By Robin Cannon, VP of Product, Knapsack

Ask an AI tool to build a screen today and watch what it reaches for. Inside tools like v0, Cursor, and Figma Make, a model will generate a working interface in seconds — and most of the time it builds from whatever components it already knows, not from yours. Often that means shadcn/ui, the open-source library these tools ship with and were trained on. Your team spent two years on a design system — the shared rulebook for how your product looks and behaves — and the fastest-growing contributor to your UI (user interface) never opened it.

That's the quiet shift worth naming: the heaviest consumer of your design system in 2026 isn't a person reading your docs. It's a model. And it doesn't read the way people do.

Figma says 8px. Your code says 12px. Your design system documentation says 10px. The AI picks one — not always the same one — and generates fast, plausible output. Your docs site, your Storybook, your brand guidelines: multiple sources, each served through their own MCP, all in competition for the model's attention.

The systems we built for human readers are straining, too. In zeroheight's 2026 Design Systems Report, 56% of teams now use AI in their design-system work — but only 15% say it's living up to the hype, and adoption has been the number-one challenge five years running. The job changed, and the artifact we keep polishing is the wrong one.

A component is a cache

AI doesn't consume components. It consumes context.

A button in your design system is the output of a hundred decisions — when to use it, what never to pair it with, which legacy variant you trust, and why. The component is what's left after that judgment gets compressed into something reusable. It's a cache, and like any cache, it's lossy: the decisions that produced it aren't stored inside it. People never noticed, because we rehydrate the missing decisions from memory and the person who built it. A model can't. It reads the cache as if it were the source and confidently ships the compression artifact as truth. AI didn't break your design system. It revealed that your design system was always a cache with no source behind it.

Picture it: an agent generates a checkout button, picks a near-match, styles it close to your brand, and ships. It looks right. But it's the deprecated variant; it misses your accessibility rules, and it drifts your checkout off-brand. The rule a human carries — never use the ghost button on a paid action — was never written anywhere the model could read it. The catalog described what the button is, not what your team decided about it. The richest part of your design system was never the components. It was the decisions behind them.

Same problem, two very different teams

This is where it splits.

If you're at a large or regulated organization, the risk is speed without control. AI generates production interfaces faster than any review process can keep up, and off-brand or non-compliant output is hard to catch after the fact. A governed answer — one trusted response when your design files, code, and docs disagree — is the guardrail that lets AI move fast while you still stand behind what it ships.

If you're on a smaller or mid-market team, the problem is the opposite. The headcount and maintenance of a traditional design system is more than you can carry. You don't need a cathedral. You need something you can point at your existing work and get value from in days.

Same architecture, two pains — and most of the market is solving neither. One camp races to automate code generation; another doubles down on documentation. Both improve the artifact that's no longer the point. The move is to treat the design system as the shared context people and machines both work from, and to govern it so it enables work instead of gatekeeping it.

That's the bet we're making with our Intelligent Product Engine — a control plane that gathers your design files, code, documentation, and guidelines, resolves what's true and what's in conflict, and serves a governed answer to whatever tool is doing the work.

A short test

You don't need a platform to start — you need to know where you stand:

• Are your tokens and names semantic — do they describe intent, not just appearance?
• Are the decisions behind your components written anywhere a tool could read them, or only in people's heads?
• Do your service journeys and blueprints live somewhere a tool can actually read — or only in Miro boards no model can parse?
• When two sources disagree, is there a governed answer — or just an argument?

The unit that's missing has a name: the decision record — the reasoning a component was compressed from. And the metric is shifting with it. For a decade we tracked component coverage (how much of the UI is in the library); the AI-era number is decision coverage — how much of the reasoning behind it a machine can actually read. Here's my bet: by 2027, design-system maturity is measured in decision coverage, not component coverage.

The design system isn't over. It's been promoted — from a site people were supposed to read into the thing both people and AI build on. The teams that notice will be the ones whose AI builds the right button without being told. Faster generation through AI is a given. We're betting on knowing what's right.

Read the full piece → and, if you're curious where your own decision coverage stands, try our short readiness assessment.

By Anamol Rajbhandari

In 1997, a small piece of software called Auto-Tune arrived in recording studios, built to do something narrow and technical, which was to nudge a slightly flat or sharp vocal back onto the note it had been reaching for. For an individual singer, the effect was close to magical, as someone who could mostly carry a tune but never quite land it could suddenly come out sounding clean and professional. Someone who could barely sing at all could be smoothed into something passable, so that almost overnight the range of what any single voice could pull off grew a great deal wider, the hardest and least forgiving part of singing having been quietly handed off to a machine.

As the tool spread through the industry, a large share of pop vocals began to converge on one another, all of them buffed toward the same glassy, centered, faultlessly on-pitch ideal. The quality that had made Auto-Tune so useful in the first place, its pull toward a single correct standard, was never going to confine itself to one voice. It worked on every voice at once, drawing each singer closer to the same target until the targets themselves grew difficult to tell apart, so that the lift and the flattening turned out to be one motion seen from two ends. Whatever raised each voice to the standard was also what drained the difference out of the whole.

Design is standing in that line, and both halves of the pattern are already visible in it. If we were to ask a designer who has spent a season working with the new generative AI tools whether they can now make more than they could before, the answer without hesitation would be a “yes”. They can do a great deal more, faster and across a range their own training never gave them. And if we ask the same designer whether the work coming out of the field as a whole is growing more varied, there is usually a pause followed by a slightly uneasy admission that everything seems to be starting to look like everything else. 

Both of those answers are honest accounts of what the designer is actually seeing, and the reason they sound like a contradiction when set beside each other is that they are describing a single event from opposite ends of it. The freedom each designer feels is what flattening the field and learning to hold those two facts in the same hand at once may be the least examined large change now underway in creative work.

The same thing has already run its course once through human faces. A decade of filters and photo-editing apps produced what one widely read essay named Instagram Face, a look that was poreless, high-cheekboned, faintly feline, and eerily consistent from one young influencer to the next. It was the machinery of the singers all over again, in that everyone was handed the power to sculpt their own image toward a standard that used to require money and surgery, and as everyone reached for that standard, the faces slid quietly toward one another. 

What a person feels in that situation is only their own face improving, while the thing no one can feel from inside a mirror is an entire generation converging on one, similar face. The gap between the private gain and the collective loss makes it difficult to notice while it is happening. These two domains have now travelled the full arc of it, which is reason enough to contemplate why the same thing happens to every craft that lets a machine of this kind stand between the maker and the thing being made.

Why the machine returns the average

The pull toward a single standard is built into the way any of these tools works, and it is there long before anyone reaches the question of taste. A generative model takes in an enormous body of existing work and forms from it a statistical picture of what such work tends to be, and then, asked to produce something, it samples from that picture, and the likeliest sample it can draw is always the crowded middle of the distribution, the average of everything it once absorbed. Producing the most probable next thing is the entire mechanism, and the most probable thing is, by definition, the average one.

It helps to picture a thousand people each asked to sketch a chair, and then every one of those sketches laid over the last until only the composite remains. The shape that shows through belongs to no one; it is a kind of ghost of a chair, the most chair-like chair it is possible to draw. It has every private quirk of every contributor cancelled quietly against every other. We run that same overlay across nearly everything a field has ever produced, and the result is roughly what the model is reaching into each time it answers a request. So it has all the feel of a considered choice drawn from the exact center of everything, averaged.

For most people most of the time, this may feel like a genuine gift and it would be dishonest to pretend otherwise. The average of all the design work ever poured into a model is a good deal better than what almost anyone could produce alone. A beginner is handed a level of competence that used to take a few years to earn. An overstretched professional is spared the tedious middle stretch of the work, and the reach of any single person expands dramatically the moment they sit down with the tool. The difficulty is only that the center everyone is being lifted toward happens to be a single location, so that raising a novice raises them toward it and quickening an expert nudges the expert toward it too. 

Thus, ten thousand people carrying ten thousand different sensibilities, all reaching into the same model on the same afternoon, are all guided gently to the same crowded middle. Each of them experiences only the lift, and none of them can perceive the crowd gathering around them. It is because the convergence of an entire field is not something that can be seen from inside one's own chair.

Instead of feeling like a limitation, it rather feels, if anything, like flight. Which is precisely why it meets so little resistance. Since sitting down with one of these tools a person finds that everything opens outward leads to a request producing one result and a second request producing a different one. The whole sensation being that of a door swinging wide onto an enormous and well-lit room, with no way of seeing that every other door opens onto the same room or that the tool is steadily guiding everyone toward the same bright patch at its center. It shows most plainly in what appears unbidden, in the way that prompting five different tools for something as ordinary as a dashboard tends to return similar, if not the same, small handful of layouts, the ones sitting in the densest part of the training data.

And because that middle is in fact good enough, the trap closes without ever announcing itself as a trap. Since leaving the average takes real effort, the effort of discarding the first answer and then the fifth, of wrestling with the prompt, of insisting on a stubborn opinion and throwing out whatever comes back merely competent, and on a deadline, with a perfectly decent option already glowing on the screen at no cost- the decent option tends to win. It is easy to argue that this seems lazy. But it is simply the arithmetic of an ordinary working day, and when that same reasonable choice is repeated across an entire profession, the field converges without a single person having ever decided that it should. From the vantage of any one desk this looks unmistakably like freedom, and it is only when the work is set side by side, across teams and companies and whole countries, that the convergence becomes visible as something closer to weather, a condition everyone turns out to be living inside without having chosen it.

As none of it began with AI, giving the technology all the credit rather flatters it, because design had been converging for a long time before any model could draw. Products imitate whatever already works, since copying the proven thing lowers risk and the lowering of risk is most of what a company exists to do. The result was plain long before generative tools appeared: dashboards resembling cousins across wholly unrelated firms, onboarding flows that reduce to the same five screens everywhere, settings pages so alike they could be swapped between products without anyone noticing. Best practice has always been a shared center under a more flattering name. What the machine changes is the speed and the completeness of the movement toward it. Because the older kind of copying still required a human being to notice a pattern, admire it, and choose to follow it—that requirement left lag in the system, and gaps, and room for a stubborn team to wander off somewhere strange, whereas the new version arrives inside the tool itself, switched on by default and working on everyone at once.

The averaging reaches the mind

There is good reason to think the averaging reaches well past the surface of the screen and works its way back into the person using the tool. Cognitive scientists have a name for the underlying mechanism, cognitive offloading, which describes the way a mental capacity tends to weaken once the job it used to do has been handed to a device. Pocket calculators dulled the ability to do arithmetic in the head and navigation apps have quietly eroded the sense most people once had of how the streets of a city connect to one another. 

Generative AI tools carry this further than any device before them because the faculty now being handed over is the act of making itself. It is the very material from which judgment is built, so that the tool shapes the output and then, in the same motion, reaches back to absorb the effort that would have shaped it, leaving the part of a person that once did that work to soften from disuse.

From there, the loop closes on the person almost completely. What makes anyone's work recognizably their own is a private accumulation of odd particulars. The things they happen to have seen and the taste they still carry over years of getting things wrong- that accumulation is drawn upon only when someone makes a thing themselves. But it is passed over entirely when they accept what the machine offers instead. When one leans on the average for long enough, the person stops generating deviations.

