Boagworld: UX, Design Leadership, Marketing & Conversion Optimization

This month we look at why most in-house UX teams are organized around work that no longer arrives in the way it used to. Product owners, developers, and marketers are all producing design with AI tools long before anyone thinks to raise a ticket, and telling them to stop has quietly stopped working. We talk about what to fund instead, how to measure a team that is no longer the bottleneck, and where Marcus thinks Paul has gone too far. We also cover a new tool that gives a freelancer something close to a staff team, a new free email course, and the unexpected link between bees, a man with a bin on his head, and the power of consistently turning up.App of the Month: Grok BotGrok Bot launched in early beta on 11 August 2026 from SpaceXAI and Cursor, and it is the first AI tool in a while that has genuinely surprised Paul. Rather than a single chat window, you spin up a set of agents with their own roles, so there is a chief operating officer to talk to day to day, plus a finance person, a marketer, a salesperson, and a designer. Each one has its own computer, they brief each other, and they keep working after you shut the laptop.What makes it interesting is the complete absence of setup. There are no API keys to find and no MCP servers to wire up. You tell it which service you use, it pops up a button, you log in as you normally would, and it handles the rest. If there is no connector, it falls back to a browser. Paul asked it about a client message sitting in Slack, and it read the thread, gave him advice, created a task for Monday, added a note to the relevant meeting, and then offered to write the Slack message telling the developers what was happening.For freelancers who lose hours to the admin they are least confident about, this is the closest thing yet to having colleagues. Marcus raised the obvious concerns about privacy and about becoming too reliant on output nobody checks, which is fair. Paul's answer to the second one is a quality control agent whose only job is to check everyone else's work, with the caveat that both can still be wrong.The catch is the price. It runs at $200 a month, although that does include Cursor.How the team is set up for the wrong jobEverybody is designing nowPaul used to insist that only proper UX people should do UX, and he is honest that a good part of that was protecting the job description rather than protecting the user. That position has become much harder to defend. Product owners come back from an hour with an AI tool holding a prototype and a flow that mostly hangs together. Developers generate interface options before the ticket has even been refined. Marketers build and test their own landing pages. Some of it is good, some of it is plausible rubbish, and all of it turns up before anyone thinks to involve design.The complaint Paul hears most often from in-house teams is some version of how do we stop people doing this. In our experience you cannot stop them, because going around you is faster than waiting three months for a slot. The consequences are real, they are just delayed. Somebody ships a prototype that looked fine, and six months later the accessibility has failed quietly, the conversion rate is poor, and the product looks like it was built by three different companies. By then everyone has moved on and the rework never gets attributed to the decision that caused it.Cutting the team is the expensive responseMeanwhile the money is going the other way. Headcount that leaves is not replaced, and the seat at the table turns into being asked to comment on decisions that have already been made. Cutting the team is the expensive response, because the work does not disappear, it just gets done worse and later.What to fund insteadThe organizations getting this right have moved design upstream, away from producing every screen. Fund the team to set the conditions instead of being the only route to a wireframe:A design system with real usage guidance, so a developer at 4pm can make a decent decision, or brief an AI to make one, without booking time with a designer.Playbooks for the work that keeps repeating. Landing pages are the obvious one, because marketers are already building them and they are not conversion experts, copywriters, or developers.A research repository that anyone can query, so the findings get used in the meeting six months after the readout instead of dying in a slide deck.Office hours, audits, and coaching as the service, with the hard political work still firmly on the books.How you measure the teamHow the team is measured has to change with it. If people are still judged on screens produced and tickets closed, the queue rebuilds itself inside a month, because nobody can afford to spend a week writing guidance that tanks their own numbers. Measure whether other people shipped something decent without design touching the file. Measure whether last quarter's research got referenced again. A quieter team is a sign the setup is working, not evidence that they are underemployed.Where Marcus disagreesMarcus pushed back, and the disagreement is worth listening to in full. He is happy with the developer querying a design system, and he has argued for years that a salesperson or researcher should not be drawing wireframes, because a wireframe is a series of design decisions dressed up as a diagram. His worry is that Paul is conceding ground that should be defended, particularly where the design is new rather than an application of existing rules. Paul's split is between taste and rules. Branding and anything close to art still needs a human leading, but interface and information design runs on accessibility standards, hierarchy, grids, and type scales, and those are rules a well-trained AI can follow. The important word is trained. Briefed cold, these tools produce something plausible and shallow. Given a design system, real guidance, and quality checks, the output gets good, and a designer spending fifteen minutes critiquing it is a better use of talent than two days pushing pixels in Figma.There is a caveat for agencies, which Marcus is right about. At Headscape every project starts from scratch with a new brand, so there is more taste involved and more reason to keep a designer in the driving seat. In a bigger in-house setting, you decide the level of human check based on risk. That might be every piece of work going past a designer, or spot checks weighted toward people who are new to the tools.Two things neither of us could answerTwo worries neither of us could answer. Plenty of designers love the craft of moving things around, and asking them to write documentation instead is not the job they signed up for, so this has to be handled with a lot more care than most organizations are known for. The bigger one is where the next generation comes from. If non-designers now do the work junior designers