strategic lifecycle without any competition. Or they can dominate downstream where the battle for AI strategy is unfolding. Design led AI. Agencies have been creating markets for other people for decades, its time to own their own.
The future agency may literally function as "software company as a service," owning everything from strategy to market delivery.
The Bottom Line
Survival won't be about defending the old process. It will be about proving you add value inside the new one.
For those who can, the upside is unlimited.