The heaviest version of this cost settles on whoever is only now arriving in the field. The designers who came of age before these tools built their taste by the slow and graceless route, producing some thousands of mediocre things and working out, gradually and often painfully, why each of them was mediocre. People did not enjoy that stretch and it was also where a sensibility was forged, because the quirks that eventually harden into a personal style are found only in the friction of doing the work badly and repairing it by hand. 

A newcomer today can skip that friction almost entirely, since the tool delivers competent work on the first attempt and accepting it is the intelligent, exhausted, wholly reasonable thing to do.  Those designers begin on a floor no earlier generation ever stood on, while the ceiling they learn to aim toward is the tool's ceiling, and the tool's ceiling is the average. Newer designers may therefore never develop the particular muscle that can feel the average as average and choose deliberately to leave it. Having never once been pinned beneath it long enough to want out, and since a field renews its supply of deviation only through people who learned somewhere how to deviate, a generation taught instead to accept the middle will thin that supply.

What raises all of this above a complaint about taste is what happens to the output once it exists. The work these models produce does not simply disappear after it has been used. It pours outward, onto websites and into applications and across every LinkedIn post and social feed, and that flood is precisely the material on which the next generation of models will be trained. Today's average, thus, becomes part of tomorrow's raw ingredients, and tomorrow's model learns a picture of the world already pulled toward the center by whatever its predecessors made. When a model is fed a sufficient quantity of its own kind of output, a now well-documented thing happens to it. The distribution it has learned grows narrower, the rarest material at the edges is the first to vanish, and the whole system drifts harder toward its own mean with each successive generation, its quality and its variety draining away together.

Culture, left to itself, tends to replenish from those edges. When someone makes an odd thing, the odd thing catches on, and it becomes in time a new region of the map that everyone else can build upon. A system trained largely on its own averages has no such source of replenishment and ends by consuming its own tail. And design is where the effect will show itself first, because design operates at enormous scale and turns over quickly. Subsequently, its edges can be watched thinning very nearly in real time with a machine now sitting in the loop and hastening the collapse by feeding steadily on its own reflection.

Where the correction has to come from

The comforting thought is that this will pass the way every earlier panic about a flattening technology has passed. Culture has survived every tool that was once supposed to level it. The printing press was meant to standardize thought and ended up multiplying it beyond recognition. Similarly, photography was going to be the death of painting and instead handed painting its freedom. And synthesizers and drum machines were each written off in their turn, and each went on to widen music the moment a restless few worked out how to push it somewhere the inventors had never pictured.

If we look closely at how any of those rescues actually happened, it was by the people who recovered and never the tool in and of itself. The variety came back because someone brought judgment the instrument did not have, a taste that ran ahead of what the machine could manage on its own. That is the uncomfortable part, because judgment is precisely what a tool reaches for from the first blank screen quietly wears down. And so, the correction, if it comes, will not arrive by itself but from the way people decide to use the thing, and from a stubborn refusal to give away the one faculty that has always done the redeeming.

That faculty is earned the slow way, over the long apprenticeship that turns a fluent hand into an expert one by sharpening against resistance, when a hard limit blocks the easy move and forces an unfamiliar one, which is why constraint has always been a quiet engine of anything genuinely new. 

A tool built to erase both the effort and the limits leaves neither one intact. There is a well-documented price for handing a skill to a machine, which is that the skill fades in whoever stops using it the way it has faded in the pilots and physicians whose instruments came to do their thinking for them.

So the useful response is to do by hand the one important part that actually matters, which is the deciding. That means using the machine to reach a competent draft in seconds and then treating that draft as the start of the work rather than the end of it. It is turning it down often enough to stay able to tell the good from the merely acceptable, inventing the constraints the tool is so eager to dissolve, and guarding the slow and unrewarded practice that is the only thing keeping a person's taste ahead of the average it serves back. None of this is efficient, and almost everything about an ordinary working day argues against it.

That, oddly, is where the hope sits. As the average becomes free and instant for everyone at once, the only thing left with any real worth is the judgment to want more than the average can give. That judgment cannot be downloaded or prompted into being, only earned. 

Whether enough people will still have trouble earning it, now that the passable answer is always waiting and always free, is the question these AI tools have raised and cannot begin to answer themselves. A whole generation is quietly answering it already, mostly without noticing there was a question at all.

The Best Use of AI Award is for teams that have meaningfully integrated AI into their workflow or product. As a change in how they work, make decisions, or build better products. This year’s final three show three different ways AI can make a real difference: 

• Intercom R&D, for making agent-first development part of how 500 people build and ship
• Vodafone Ireland, for using AI to accelerate design thinking and move teams from assumptions to validated decisions
• PageOn, for moving from Figma handoff to AI-native product delivery

Each case study stands out because AI was used to change the workflow. One helped an R&D organisation ship significantly more. One helped product teams turn research and alignment into 90-minute workshops. One changed how designers move from idea to production.

Vote here: Take me to the LinkedIn Poll

Intercom R&D

In June 2025, Intercom’s CTO set a clear goal: double R&D productivity within 12 months. The metric they used was merged Pull Requests per R&D employee across 500 people. Instead of treating AI as an optional tool, Intercom made agent-first development an operating principle.

Engineers shifted from writing code in IDEs to directing agents. The team first set a target of 80% agent-driven PRs within six weeks, then raised that target to 95%. The team used AI in several parts of the workflow. Designers started shipping code directly to production, beginning with small copy and CSS changes before taking on more complex work. Researchers and analysts could self-serve data through internal tooling built on Claude Code, instead of waiting in analysis queues.

One of the biggest shifts was in code review. PR reviews had become a blocker, so Intercom built an AI approval system, starting with lower-risk changes before expanding. Now, 93.6% of PRs are agent-driven, and 19.2% are approved entirely by AI with no human reviewer. Auto-approved PRs merge in around 15 minutes, compared with more than an hour across the wider organisation.

The impact of these changes was significant. After nine months, the organization hit 2x the productivity goal, and across 16 months, they reached 3x and were still climbing. The defect backlog dropped by 54%, the median time from idea to shipped fell by 39%, downtime from breaking changes dropped by 35%, and the cost per PR halved.

Vodafone Ireland

Vodafone Ireland had a different problem. Design thinking was seen as too slow, research synthesis took weeks, and teams were often making decisions based on assumptions instead of evidence. As a result, 25 to 26 key digital KPIs were blocked.

Vodafone Ireland introduced an AI-accelerated design thinking framework to reduce the time between insight, alignment, and action. Instead of relying on multi-week discovery cycles, teams used custom GPT agents to bring together fragmented inputs such as analytics, stakeholder knowledge, and previous research. The AI helped generate early personas, problem framing, opportunity areas, solution directions, UX copy, interaction patterns, and testable concepts. Every output was treated as a hypothesis and then validated through real user testing, usually with 12 to 20 users per project.

As a result, Vodafone Ireland achieved a 77% login conversion rate, 91.7% task completion on key journeys, increased clarity scores, and 30+ teams adopted the methodology. The framework helped unlock 25 to 26 previously blocked KPIs. This changed both the speed and quality of decision-making. Research synthesis and alignment moved from weeks into 90-minute workshops, which helped teams ideate faster, align earlier, and validate more often.

PageOn

PageOn shows how AI can change the relationship between design, prototyping, and production. Before that shift, even small feature work usually took around five days to move from design to release. Most of the work happened in Figma and then passed over to engineering through handoff, which created delays, rework, and limited validation of how the product would actually behave.

AI changed that workflow by moving key parts out of static mockups and into AI-assisted, runnable environments. The team used tools like Claude to explore interaction ideas faster than traditional wireframing, then gradually moved closer to the real product, where flows, states, and constraints could be tested more realistically. As the codebase became more modular and AI-friendly, designers were able to go beyond prototyping and make production-ready UI changes, with engineering review still in place.

That shift made a big difference. For smaller features, design could often be completed in one day and shipped the next. Therefore, Figma dropped from around 80% of the workflow to about 5% for this type of work as more design activity moved into runnable environments. This way, the team could explore more alternatives, test ideas faster, and reduce the risk of experimentation. If a variation did not work, it could be adjusted or replaced quickly. As people experienced realistic versions of features earlier, feedback became more concrete and decision-making became faster.

Why these three matter

These three finalists show that AI is most useful when it changes real behaviour. Intercom used it to scale delivery. Vodafone used it to speed up evidence-based decision-making. PageOn used it to remove friction between design and shipping.

The public vote counts for 25% of the final score, and now it’s over to you. The winners will be announced live on stage at UXDX EMEA 2026 in Berlin, where we’ll celebrate the teams turning evidence, research, and design into measurable outcomes. Voting closes in 7 days, so make sure your vote is counted.

Vote here: Take me to the LinkedIn Poll

*Get your tickets here to see the winners live: https://uxdx.com/berlin/2026/tickets/

The Unblocker Award is for teams that successfully removed organisational, structural, or process barriers that slow cross-functional collaboration down, and can show the difference it made. This year’s final three are:

• Delivery Hero - Cape Multibrand Design System: Connecting 10 Design Systems at Global Scale
• PLUS - Ending Redundant Work on a Rotating Design Team
• Heineken International - Unblocking Delivery Through Product Leadership Alignment

Each one solved a different kind of blocker: one tackled fragmentation across systems and brands, one reduced knowledge loss during team rotations, and one removed the delivery friction that was stopping a team from shipping predictably.
What connects these three is that they did not just improve one workflow. They changed the system around the work, and that is what made the difference.

Vote here: Take me to the LinkedIn Poll

Delivery Hero

Delivery Hero’s customer support and help center experiences had grown across a huge global ecosystem: 10 design systems, 35+ themes, 17 products, 5 service types, 9 brands, 25+ languages, and 70+ countries. That scale created a serious problem. Users saw inconsistent experiences when moving between a host app and support journeys, and internally, designers and engineers kept re-solving the same problems instead of building on shared foundations.

The team’s goal was to reach Level 2 of the Seamless Experience Scale across customer support and help center products. That meant creating a consistent visual and interaction foundation across products, with shared tokens for typography, colour, spacing, border radius, icons, and illustrations. Instead of debating consistency in abstract terms, the team created a clear framework that made progress measurable.

The biggest change was structural. They did not replace the existing design systems. Instead, they introduced the Cape Design System as a multibrand orchestration layer that connected them. That gave teams shared rules, shared priorities, and shared accountability, while still allowing local flexibility where needed.

The result was a move from Level 1 to Level 2 within 6–12 months, with the foundation now spanning 10 design systems and 35+ themes. That meant a more consistent, scalable, and usable experience for customers, and far less duplication for the teams building it.

PLUS

PLUS had a different kind of blocker. Its non-profit product team depended on around 15 part-time student designers, but those students rotated every 4–5 months. That meant knowledge kept disappearing. Every new cohort had to relearn the design system, product context, and team conventions, which slowed onboarding and created rework for developers too.

The team set a clear success criterion: cut onboarding from 4–6 weeks to under 2 weeks. They also wanted to reduce repeated AI context setup, improve design system health, and unblock adjacent roles like product managers and developers.
To solve this, they built plus-uno, a GitHub-native system that combines a component library, a prototyping workspace, and a structured knowledge base for AI-assisted design workflows. The important change was not just the tool itself, but the way knowledge now persists. Conventions, product context, and design decisions are stored in structured documents, then loaded automatically at the start of each session.

That changed the workflow from “rebuild context every time” to “start with shared knowledge already in place.” The impact was clear: onboarding time dropped from 4–6 weeks to within 1 week for new designers, design system coverage improved, and two product managers were able to handle minor design tweaks that used to be routed through designers.

Heineken International

Heineken International’s team was struggling with delivery predictability. On average, they were completing only 47% of their sprint commitment, and work was often started but not finished. The problem was not team capability. It was a system problem: unclear priorities, weak backlog structure, missing review loops, and conflicting signals from different stakeholders.

The team set out to increase delivery predictability, protect sprint execution, improve backlog quality, restore stakeholder reviews, and onboard new team members without losing momentum. They also needed to keep the team stable during a wider company reorganisation, which added more pressure.

The fix was not a local process tweak. The team changed how work was owned and managed. They introduced strategic alignment, a single voice on priorities, better backlog templates, protected sprint commitments, consistent reviews, and stronger onboarding. In other words, they removed the blockers around the team rather than asking the team to work harder inside the same broken system.

The result was a jump from 47% delivery predictability to 100% of sprint commitment in under 6 months. That happened even while the team was going through turnover and organisational uncertainty, which makes the improvement even more telling.

Why these three matter

These finalists show that the best way to unblock a team is not always to add more process. Sometimes it is to create shared standards, preserve knowledge properly, or fix the decision-making system around delivery.

They also show that barriers can look very different. One is fragmentation across products and markets. One is knowledge loss during team rotation. One is delivery friction caused by misalignment and weak structure. But in each case, the team removed the thing that was slowing everyone down.

The public vote counts for 25% of the final score, and now it’s over to you. The winners will be announced live on stage at UXDX EMEA 2026 in Berlin, where we’ll celebrate the teams turning evidence, research, and design into measurable outcomes.

Voting closes in 7 days, so make sure your vote is counted. Vote here: Take me to the LinkedIn Poll

*Get your tickets here to see the winners live: https://uxdx.com/berlin/2026/tickets/

The Impact Award is for teams whose evidence directly drove real outcomes, including user adoption, retention, and conversion. This year’s final three tell three very different but equally strong stories: 

• All human & Power NI, for redesigning Help & Support to drive digital self-serve
• De Gruyter Brill, for unifying two academic publishing platforms into one clearer experience
• LiveEO, for helping utility teams make smarter vegetation management decisions and reduce outages

What connects these three case studies is that each team used evidence to make clear decisions, and those decisions led to measurable change. One reduced pressure on customer support. One improved confidence in a complex purchase journey. And one helped prevent real-world outages by giving teams clearer risk signals.

Vote here: Take me to the LinkedIn Poll

All human & Power NI

This case study comes from All human’s redesign of Power NI’s Help & Support experience. Power NI’s Help & Support area was getting around 21,000 visits a month, but the experience was pushing too many customers towards the most expensive support channel: calling. Customers were coming to the website to complete high-intent tasks, such as moving home, switching provider, or submitting meter readings. But instead of finding the right online journey, many were ending up on the Contact page.

The team could see that the navigation was organised around internal categories, not customer intent. Important self-serve tasks were buried, and the Contact page mixed emergencies, sales, and service queries in a way that made it harder for people to find the right path.

To understand the issue properly, the team used three evidence sources: heuristic analysis, tree testing, and analytics plus call data. Tree testing showed that only 59% of users could find what they were looking for, while 20.6% ended up on the Contact page. Analytics also showed that users who clicked the phone number were 7x more likely to exit the site.

That evidence helped the team avoid a complete redesign of the website. Instead, they focused on the journeys that were creating the highest call volumes and made those tasks easier to find. They made self-serve options easier to find, lowering the visual priority of contact routes, and added better tracking so they could understand what users did next. The result was a major shift from calling the company towards self-serving. An increase of 138% of self-serve logins, Contact Us submissions dropped by 29%, and phone clicks dropped by 14%.

De Gruyter Brill

After the merger of De Gruyter and Brill, two academic publishing platforms needed to become one. The challenge they faced was that researchers, students, and librarians were dealing with inconsistent navigation, unclear access status, and a checkout flow that created too much friction. The platform needed to feel like one coherent experience without losing institutional access or weakening trust.

The team used three main methods for their research: moderated usability testing, analytics analysis, and stakeholder interviews and workshops. Across three rounds of moderated usability testing with 75 participants, the team found several points of friction. One important insight was that hover-triggered menus were causing mistakes, so they moved to click-triggered navigation. Analytics helped identify where users were dropping off, while stakeholder workshops with Editorial, Sales, Customer Service, and Marketing helped the team understand content priorities and business constraints.

After implementing the changes, the results showed movement across the success criteria. Checkout drop-off fell by 21%, homepage engagement rose by 32%, and views and engagement on subject landing pages increased by 15%. Just as importantly, the redesign helped create a more unified experience for a global academic audience.

LiveEO

LiveEO’s Treeline product was built for vegetation managers who need to prevent outages before they happen. When vegetation is not managed in time, power lines can fail, which can lead to outages, safety risks, wildfire risk, and major costs for utility providers and communities. Before the redesign, data was spread across different layers and users had to manually compare static snapshots, switch between systems, and rely heavily on experience to decide which lines needed attention first. The goal was to move teams from fixed-cycle planning to insight-led planning, where risk, urgency, and context could be understood in one place.

LiveEO ran continuous research across North America, Europe, Australia, and Japan. They combined interviews, usability testing with early concepts and AI-based prototypes, and direct field observation. Seeing vegetation managers in their real working environment showed how much they were relying on workarounds, including multiple phones, printed maps, to-do lists, and local knowledge.

The redesign brought risk, urgency, and workload into one clearer view, helping vegetation managers see which lines needed attention first. This meant teams could move away from fixed-cycle planning and make decisions based on actual risk. Over 12 to 18 months, customers saw a 40% reduction in outages, alongside cost-per-mile improvements.

Why these three matter

These three finalists show different kinds of impact while having the same pattern. Each team used evidence to make sharper decisions, avoided guesswork, and tied their work to a measurable outcome.

They also show that impact does not always look the same. Sometimes it is more self-serve completions, sometimes it is lower checkout drop-off, and sometimes it is fewer outages. But in every case, the research led to a change that mattered.

The public vote counts for 25% of the final score, and now it’s over to you. The winners will be announced live on stage at UXDX EMEA 2026 in Berlin, where we’ll celebrate the teams turning evidence, research, and design into measurable outcomes.

Voting closes in 7 days, so make sure your vote is counted. Vote here: Take me to the LinkedIn Poll

*Get your tickets here to see the winners live: https://uxdx.com/berlin/2026/tickets/

Here are the speaker announcements shared in March for UXDX 2026. Speakers from Siemens, Taxfix, Product People, McKesson, Electronic Arts, Amazon, and Llewyn Paine Consulting are taking on the issues teams are dealing with right now. From AI adoption and healthcare UX to design systems, cross-functional alignment, and clearer decision-making, there is plenty here for teams looking for practical ideas they can apply straight away.

UXDX EMEA 2026

David Sward from Siemens will share what it takes to build one design language across highly complex industrial products with very different foundations and lifecycles. In “Solving the Design System Problem When Products Live for Decades,” he will unpack Siemens’ move from multiple specialised systems to a more unified approach, and explain why one tech stack is not always the answer. Expect practical lessons on legacy products, acquisitions, open source design assets, and how to build a system that can evolve over time.

When teams try to become AI-native, the hard part is not the tooling but changing how product, design, and engineering actually work together. In “Supercharging the Product Cycle: Practical Lessons in AI-Driven Product Development,” Timo Ilola, VP of Product Design at Taxfix, shares how the company is embedding AI across the product cycle. Expect a grounded look at mindset, team upskilling, and how roles change when AI shifts the focus from delivery to more strategic impact.

Pooja Dey from Sage will explore how product marketing and UX testing can do more than support go-to-market. In “From Buyer Journey to Product Investment: How Product Marketing and UX Testing Shape Product,” she will show how teams can use buyer journeys, onboarding insights, and in-life usage to shape product decisions earlier. Expect practical lessons on prioritisation, risk reduction, and building stronger feedback loops between product, UX, and marketing.

The Founder of Product People, Mirela Mus, will lead an interactive session on how to spot AI-native opportunities worth building. In “AI First Product Strategy Workshop: Find and Prioritise AI Native Opportunities That Ship,” she will show how teams can move beyond using LLMs for efficiency and start identifying where AI can create real product value. Through three case studies, Mirela will share practical ways to evaluate and prioritise AI-first bets that teams can act on straight away.

Kristina Gibson will lead a hands-on workshop for teams that feel stuck waiting for perfect data before making a strategic move. In “Do you feel like your team is waiting on perfect data to validate a strategic change?” she will share a practical framework for exploring new opportunities, evaluating bets, and turning product strategy into roadmap action. Expect useful guidance on aligning stakeholders, assessing new directions quickly, and moving beyond incremental experiments.

Rahul Balu (Senior UX Designer at Amazon) is taking the final wildcard slot for UXDX EMEA 2026 in Berlin with his talk: “From Opinions to Outcomes: How Clarity Builds Trust and Better Design.” His session tackles a familiar problem for many teams: design work does not usually stall because of a lack of talent, research, or effort, but because different functions are working from different assumptions and speaking in different terms. Rahul will share practical ways to build a shared design language that helps teams move from debate to decision-making, align around outcomes, and build trust across stakeholders without losing the judgement that good design depends on.

UXDX USA 2026

When AI moves into healthcare, the question is not just what it can improve, but where it needs limits. In “AI in Healthcare UX: From Burden to Breakthrough,” Anya Gerasimchuk, Senior Director of UX/UI Engineering at McKesson, draws on the company’s three-year AI transformation across oncology and multispecialty care. The session explores where AI can genuinely help in healthcare, alongside the guardrails and constraints that shape responsible adoption.

Electronic Arts’ Andra Bond and Olivia Lucas will take the audience inside the work behind EA’s shift to experience-led roadmaps. The session “How EA Unified Product, Design and Ops with Experience-Led Roadmaps” looks at how experience Atlas and evolution mapping helped connect design, research, data science, and operations around shared outcomes. Expect practical lessons on journey mapping, measurement, and reducing friction across identity, loyalty, and support.

As Notion scaled from personal users to global enterprises, Head of Design Randy Hunt saw how quickly traditional structures can start to slow teams down. With “Kill the Org Chart: Building Skill-Based Design Teams at Notion,” he will unpack how the company reshaped its design organisation around skills and behaviours rather than titles and functions. Expect practical lessons on building more adaptable teams, preserving creative craft through scale, and creating the conditions for ownership and speed.

When nearly half the team leaves, the problem is bigger than hiring. In “When 45% of Your Team Walks Out: Inside Wise’s Design Culture Reboot,” Josh Payton from Wise shares how the company rebuilt its design organisation after a period of burnout and constant firefighting. Expect practical lessons on rebuilding trust, redesigning workflows, and creating a culture built on autonomy, balance, and impact.

Llewyn Paine (VP Innovation Strategy & Operations from Llewyn Paine Consulting) will take a strategic look at what happens when products are tested by AI agents instead of humans. In “Strategic Readiness for AI Agents: A Workshop on Agent Experience (AX) for Product Leaders,” she will explore what those breakages reveal about commercial risk, operational readiness, and the wider gaps teams need to address. Expect practical takeaways on assessing the Agent Gap and coordinating the cross-functional response needed to close it.

Why These Topics Matter

These talks are not about future promises. They are about what teams are dealing with right now: making AI useful, fixing complexity, aligning across functions, and building products that can keep up with change. If you are trying to move faster without losing trust, clarity, or quality, these are the kinds of sessions worth paying attention to.

Secure your spot now:

• USA tickets: https://uxdx.com/usa/2026/tickets/
• EMEA tickets: https://uxdx.com/berlin/2026/tickets/

Catch up on earlier UXDX 2026 speaker announcements here:

The UXDX 2026 Speaker Lineup Is Here (January Edition)

We’ve been locking in speakers for UXDX 2026, and honestly? This lineup might be our strongest yet. We’re bringing together people who are actually building and shipping products at scale. Great speakers from Booking.com, N26, monday.com, Dashlane, and Netlify who are dealing with the same messy rea…

UXDX

UXDX 2026 Speaker Updates (February Edition)

Here are the speaker announcements shared in February for UXDX 2026! With AI already on the roadmap, the question is whether you can ship it without losing trust, getting stuck in approvals, or watching quality slip between design, research, and engineering. Speakers from AstraZeneca, Miro, Pinteres…

UXDX

Three new awards celebrate teams who can demonstrate a clear link between evidence, decision-making, and measurable results.

UXDX is proud to announce the launch of The Unblockers: a new awards program launching at UXDX EMEA 2026 in Berlin. Built for cross-functional teams, The Unblockers recognises real-world work that demonstrates a measurable chain from evidence to decision to impact.

Unlike other industry awards that prioritise presentation quality or brand visibility, The Unblockers is built around practical case studies from which the wider community can learn. UXDX evaluate entries on the strength of evidence, the quality of decision-making, and the clarity of outcomes.

Three Award Categories

• Impact Award
  For teams whose evidence directly drove real outcomes, including user adoption, retention, and conversion. We want to see a clear, documented connection between research or data and the business result it produced.
• Best Use of AI Award
  For teams that have meaningfully integrated AI into their workflow or product. Not as an experiment, but as a genuine change in how you work or what you were able to build.
• The Unblocker Award
  For teams that successfully removed the organisational, structural, or process barriers that slow cross-functional collaboration down; and can show the difference it made.

Recognition and Rewards

Winners will be announced live on the UXDX EMEA main stage in Berlin. All finalists will receive direct feedback, a Finalist badge for LinkedIn, professional photos taken, and exposure to the wider UXDX community. Top entries will additionally receive team tickets to UXDX EMEA 2027, a published case study, and sponsored gifts.

How to Enter?

The entry process begins with a short expression of interest form, designed to take no more than three minutes to complete. No presentation deck is required at this stage. Teams whose entries show strong potential will be contacted with the next steps for a full submission.

The deadline for expressions of interest is 10 April 2026. Submit your quick entry now here to participate.

Here are the speaker announcements shared in February for UXDX 2026! With AI already on the roadmap, the question is whether you can ship it without losing trust, getting stuck in approvals, or watching quality slip between design, research, and engineering. Speakers from AstraZeneca, Miro, Pinterest, AccessiBe, Allied Solutions, and Product Discovery Group are sharing what it takes to keep momentum when the model changes and the pressure ramp up. They will get into how to stop accessibility falling through the cracks and prevent “transformation” from turning into busywork.

UXDX EMEA 2026 Berlin Speakers and Workshops

Daria Tarawneh, Head of Enterprise Design at Miro, is tackling the reality of enterprise AI, the work that determines whether anything ships. Her talk: “You Can’t Just Turn It On”: The Hidden Work Behind Enterprise AI" is going to break down how to build the foundations that keep teams moving. This is about building a pace that holds up, without turning delivery into a waiting game or a constant reset cycle.

When the model changes halfway through the build, most teams slow down or spin out. In “0 to 1 Product Orchestration: Building AI as it Changes the Way We Work,” Haydyn Phillips from AstraZeneca shares a case study on how to keep delivery moving anyway. This session gets into: orchestrating research, design, analytics, and engineering around a shared AI context and an adaptive delivery rhythm, while defining success beyond productivity metrics.

Rina Volovich and Haim Repael Azoulay from accessiBe lead a hands-on workshop: "AI & Accessibility in Practice: Closing the Design-to-Code Gap." Accessibility often gets lost somewhere between Figma, handoff, and production. This workshop is built to fix that. You’ll see an end-to-end workflow using AI tools, and leave with a clear mental model for building accessibility into design, engineering, and deployment.

UXDX USA 2026 New York Speakers, Talks and Hackathons

Over in the USA, Larkin Brown (Senior Director of Product Research at Pinterest) is coming to UXDX USA 2026 to show how Pinterest keeps humans in the loop while shipping GenAI at scale. In “Bridging Human Insight and AI: How Pinterest Builds Visual Experiences That Connect”, she shares a rare behind-the-scenes look at how Pinterest teams are reshaping visual search, shopping, and creation with AI.

At EMD Digital, Paul Svoboda leads UX strategy and design while working on the product side. Most enterprise transformations get trapped in meeting cycles and slide decks. Paul is bringing a different model from EMD Digital, where change is treated like a product and shipped as software, so better ways of working become the default through use.

If you want to stop talking about AI and start building with it, Jim Morris is your session. Jim, from Product Discovery Group, is bringing “AI Prototyping for Non Engineers: Demo + Hackathon” to New York, and it is designed to get you from idea to working prototype fast. You will watch a live build, then jump into a hands on hackathon to create your own interactive prototype and share what you learned with the room.

Our UXDX USA 2026 wildcard goes to Benjamin Hewett (Director of UX at Allied Solutions) with “Never Done: Evolving UX teams who earn influence.” This is for leaders who know a UX team is never finished, it either evolves or loses ground. Benjamin will share how teams earn trust in the moments that matter, design their own growth, and build influence inside the messy reality of real organisations.

Why These Topics Matter

These are not talks about what might work someday, but what is working now, what is failing, and what teams are learning in real time. If you are trying to ship AI responsibly, keep accessibility from slipping, protect research quality, and build teams that can adapt without burning out, this is exactly the work you need to hear about.

Secure your spot now:

• USA tickets: https://uxdx.com/usa/2026/tickets/
• EMEA tickets: https://uxdx.com/berlin/2026/tickets/

We've been locking in speakers for UXDX 2026, and honestly? This lineup might be our strongest yet. We're bringing together people who are actually building and shipping products at scale. Great speakers from Booking.com, N26, monday.com, Dashlane, and Netlify who are dealing with the same messy realities you are.

UXDX EMEA 2026

Romain Berthomé runs product at Booking.com, where he manages 80+ people spread across continents. His talk, "Scaling Impact: Building High-Performing Product Teams Across Borders," gets into the practical side of distributed teamwork; How do you maintain alignment when your team spans time zones? How do you balance autonomy with direction? If you've ever struggled with a team that's growing too fast or spread too thin, this one's for you.

Marcus Knight from N26 is tackling something every team claims to care about, but few actually fix: the design-to-engineering handoff. "Zero-Blocker Delivery: Closing the Design–Engineering Gap (for Real)" isn't about just the theory. This talk is about the friction points that slow teams down and how to actually eliminate them. Plus, he'll cover how AI fits into the design process without creating more problems than it solves.

Then there's Tanya Adlam from monday.com, who's been in UX research for over 20 years, including time at Meta's Reality Labs. Her talk, "When AI and Teams Blur the Lines: Who Owns the Research?" addresses a question a lot of teams are quietly grappling with: when AI can generate insights, what's the role of dedicated researchers? She'll share frameworks for keeping research quality high even as the tools and team structures evolve.

UXDX USA 2026

On the other side of the pond in the US, we've got Frédéric Rivain, CTO at Dashlane, presenting "AI Without Access: Shipping Intelligence When You Can't See User Data." Dashlane's entire business model depends on never seeing user data, so how do they integrate AI? It's a fascinating constraint that forces creative solutions, and Frédéric will walk through how they've approached it.

Dana Lawson, CTO at Netlify, is speaking about what might be the most provocative talk of the conference: "Are We Designing for Humans or for Agents?" As AI agents become more common, are we still designing experiences for people, or are we optimizing for bots? It's an uncomfortable question, and Dana's not going to shy away from it.

Why These Topics Matter

These aren't aspirational talks about what might work someday. They're about what's working now, what's failing, and what people are learning in real time. If you're managing teams, shipping products, or trying to figure out where AI actually fits into your workflow, this is worth your time.

Secure your spot now:

• USA tickets: https://uxdx.com/usa/2026/tickets/
• EMEA tickets: https://uxdx.com/berlin/2026/tickets/

"Agile’s original purpose was not cultural enlightenment. It was building better software. Working software over documentation. Faster, more reliable delivery."

And that is exactly why Jay thinks we are stuck.

Because most teams have become proficient at shipping. They did not necessarily get good at delivering value.

Jay threads two quotes through his talk to reset the room’s focus. The first, via Marty Cagan, is the reminder that the best companies do not just deliver software. They deliver value that changes customer behavior. The second, from Jez Humble, is even more explicit: winning is about outcomes, not outputs, solving real problems for real people.

This is where Jay lands his critique. Agile helped teams learn to ship features quickly and reliably, but many organizations turned that capability into the goal. Speed became the metric. Shipping became the winner. Teams started optimizing for output, and once you are in that loop, it is very easy to mistake activity for impact.

He throws in the number that many people in the industry quote: somewhere between 50 and 80 percent of features fail to create meaningful customer value. He is careful about the range, because it varies by company, but the direction is clear. Even when delivery is flawless, value can still be missing.

Jay’s argument is not that Agile failed. It is that Agile solved a different problem, and now we have a new one.

Why this matters to him

Jay has been building digital products for close to two decades, and for the last ten years, he has been working in the messy overlap where research, design, and agile delivery collide. He has been trying to help organizations shift from feature factories into learning organizations. The case he shares is one of his more recent transformations, and he frames it with the line everyone knows, even if no one can prove who said it first: culture eats strategy for breakfast.

He treats it less like a meme and more like an operating constraint. You can have a strategy. You can have OKRs. You can have a beautiful roadmap. If the culture rewards output, you will get output. If the culture punishes uncertainty, you will get certainty theater. If the culture treats “build this feature” as the entire job, the team will stop asking why.

When Jay joined his organization about a year and a half ago, that was the reality. There was “the business,” and there was “tech.” Business requested features. Tech delivered them. The core measure of success was predictable: did we ship on time?

No questions asked.

Purpose that actually connects to the work

Jay is clear that the team's purpose was not imposed top-down. They built it through co-creation. The wording they landed on is strikingly human: solve people’s problems through collaboration and a human-centered approach to create a sustainable world for us and future generations.

He calls out something many companies miss. Corporate missions often exist at a level that feels abstract to delivery teams. In their case, the company's mission was to be carbon neutral by 2035. That is a strong goal, but a customer service product team can struggle to see how their daily work connects to it. A team-level purpose creates a bridge between the corporate ambition and the practical work.

The values they chose were equally telling: empathy for customers and coworkers, curiosity about market and users, collaboration, trust, innovation, and impact. These are not decorative words in his story. They were chosen to correct real gaps. If trust is missing, naming trust as a value is a way to make the absence discussable.

A structure that makes outcomes a shared responsibility

Jay then moves from purpose to structure. The old setup was siloed: business had its goals, product had its metrics, design and tech had their own measures. In that environment, everyone can succeed locally while the customer experience fails globally.

So they reorganized around a cross-functional value stream. At the core sat a product manager, engineers, and a designer for customer-facing products, aligned in the same team. Around them sat an extended team: data analysts, solution architects, developers, and DevOps. Above, but not separate, sat business stakeholders like marketing, sales, and customer experience.

The key change is not the org chart. It is the ownership model. They planned quarterly OKRs together. They stopped measuring design success separately from business success. Metrics like CSAT, churn reduction, conversion increases, operational efficiency, and cost-to-serve became shared team outcomes. Jay’s subtext is that you cannot ask teams to care about outcomes while measuring them on output. The measurement system is the culture.

Ways of working that prioritize problem framing

Jay is careful with the word “process.” He knows people flinch at it. But he argues that process is necessary, and in this transformation, process was the bridge between purpose and execution.

Their approach starts with problem framing and opportunity framing. Not “what feature should we build next,” but “what problem is worth solving,” “how big is it,” and “what would value look like.” He describes their method as a tailored, opportunity-solution-tree style approach that helped the team quantify problems and avoid the trap of simply pulling the next ticket.

Then he introduces the discovery delivery loop they implemented. Discovery and delivery were not treated as separate phases that throw work over a wall. They ran together, with cross-functional involvement throughout. Discovery itself had three phases: insights, ideation, and validation. Data analysts and researchers brought quantitative and qualitative input. The team translated insights into hypotheses. Those hypotheses were validated through research and experimentation. The output of the system did not feature. It was learning.

Jay repeats this idea because it is the hinge of his whole talk. If you want to innovate, the differentiator is what you learn about your customers, how quickly you learn it, and how widely that learning spreads.

Redefining success so people can actually change

Jay knows that culture change does not happen on a quarterly deadline. If you set aggressive business KPIs and expect instant outcomes, you create panic. Panic drives teams back to the old habits: ship more, faster, just to prove progress. So they redefined success in a way that reduced fear. He uses a SpaceX line he loves: every rocket failure gets us closer to Mars. It is not a celebration of failure for its own sake. It is a celebration of learning that reduces future risk.

They made a simple equation explicit: learning equals success.

Success could come from shipping a feature, but only if they measured it and understood its impact. Success could come from a research effort that disproves an assumption. Success could come from an experiment that failed, as long as the learning was captured and shared.

To make that real, they rewarded knowledge sharing. People were encouraged to present learnings in all hands, including the failures, not just the wins. They also rewarded the process through performance management: how many discoveries did you initiate, how much research did you trigger, what did you validate or invalidate?

In other words, they rewarded the behaviors that create outcomes, not just the deliverables that look like progress.

What rewarding failure looked like in practice

In the Q&A, someone asks the question everyone is secretly thinking: what does rewarding failure actually look like?

Jay’s answer is refreshingly concrete. They offered small rewards for people who presented a failure in all hands, because people were hesitant to stand up and say, “We tried this, and it did not work.” Once they signaled that sharing a failure was valued, people started doing it, and other teams avoided repeating the same mistakes.

It is a small intervention, but it changes the social risk. When it becomes safe to share what did not work, learning scales.

Balancing learning with quality

Another question challenges the learning narrative: what if teams pursue learning at the expense of quality for users?

Jay pushes back on the false tradeoff. Learning does not mean chaos. Teams still ship features. The difference is that they do not ship and immediately sprint away to the next backlog item without measuring, reflecting, and adjusting. If you do not measure, you are not learning. If you do not learn, you are just producing.

His point is subtle: the enemy is not delivery. The enemy is unexamined delivery.

The results he could share

Jay says he cannot show all the numbers, but he does share one meaningful outcome from the transformation. They had a financial goal to reduce the operational cost and bring down the cost to serve customers. After piloting the approach, particularly with the customer service team using their problem framing and discovery delivery methods, they exceeded the target by 28 percent.

Jay uses that result carefully. It is not “look at the magic process.” It is “this is what becomes possible when teams have purpose, shared outcomes, and the cultural permission to challenge requests and learn fast.”

The real takeaway: outcomes are a culture, not a KPI

By the end, Jay’s story lands in a place that feels both obvious and hard.

You cannot KPI your way out of an output culture.
You cannot roadmap your way into curiosity.
You cannot demand outcomes from teams you measure on shipping.

The shift from output to outcome is cultural. It starts with a shared purpose that feels real at the team level. It becomes durable when the structure aligns incentives around shared goals. It sticks when teams have a repeatable way to frame problems and validate solutions. And it survives when success is defined as learning and impact, not only delivery.

Jay ends with a practical invitation. Pick something small from his playbook: co-create a team purpose, align OKRs cross-functionally, measure outcomes together, celebrate learning publicly, and reward the behaviors that create insight. Cultural change is a long game, but it is not abstract.

It is built, one reinforced behavior at a time.

Want to watch the full talk?

You can find it here on UXDX: https://uxdx.com/session/from-output-to-outcome-driving-cultural-change-in-product-teams/

Or explore all the insights in the UXDX EMEA 2025 Post Show Report: https://uxdx.com/post-show-report

"When you use ChatGPT day to day, are you really using one model, or is there an agent behind it?"

He does not ask for an answer in the room. He wants the question to stick because the rest of his talk is about what happens when enterprises stop treating generative AI like a demo and start treating it like production software with real users, real data, and real consequences.

Mahmoud is a Lead AI Engineer at Mastercard, working in the unglamorous, high-impact zone where models meet reality: deployment, scaling, governance, and keeping systems safe. His message is not that large language models are magic. It is that shipping them responsibly is hard, and the hardest parts are rarely the model.

Why enterprises struggle with classic ML

Mahmoud starts with a quick rewind. For two decades, machine learning and deep learning have delivered wins across classification tasks: image diagnostics, forecasting, fraud detection, and more. The problem is that enterprise environments are different.

Enterprises run on unstructured data. Mahmoud quotes a common estimate that 80 to 90 percent of enterprise data is unstructured. That is the heart of the dilemma. Traditional ML often wants labeled datasets and clean pipelines. Labeling at enterprise scale is slow, expensive, and politically messy.

Generative AI changes the shape of the problem. If you can take unstructured data, ground a model in it, and generate useful answers, you suddenly have a path to value that does not depend on labeling everything first. This is the promise that has pulled so many companies into the LLM wave.

But Mahmoud draws a sharp boundary between promise and production.

Stop obsessing over tomorrow, worry about today’s risks

He calls out a pattern he sees everywhere: people jumping straight to AI doomsday narratives about the future. Mahmoud’s view is more immediate. The biggest problem is not that AI will build a distant future. The biggest problem is the risk it creates right now.

In enterprises, the threats are not abstract. They are compliance risks, data exposure, governance gaps, and operational failures. He references a Nature piece that argues for shifting attention away from speculative catastrophe and toward present-day harms and controls.

This framing matters because it changes what teams prioritize. If you think the main issue is the future, you debate philosophy. If you think the main issue is today, you build safeguards.

The misconception that LLMs began with OpenAI

Mahmoud also tackles a myth that distorts how leaders make decisions. Generative AI did not begin with OpenAI. Language models existed decades earlier. What OpenAI changed, in his telling, was the interface and usability. The leap was not only the model. It was making the system conversational and accessible enough that anyone could “talk to it” naturally.

In enterprise settings, that usability shift is double-edged. The easier it is to use, the easier it is to misuse. A chat interface can become a shortcut to sensitive data if guardrails are weak.

The real stack: model, environment, software layer, infrastructure

Mahmoud lays out the major layers you have to think about when deploying LLMs:

The foundation model, whether you build, fine-tune, or adopt open source.
The experimentation environment, where teams test and iterate.
The infrastructure, often GPUs, and serving capacity.
And the software layer that holds everything together.

He emphasizes that enterprises can often acquire models and hardware. Open source platforms and model hubs make it easy to start. Cloud and procurement budgets make infrastructure attainable. The hard part is the layer in between: integration, orchestration, user-facing reliability, monitoring, and the plumbing that connects models to the right data and the right controls. Without that software layer, you do not have a scalable product. You have a brittle experiment.

He reinforces this with a familiar idea from the ML world: only a small percentage of the work is the model itself. The rest is data governance, pipelines, observability, security, and operations. In his own Mastercard experience, he says the LLM is only a tiny fraction of the workload.

The “closed book” problem: why raw LLMs fail in enterprise

Mahmoud then zooms in on what happens if you try to use an LLM without grounding it in an enterprise context. He describes a “closed book” approach where the model answers from its training knowledge, plus the prompt. That approach breaks down quickly in real organizations.

He lists the failure modes his team has to manage:

Hallucination. Confident answers that are not true.
Attribution. The need to explain how a response was produced, especially in risk-oriented domains.
Model staleness. The fact that models age fast as better ones appear and contexts change.
Revision and deletion. When data rights change, you must remove data and potentially retrain or rebuild.
Customization. The explosion of domain-specific needs across business units.

That last one is a common enterprise trap. If every domain needs its own model, you quickly end up with dozens of parallel deployments that are impossible to maintain. So the obvious response is the one the industry has rallied around.

RAG: adding memory outside the model

Mahmoud describes the shift toward coupling a model with an external memory, most commonly Retrieval Augmented Generation. Instead of asking the model to “know” the domain, you retrieve relevant enterprise knowledge and feed it as context. This can reduce hallucination and improve relevance.

But he is careful not to oversell it. RAG introduces its own complexity, and he hints at the kinds of design questions that become engineering headaches:

How should you chunk content?
How long should chunks be?
How do you structure retrieval?
How do you keep context fresh?
How do you prevent leakage across users and teams?

RAG is not a silver bullet. It is a trade. You swap “model knows everything” for “system retrieves the right things,” which turns model deployment into distributed systems engineering plus governance.

This is where his talk becomes very enterprise-specific.

The rise of agents and the danger of building them for everything

Then Mahmoud pivots to what he sees as the next evolution: agents.

He distinguishes three stages:

Single LLM is the basic prompt response pattern that many teams used in 2022.
Workflows, where you send a task across multiple models or steps and aggregate results.
Agents, where the system reasons, acts, uses tools, and loops through feedback.

He warns against treating agents like a trend. Do not build agents for everything. Keep it simple. The more autonomous the system, the more risk and operational overhead you inherit.

To help teams decide, he offers a simple mental checklist. You should build an agent when the task is complex and multi-step. If it is not complex, use workflows. If the task is complex but not fully doable, reduce the scope. If the task is complex and doable but the cost of error is high, put a human in the loop.

In a company like Mastercard, that last condition is not optional. It is the default. Humans remain part of the system when errors carry legal, financial, or trust consequences.

Think like your agent

Mahmoud’s most memorable teaching moment is his “think like your agent” example. He describes a simple task: searching for “Mastercard Global.”

A human opens a browser, waits, types, clicks, and reads. An agent does the same thing through tools: click, type, navigate, retrieve, and respond. The point is not that this is impressive. The point is that it demystifies agents. They are models using tools in a loop, not mystical intelligence.

That framing is useful for product teams, too. If you can describe what a human would do step by step, you can design the agent’s tool access, constraints, and evaluation criteria. If you cannot describe it clearly, you probably are not ready for agent autonomy.

Multi-agent collaboration and the protocols that enable it

Mahmoud looks beyond single agents toward multi-agent systems, where specialized agents collaborate like a team of experts, coordinated to solve complex tasks. He describes roles like a mathematician, statistician, and AI specialist working together, with a coordinator in the middle.

To make that possible, systems need standardized ways to connect agents to tools and agents to each other. He references two emerging protocol ideas:

Model Context Protocol, a unified way for models to connect to external tools and data sources.
Agent-to-agent protocols are a way for agents to communicate with other agents.

His message is that these approaches complement each other. One helps connect an agent to the world of tools. The other helps connect agents to each other.

It is also the moment where the host draws an interesting parallel: we spent years preaching cross-functional collaboration in human teams, and now we are building similar dynamics into our AI systems.

Mahmoud does not over-philosophize it, but the echo is there. Enterprise AI is not only about smarter models. It is about orchestrating a system of capabilities, just like product building has always been.

Mahmoud ends on a question a reviewer once posed to him: after all the risks and challenges, will LLMs and agents help us, or will we help them?

Want to watch the full talk?

You can find it here on UXDX: https://uxdx.com/session/transforming-enterprise-ai-with-scalable-llm-deployments

Want to make your career AI-proof? Make sure to read our new ebook: https://uxdx.com/ebook/career-compression/

The question is not whether design still matters. It is whether the story we tell about design has started to work against us.

Berlin, Boom Cycles, and the Same Old Questions

Pamela opens with warmth for UXDX being in Berlin, then goes straight into the state of the industry. She has lived through enough cycles to recognize the rhythm. A surge of enthusiasm. A hiring boom. Then the crash. Then the soul searching.

And the soul searching always sounds familiar. Why are we still misunderstood? Why are we undervalued? Why do we still fight for a seat at the table?

Her point is not that these questions are wrong. It is that asking them again, in the same way, will keep producing the same outcomes. Instead, she wants to change the narratives designers bring into the next cycle, because the next cycle will come whether we feel ready or not.

Where We Are in the Cycle

Pamela sketches the current landscape in practical terms. Most tech companies did fine during the COVID boom, then got swept into the 2022 hiring frenzy when money was cheap, particularly in the United States. Then came the crash.

She references the flatline many people have felt in their job searches. Since around 2023, tech job openings have stayed pretty flat. Layoffs are slowing, but the market is still saturated. There are simply more people looking than there are roles available.

Then she adds a detail that stings because it matches lived experience. An analysis of layoffs across dozens of companies found that product designers and product managers are significantly more likely to be laid off than engineers. Research is hit even harder.

Pamela does not linger on the numbers to scare the room. She uses them to make a different point. If your function is seen as optional in the hard moments, you cannot solve that with better portfolio presentations. You solve it by changing what the organization believes your work actually is.

AI Enters the Story, Whether We Like It or Not

Pamela is honest about the new ingredient in this downturn. AI.

Almost 80 percent of companies are now investing in AI. Many of the early wins are predictable, like customer support automation, risk and fraud, and operational efficiency. But the use case she highlights for product development is content creation, including translation for some organizations.

In Pamela’s world, expectations are already being set that content creation could be fully automated within a year. She has opinions about that, but her bigger point is more strategic than argumentative.

If AI changes how content is produced, it changes the product development cycle. Every discipline will have to evolve how it works. Design is not the only function that will be reshaped. But design is the function currently trapped in the most fragile narrative, and that makes it vulnerable.

History is Repeating, Quietly and Regularly

Pamela steps back and gives the room a longer view. Boom and bust cycles are not rare. They are recurring. Since 1850, there have been dozens of these cycles. They tend to last seven to ten years, sometimes shorter, like the one we just lived through.

The booms tend to be triggered by innovation or regulation. Financial shifts. New platforms. New distribution. The dot com era. The post 2009 rebound. The long stretch of growth driven by mobile, cloud, and digitization.

Then the bust comes, not because the innovation disappears, but because enthusiasm gets ahead of reality.

This is where Pamela’s tone becomes quietly optimistic. If these cycles are predictable, then the goal is not to avoid them. The goal is to build a career and a discipline that can ride them without losing its identity every time the market tightens.

How Design Evolved, and How We Got Cornered

Pamela traces the evolution of design roles in tech. Graphic design moved to interaction design. Interaction grew into experience and service design. The dot-com explosion multiplied titles and specializations. Then, more recently, it all compressed again into the modern product designer.

At the same time, product development processes kept evolving too. Agile and its derivatives. Design thinking as a bridge to business. Lean startup. Design sprints. Scaling frameworks. Endless attempts to make delivery faster and more predictable.

And somewhere in that acceleration, Pamela argues, design commoditized itself.

The modern model often puts one designer inside a “pizza team” working in two-week sprints. That designer is expected to cover UI, UX, motion, design systems, pixel perfection, sometimes even code, while also working ahead for the next sprint, supporting the current sprint, and QA’ing what just shipped.

Pamela calls it insane. Not as a dramatic flourish, but as a diagnosis. The workload pushes designers toward what can be produced quickly and defended easily. The UI layer. The surface. The happy path. The MVP.

Tools like design systems and Figma make speed possible, but speed can also become a trap. When design is constantly squeezed, it regresses into decoration. And once design is framed as decoration, it becomes easier for organizations to imagine that AI can replace it.

That is not a technology problem. It is a story problem.

Narrative One, Stop Selling “Polish” and Start Reclaiming Problem Solving

Pamela’s first narrative shift is blunt. Design cannot keep presenting itself as the UI layer.

She calls design a discipline with deep scientific foundations, even if designers do not always name them. When designers build solutions that work, they draw from psychology and flow. From linguistics and semiotics to shape meaning. From information architecture to support attention in real-world contexts. From learning theory and behavioral principles to help users recover when they get stuck.

Her point is not that design needs to become academic. Her point is that design needs to be explicit about why decisions are made.

She quotes a line she credits to Erika Hall: design is choices, not artifacts. If you reduce your work to the artifact, you invite replacement. If you articulate the decisions, you communicate value.

Then Pamela zooms in on a phrase she cannot stand, because it quietly destroys credibility. Designers describe their work as playing around.

She has heard it from thoughtful designers. She has heard it at major conferences. And she stops people when she hears it, because she knows what executives hear. They hear something simplistic and repeatable. They hear something that a machine can do faster.

Pamela’s provocation is not “stop being playful.” It is “stop framing rigor as play.” When you say you are playing around, you are helping commoditize your own discipline.

Narrative Two, Stop Being the Misunderstood Hero

Pamela’s second narrative shift targets a comforting identity many designers hold, often without realizing it. The misunderstood hero.

The story goes like this. Designers are the only true user advocates. Designers are the only ones who care. Designers are the only ones who see the real problem.

Pamela pushes back. Many decisions impact users, and many of those decisions are not owned by design. Pricing. Support models. Automation choices. Business policies. These are user experience decisions too, whether they sit in a design org chart or not.

She also challenges the phrase “good design.” Good for whom? Good for the user in one moment might be bad for the business model. Good for the business might not require good design at all. Those tensions exist, and pretending they do not exist is part of why design gets stuck arguing for purity rather than building influence.

Pamela shares an anecdote from Telefonica that captures a common leadership blind spot. In trying to bridge an internal divide, she kept offering “design can help” as the answer. A VP finally said, half-amused and half serious, “It sounds like design has the answers for everything.”

Pamela admits she felt that way. Many designers do.

But the correction is not to shrink the design’s ambition. The correction is to shift from hero to collaborator. The problems companies face are solved by the room, not by one function. The designer’s leverage comes from lateral leadership, storytelling, and communication that brings others into a shared understanding of value.

At SumUp, she notes, the written word matters. Writing forces clarity. Communication becomes part of the craft, not an afterthought.

She returns again to Erika Hall with another sharp reminder. Design can only be as user-centered as the business model allows. So the move is not to cling to “user-centered” as an identity. It is to become value-centered, and then connect value back to users in a way that the business can sustain.

Narrative Three: AI Is Not a Winner-Takes-All Game

Pamela does not deny the anxiety. Content is under pressure. Engineering is being accelerated by AI tooling. Designers see engineers experimenting with interface generation and wonder what happens next.

Her response is not denial, and it is not fear. It is discernment.

Design has strengths that matter in an AI-shaped workflow. Judgment. Decision making grounded in human behavior. The ability to frame problems. The ability to evaluate tradeoffs. The ability to steer, not just produce.

But Pamela also refuses the fantasy that one function will “win” AI. This is not a winner-takes-all game. It will be about how disciplines collaborate, how they adopt tools, and how they rethink the system of product development together.

If the next boom is driven by AI, then the teams that thrive will be the teams that treat AI as a shared capability, not a design threat and not an engineering toy.

Want to watch the full talk?

You can find it here on UXDX: https://uxdx.com/session/design-is-not-dead-but-the-narrative-might-be/

Or explore all the insights in the UXDX EMEA 2025 Post Show Report: https://uxdx.com/post-show-report

Two years ago, Mihaela Draghici stood on the UXDX stage in Dublin and told a hopeful story. Her teams at Volkswagen Digital Solutions had taken deliberate action to break down the silos between product teams and the business departments they serve. Cross-functional teams were finally being brought into conversations about vision, strategy, and outcomes, not just handed a plan and told to deliver it.

A Talk About People, Not Tools

Mihaela starts with a disclaimer that gets a laugh and a little relief. This is not a talk about AI. It is a talk about people working together.

She gives the context quickly. Volkswagen Digital Solutions builds digital products for multiple departments across the Volkswagen Group, including procurement, production, logistics, and sales. The delivery model is cross-functional product teams, designers, engineers, product managers, data roles, and other technical specialists, working alongside business experts and domain specialists.

On paper, it is the model everyone wants. In practice, it still breaks.

Her focus is not on the first moment silos appear. It is on what happens after you do all the right things to remove them, and still find the organization drifting apart again.

The Original Problem: Being Asked to Deliver Someone Else’s Plan

Mihaela describes the earlier friction that pushed them to change. Business leaders set plans in executive committees. Product teams would then be presented with those plans and expected to deliver. That structure left little room for negotiation, little room to adapt based on data, and little room to apply user-centered or outcome-driven thinking.

From the product side, it was hard to show progress in a meaningful way and hard to explain why certain practices mattered. Prototyping to validate ideas. Setting up analytics to understand actual usage. Running discovery to make sure the team was solving the right problem. All of it can look like a delay when the organization is in a project delivery mindset.

From the business side, the pain was real, too. Stakeholders wanted clearer alignment on product decisions and a better understanding of what product teams were doing. They also wanted product teams to understand their own constraints and realities better.

Both sides were meeting. They were planning. They were updating each other. And yet, over time, they were still working in isolation.

The Fix: Hybrid Teams With Shared Ownership

The solution Mihaela’s teams introduced was structural. They reconfigured teams into what they call hybrid cross-functional product teams, meaning business stakeholders were integrated into the product team rather than kept adjacent to it.

The intent was not symbolic inclusion. It was shared ownership and shared responsibility for success and failure. Shared vision. Shared tools. Shared ways of working. Business experts and domain specialists were included throughout the product lifecycle, not only at checkpoints.

To make that practical, they created a working model agreement that defined responsibilities across each stage of the cycle. They included business stakeholders in team ceremonies. In return, product people gained access to high-level steering committees and management groups where they previously had no seat.

Mihaela also talks about pairing as a deliberate lever. The engineering teams already worked in pairs through extreme programming, so they extended the concept to other roles. Pairing PMs with business product owners. Designers with business stakeholders. Designers with engineers. Creating regular contexts where people from different functions work side by side and exchange knowledge continuously.

The aim was closeness. Trust. Empathy. Shared context instead of handoffs.

What Improved When People Worked Together

Mihaela explains how they checked whether the change was working. They ran team health checks, interviews, and surveys to capture qualitative feedback. They looked at product success, whether the products were used and created the expected impact. They also looked at investment patterns, whether leaders were willing to fund continued improvement and future initiatives.

Over the last two years, there were real wins. Multiple products were rolled out across sales, procurement, production, and logistics. Continuous discovery practices were introduced in several teams. Experiments were led by teams, not forced from above, and they produced business value.

One example is wonderfully concrete. For a product built for factories, the team paired with an IT specialist working on site and built a tool that simulated a machine connected to the factory. It created a small testing lab in the office, allowing the team to experiment incrementally before investing in a full solution. It is the kind of work that only becomes possible when domain expertise is inside the loop rather than requested from afar.

The feedback Mihaela describes is encouraging. People in teams felt more confident about understanding the problems they were solving and the users they were solving them for. Business stakeholders were engaged and responsive, contributing insights that improved solutions. And business colleagues appreciated how engineers, architects, and designers made an effort to understand why procurement works a certain way, or why a factory process exists the way it does.

So far, it sounds like a success story.

And then Mihaela delivers the twist that makes the talk matter. Even with those changes, collaboration still fell apart over time.

Why Collaboration Falls Apart Even When You “Fixed” It

Mihaela’s key insight is that drift is not a mystery. It is the result of forces that quietly pull teams back to old patterns.

She groups the causes into three areas.

Hidden structural forces pull people back to previous behaviors. Those behaviors create habits. Habits become mindsets. And mindsets are slow to change, especially when the organization does not notice they are reappearing.

The second problem is language. People often do not speak the same language, even when they believe they do.

The third is what happens when expectations and perceptions diverge. Misalignments create communication gaps, and communication gaps lead to lower motivation, lower engagement, and in some cases, loss of trust.

Mihaela then makes the reasons tangible through examples.

The Structural Forces That Undo Good Intentions

Mihaela starts with processes, not because she loves bureaucracy, but because processes are powerful. They are often necessary. And they can still create friction.

One example is budgeting cadence versus product planning cadence. Business departments approve budgets annually, sometimes over three or five years, often in waterfall style with preset milestones and requirements that are expected not to change. Product teams plan quarterly and want space for experimentation and iteration.

The result is predictable. Business leaders want commitments. “We already promised this to management.” Product teams want to stay adaptive. Both are rational within their systems, and the collision can quietly break collaboration. Another structural force is regulation and operational constraints, especially in an industry like automotive with strong limitations around production and logistics. Those constraints must be understood early, or teams will design solutions that cannot exist.

Mihaela also points to role ambiguity inside a complex organization. When responsibilities are unclear, people duplicate work or compete for the same outcomes without realizing it. And decision-making becomes slow because of steering committees and layers of governance. Product teams want speed. The system is built for careful consensus. Friction grows.

Over time, these structures reintroduce the project mindset. “Here is the list of features, deliver them, move on.” That mindset may never be stated as policy, but it reappears through the gravitational pull of governance. Then there is a quieter issue: time and availability. Hybrid collaboration asks business colleagues to dedicate significant time to product work. At first, enthusiasm is high. Over time, the effort becomes harder to sustain, especially when people have their “real job” pulling them back.

Finally, team rituals can lose meaning. Meetings stay on the calendar even after they stop helping. People attend because attendance is expected, not because it drives outcomes. When rituals become performative, collaboration starts to rot from the inside.

The Strategies That Keep Teams From Drifting Apart

Mihaela ends with optimism, but it is practical optimism.

First, keep investing in pairing. Diverse pairs create shared context and reduce the “us versus them” dynamic.

Second, stop assuming shared understanding. Build a shared glossary that teams can update and revisit. Make it easy to contribute, not a document that becomes stale.

Third, use rotations and work exchanges. Send people into each other’s worlds. Spend time in procurement. Spend time in factories. Move between locations. Shared reality builds empathy faster than any workshop.

Fourth, check in regularly. Revisit the working model agreement when the product lifecycle changes and when team members change. Revisit role expectations often. Do not treat the original agreement as valid for years.

Fifth, adapt structures to context. Do not assume one collaboration model fits every department. Use processes in your favor, and tune them to the realities of the domain.

Mihaela’s final framing is the one that ties the whole story together. The best cross-functional teams are not the teams without collaboration problems. Those do not exist. The best teams notice problems sooner and address them faster.

That is the difference between cross-functional in name and truly collaborative over time.

And then she ends with a question back to the room, the one that turns her sequel into an invitation. What collaboration challenges are you living with right now, and what are you doing to address them?

Want to watch the full talk?

You can find it here on UXDX: https://uxdx.com/session/from-cross-functional-to-cross-purposes-why-collaboration-falls-apart-over-time/

Or explore all the insights in the UXDX EMEA 2025 Post Show Report: https://uxdx.com/post-show-report

Tasha Melchior walks on stage with the kind of calm you only get after you have lived it enough times that you stop counting. Seven acquisitions. One major merger. A company story that keeps rewriting itself, while the product team is still expected to ship, keep customers happy, and somehow feel “aligned.”

The Board Game Version of M&A Is the Fun Part

Tasha introduces herself as VP of Product at Everway, and she makes a point of sharing her non-traditional path. She was a history teacher before she moved into software and product, and she has worked across industries before circling back into EdTech. She is also the UXDX Ambassador in Copenhagen, which explains the energy she brings when she invites people to come speak at meetups.

Then she shares a personal detail that turns into a metaphor for the entire talk. She loves board games. Especially one called Acquire, a game where you invest in companies, grow them, and watch them swallow each other through mergers and acquisitions.

The game captures the excitement of M&A, the strategic wins, and the feeling of being on the right side of a deal. But, Tasha says, the real-world version has far more friction than the board game box ever mentions. When M&A happen to your company, your product team does not experience it as a clean transaction.

It feels like the rules changed mid-play.

Everway, A Mission That Makes the Chaos Worth It

Before Tasha gets into the messy part, she grounds the room in what Everway actually does. The company was formed through the merger of Texthelp and n2y, two similarly sized organizations of roughly 300 employees each, now combined into a team of around 600.

Everway builds software that helps people understand and be understood. The scale is bigger than many people realize. Products delivered to over 250 million people across around 150 countries, supported by teams spread across 10 global offices. Tasha is based in Copenhagen, and her own product team spans Scandinavia, the UK, the US, and even Melbourne. Scheduling alone is a weekly puzzle.

The mission is rooted in a belief that every mind is unique. Traditional one-size-fits-all learning and work systems leave people behind, especially neurodiverse learners. Tasha explains that many learning differences still carry stigma, even though people who are neurodiverse often develop strengths that neurotypical systems fail to recognize.

Everway’s expanded strategy reflects something Tasha sees as obvious once you say it out loud. Students grow up. They leave school. They reach university and work and lose support. The company now frames its approach as “k to gray,” kindergarten until you turn gray, supporting people across life stages, not only in classrooms.

She offers a metaphor that makes the mission feel tangible. Assistive tools should become like eyeglasses. Once bulky, stigmatized, and only for people who truly needed them. Now normal, stylish, and even worn by people who just like how they look. That is the future she wants for learning and communication support. Necessary for some, useful for all.

The Holy Trinity That M&A Loves to Disrupt

Now Tasha pivots to the behind-the-scenes story. She frames her leadership lens as three areas she constantly monitors: people, product, and process.

For people and culture, she watches for empowerment, clarity, and whether the environment helps the team thrive.

For the product, she watches strategy clarity, alignment, and whether what the team builds will truly be “hired” by users to get their jobs done.

For the process, she is unapologetically enthusiastic. She believes repeatable, measurable, scalable processes are what enable teams to deliver reliably. She borrows a line from her CEO, Craig. Process is like the frame of a bicycle. If the frame is broken, you can pedal as fast as you want and still go in the wrong direction.

Then she asks the question that shapes the rest of the talk. What happens to people, products, and processes when you throw mergers and acquisitions into the mix?

The Scale Tips, and Everyone Storms

Tasha uses an analogy that is almost uncomfortably accurate. A friend once told her that a family is like a scale, more or less balanced. When a new member joins, the scale tips and everyone loses their footing. It takes about a year to rebalance.

In Tasha’s experience, every acquisition is like adding a new family member. The scale tips. The team enters the storming phase, the part of Tuckman’s model that no one romanticizes. Suddenly, there are new colleagues, new identities, new ways of working, and often a new story about what “good” looks like.

Tasha shares an example from outside Everway that still maps perfectly. A CTO friend described a merger that looked like a perfect match at first because the values sounded aligned. Then reality arrived. Differences in culture and process surfaced fast.

Everway has seen similar tension, amplified by the fact that some acquired companies had competed with each other for years. People had built their professional identity on saying, loudly, why the other product was worse. Then, overnight, those people become colleagues, under a new brand name, expected to collaborate as if the past did not exist.

Tasha does not sugarcoat it. Change requires emotional agility, not just skill. It asks people to unlearn, let go of comfort, and operate without the steady waters they are used to navigating.

The Product Portfolio Problem Nobody Can Ignore

M&A create a growing product suite, and Tasha shows how quickly that becomes unsustainable. Each product is tied to revenue. Each has loyal users. Each has customers who do not want to hear the word “sunset.”

But sales teams cannot sell a sprawling set of overlapping tools forever. Brand consistency, UI consistency, and a cohesive suite become essential, and none of that happens automatically.

Tasha gives a concrete example through literacy toolbars. Everway’s Read and Write toolbar supports diverse and multilingual learners, with features like read aloud, screen masking, dictionaries, and word prediction across platforms.

When Everway acquired Don Johnston in the US, those products included CoWriter and Snap and Read. The decision was to transition customers into Read and Write, but to do it without making users feel like they had been dumped into a foreign interface. That meant integrating features and even underlying language engines into the core product, and redesigning the experience so users could move smoothly between different display modes.

The end goal is not only feature parity. It is consolidation, sunsetting products carefully while protecting trust.

In Sweden, the approach differed. Three acquired companies had been selling three different literacy toolbars. Instead of choosing a winner, Everway built a new “super product,” designed with Scandinavian minimalism and simplicity in mind. Since launching in May last year, active usage has grown significantly, and feedback has been strong.

Tasha’s underlying lesson is that consolidation is not a single decision. It is a long sequence of design, technical, and change management choices, all made while the team is still responsible for keeping the current customers happy.

When Your Best Process Gets Replaced

Process is where Tasha’s story becomes painfully relatable. Before the merger, her team had built a roadmap process that stakeholders loved. Simple, clear, and scalable.

They kept all product roadmaps in a single slide deck. For each product, the PM maintained four key views: deliveries from the past six months, the next two quarters, an annual timeline, and a top ten ideas list. The roadmap used a now-next-later structure and included an out-of-scope frame so stakeholders could see what was truly shipping versus what was being worked on.

Each item is linked to a one-page “hero slide” that explains the what and why, how success would be measured, and is linked to marketing materials and relevant UI imagery. Sales and customer success used these slides with customers. Engineers used the deck for context. Leadership lived in it.

Then the merger happened. New CEO. New CPO. New preferences. New way of working.

The company moved to Productboard. Tasha admits she resisted. Her thinking was simple: if it is not broken, do not fix it. But she also acknowledges the reality. Standardizing processes across a bigger org matters, and new leaders need a system that scales for them too.

The point is not that Productboard is good or bad. The point is that M&A often replace a working process with a different one, and even process people like Tasha still have to move through resistance before acceptance.

The Real Requirement Is Emotional Agility

Tasha names the concept she wants people to take away. Mergers and acquisitions require not only agile development, but emotional agility.

There is friction. Personality differences. Silos. Sometimes chaos. Sometimes loss of expertise because not everyone stays through the transition. It can feel like an emotional rollercoaster.

To cope, she uses an emotions wheel. Identify what you are feeling. Accept it. Do not judge it or fight it. Then step back, lift yourself out of the fog, and choose actions that align with your values.

Her closing takeaway is honest. It is messy. It is hard. It is also exciting, and it teaches you constantly. Over time, balance returns. And even though the scale will tip again with future deals, resilience grows in the process.

If your product team is in the middle of an acquisition right now, Tasha’s talk offers one clear reassurance. Feeling off balance does not mean you are failing.

It means the family just changed, and you are doing the work of becoming one team.

Want to watch the full talk?

You can find it here on UXDX: https://uxdx.com/session/the-real-impact-of-mergers-acquisitions-on-your-product-team/

Want to make your product career AI-Proof? Download our free ebook: https://uxdx.com/ebook/career-compression/

Donal O’Mahony has heard the phrase “once in a lifetime shift” so many times that it has lost its meaning. The dot com crash. Flash is dying overnight. A multi-billion-dollar acquisition that rewired a company’s culture. Now, generative AI is arriving with enough speed and noise to make even seasoned leaders feel like their roadmap has been stolen in broad daylight.

Donal steps on stage as Vice President of Experience at Contentful, ending the conference with a deliberately optimistic note. Contentful has a Berlin origin story, he says, and there is something fitting about talking in a cinema, in a city built on reinvention.

The talk title, “Who Moved My Roadmap?”, is a wink at anyone who has heard, “The roadmap has changed again.” In a room full of product and development leaders, that line lands instantly. Donal’s first move is to remove the surprise. Change is always happening. It is always going to happen. If you are waiting for a stable period where priorities settle down for good, you are waiting for something that does not exist.

Instead of treating every reset like a crisis, Donal wants teams to build the muscle to navigate it, without burning themselves out or losing momentum. The goal is not a perfect plan. The goal is direction and the ability to adjust without panic.

The Cheese Is Gone, and People React Exactly as You’d Expect

To explain why change feels so personal, Donal brings in a small book that helped him through upheaval at work and at home. It is called Who Moved My Cheese. He jokes that he is dyslexic and recommends the audiobook, but the point is serious. The story is a parable about four characters who find their cheese in the same place every morning, until one day it is gone.

Some characters return to the empty spot, convinced the cheese will come back if they wait. Some freeze, terrified of the maze. Some nibble the crumbs, trying to pretend that what is left will be enough. And some move, explore, and adapt.

Donal’s reason for sharing it is simple. People are hardwired to respond differently to disruption. When the roadmap shifts, the reactions you see across your team are not random. They are predictable. And the uncomfortable truth is that the response you choose shapes what happens next.

He is also careful to say this is not about forced positivity. He jokes about his hairline as evidence that stress is real. The message is more grounded. These moments will become your portfolio pieces. You want to look back on them with some fondness, not only as periods where you were stressed for longer than you needed to be.

Seismic Shift One: When Flash Died Overnight

Donal’s first story goes back to an era that now feels distant, but at the time felt like the future arriving. He joined an Irish company that was moving through a reverse takeover into Houghton Mifflin Harcourt, with an ambitious mission to digitize educational content for e learning.

Then the world changed in a chain reaction. The iPad arrived and instantly made tablets feel like the obvious platform for learning, especially for young children. The opportunity looked enormous. Interactive textbooks, 3D models, videos, science brought to life, all of it suddenly felt possible at scale.

And then Flash began to die.

Steve Jobs published his thoughts on Flash, calling out issues like security and performance. Whatever was happening between companies behind the scenes, the impact was clear. Flash, which had powered so much interactive content across education and media, was on borrowed time. Boardrooms that had invested heavily in Flash-based content woke up to a new reality. Everything had to be converted. Tools, workflows, and assumptions had to change quickly.

Inside teams, Donal saw the full spectrum of reactions. Some people saw opportunity, a chance to pivot from print into digital design, to build new skills, to shape a new craft. Some mourned the loss of Flash, then later thrived when they moved into newer ecosystems like Unity. Some played the familiar role of the person who lists everything that cannot be done, and why it will never work.

But the group Donal warns against is quieter. It is the people who see the change coming and do nothing. They keep building in the old way after hours. They keep using the old tools out of habit. They wait for the cheese to return.

Donal calls this becoming an OSA team, an “oh shit asteroid” moment, where you can see the comet coming and still carry on as normal. His advice is not to shame anyone for fear. Fear is natural. His advice is to notice it and then choose something else.

He shares a personal tactic. When a negative thought shows up, counter it with several positive ones. It can feel forced, but it stops fear from becoming your default. And he offers a better question than the one many teams use under pressure. Instead of “what’s the worst that could happen?”, try asking, "What’s the best that could happen?”

For Donal, that question led to real outcomes. He built a UX team. The work moved from static content toward user-centered digital experiences. One of the apps connected to the company’s work even landed in Apple’s App Store Hall of Fame. For a parent with young children, that was not just a career milestone. It was meaning.

Seismic Shift Two, A 2.4 Billion Reset

Donal’s second chapter jumps into telematics, joining Fleetmatics, an Irish success story that had floated on the New York Stock Exchange and had serious scale. He saw a mission to build user centricity and grow UX maturity, and he remembers being invited to UXDX around that time to talk about his plans.

Then the story takes a turn that anyone who has lived through an acquisition recognizes instantly. Donal returns from a family holiday and steps off a ferry to find several missed calls from his boss and the founder. No signal for hours. The kind of silence where you rehearse both outcomes before you finally hear the truth.

The company was acquired for 2.4 billion. Competitors were being acquired too. Leadership shifted. Org charts changed. The culture was about to be tested.

Donal describes the reactions as the same maze again. Some people fought the change. Some wanted everything to go back to how it was, the smaller office, the founder led certainty. People resigned. Others waited, anxious about what the new world would reward.

But Donal highlights the person who set the tone: the founder, Peter Mitchell. Donal speaks about him with genuine warmth, describing him as a change agent and a people person, someone who knew names and family stories deep into the company’s growth. That leadership approach mattered because it shaped how others responded.

The best that happened was bigger than the plan Donal originally pitched. Years earlier, he had talked about building a UX team of 25, ambitious for Dublin at the time. After the acquisition, it scaled to around 80 across continents. The company shipped more ambitious work, including safety technology that used AI-driven insights before it became fashionable to label everything as AI.

Donal’s point is not that acquisitions are pleasant. It is that they can unlock scale, investment, and impact, but only if leaders model steadiness and if teams stop treating change as betrayal.

Seismic Shift Three, AI Arrives on Day Thirty

Then Donal lands in the present. He joined Contentful, attracted by an API first, open, future-focused platform. One month in, generative AI arrives with force. He jokes that it felt like being handed dynamite. For some people, it was pure excitement. For others, it was dread.

Donal reframes the moment with a question. What is generative AI generating?

Yes, it is generating content, and he laughs at the timing of joining a company called Contentful right as content generation explodes. But it is also generating fear, anxiety, stress, and uncertainty. It can generate opportunities and innovation, too. The human reaction is part of the output.

So Donal returns again to the choice. You can choose fun, even inside confusion, or you can choose fear. Fear is natural. The work is noticing it and then counterbalancing it, so it does not drive the room.

North Stars, Not Fantasy Plans

When roadmaps feel unstable, Donal suggests a practical anchor: directional North Stars. He references Jared Spool and the idea of “concept cars,” visions that point two to four years ahead. Donal likes them because they create alignment without pretending to know the exact sequence of steps.

In Contentful’s early AI phase, the team created a provocative concept video. It was meant to spark conversation and provide direction, not to claim certainty.

But Donal adds a warning. Future thinking is useless if it is done in a bubble. The concept has to be shaped by the teams closest to customers, research, and delivery. The magic happens in the two-way exchange: strategic direction informed by real user contact, and tactical learning guided by a shared horizon.

Your Mindset Is the Real Portfolio

Donal closes with a reflection that feels almost like advice to his younger self. Enjoy this moment in your career. You will look back and realize these periods of upheaval were the work that shaped you.

Then he leaves the room with a simple truth. Whether you think you can or you think you cannot, you are right.

For Donal, the future of UX and product in the era of AI is not about clinging to a perfect roadmap. It is about building direction, staying close to customers, and choosing a response to change that you will be proud of later.

Want to watch the full talk?

You can find it here on UXDX: https://uxdx.com/session/who-moved-my-roadmap-navigating-the-unpredictable-future-of-ux-and-product-in-the-era-of-ai/

Or explore all the insights in the UXDX USA 2025 Post Show Report: https://uxdx.com/post-show-report