used to learn on, and the design knowledge in the tools stops being refreshed, the whole thing stagnates.Where to start on MondayIf you want somewhere to start on Monday, ask whoever runs design what people keep coming to them for week after week, fund turning one of those into something the rest of the business can use without them, and stop routing every request through the person who still says yes.If that conversation is one you are having with your own team, the UX design leadership workshop covers the same ground in more depth.Read of the MonthPaul has released a free 14-lesson email course, Become the UX practitioner your organization actually needs. It makes the same argument as this episode with the detail we could not fit into an hour, delivered as short emails every couple of days, plus the full workshop slide deck as a PDF.The written version of this month's topic is also up as Your UX team is set up for the wrong job.Marcus' JokeI just got a new job making plastic Draculas. There's only two of us on the assembly line, so I have to make every second count! Find The Latest Show Notes

Why print-first brand systems break online, and how to push back without demanding a full rebrand. I've been grumbling about this one for years, because although almost everyone now meets a brand through a screen, we're still handed identities designed for brochures, business cards, and the occasional exhibition stand, and then told to make the website behave itself. So this month I have a proper rant about it and Marcus does his best to talk me down, which he largely manages. We also get into top task analysis, which is one of my favorite research techniques, and the free app I've built because I got fed up running it the hard way. App of the Week Top Task Analysis is a free app I've built to make top task analysis considerably less painful, mainly because I'd spent years running it with a spreadsheet, a survey tool, and a fair amount of swearing. How top task analysis works Top task analysis is a technique Gerry McGovern came up with, and it works a bit like a survey except that it stops people from being greedy, because an ordinary survey hands you a long list of everything your users say they'd quite like and no sense whatsoever of what they'd actually walk over hot coals for. This forces them to prioritize, so you end up knowing which small handful of things the majority genuinely care about. Gerry's classic example is the Microsoft Office knowledge base that kept answering more and more questions and kept watching its satisfaction score slide, because the answers people actually needed were buried under a hundred they didn't. I use it for all sorts, from shaping information architecture and working out which objections are worth answering, through to prioritizing features and deciding what has earned its place on a dashboard. Roughly 80% of your users want about 20% of what's on your site, and this is how you find the 20%. What the app does The reason nobody runs it as often as they should is that it needs at least 2 rounds, one to gather the tasks and another to vote on them, so the app handles both in a single pass. You seed it with a starter list from AI, from the client, or from whichever stakeholder shouts loudest, and it genuinely doesn't matter if that list is a bit rubbish, because visitors can search it, pick the 5 tasks that matter most to them, and add anything that's missing, which then appears for everyone who arrives after them. A second screen asks them to put their 5 in order, and the back end lets you see the lot, merge the 14 different ways people phrase the same task, and tidy up anything unhelpful. And yes, there's a profanity filter, because I have met the internet before. Marcus made the good point that you want a segmentation question alongside it so you can see top tasks broken down by audience, which is already in there, and he also dug up a hospital trust site we worked on years back where we dropped the top 8 tasks straight under the main navigation. They're still sitting there 8 years later, which either says something flattering about the technique or something less flattering about how often that site gets touched. Give it a go It's free, because I did think about charging for it and then couldn't be arsed, which is not the sharpest business decision I've ever made. Have a play and tell me what's broken. The app: Top Task Analysis The step-by-step guide: Top Task Analysis: A Free App And Step-by-Step Guide Feedback: [email protected] When brand guidelines fight the web Most people meet a brand through a screen these days, and yet the web is still treated as the place where you paste in whatever was designed for print. A branding agency mocks up a homepage that has never met a real sentence, the guidelines get signed off after 18 months and roughly all of somebody's political capital, and then some poor soul is handed the job of protecting it, at which point they'll defend an unreadable contrast ratio to the death because it's sitting on page 47 of the PDF. Consistency matters and I'm not arguing otherwise, but being consistently difficult to read isn't much of an achievement. Marcus agreed with the general complaint and pointed out that Headscape has described itself as a brand interpreter for digital for something like 15 years now, for exactly this reason. He also told the story of a charity working with deafblind people whose shiny new sub-brand logo failed color contrast checks, and if I'd invented that example nobody would have believed me. A brand should serve the organization and its audience, so the moment its rules make the website harder to use, those rules need to change. What a brand actually is A lot of the confusion comes from shrinking brand down to a logo, some colors, a typeface, and a photography style, when all of that is really just the clothes the organization turns up in. The brand itself is closer to a personality, made up of what the organization believes, how it talks, and how it treats people when nobody important is watching, and your copy and your customer service will say far more about that personality than a logo ever manages. Which means we have considerably more room to move online than the brand police like to let on. Why print-first branding breaks online Print gives a designer a beautifully controlled environment and the web gives them almost none of it, which is where most of the trouble starts. Somebody working on a poster knows the exact dimensions, the exact paper, and the exact ink, whereas online you're designing for a canvas you can't see, on screens that run from a cracked phone in bright sunlight to a 32 inch monitor with the brightness turned up to painful, at whatever zoom level somebody's eyesight demands that day. Pantone certainty becomes display roulette, where pale colors wash out, dark colors go muddy, and the elegant light gray text you signed off on a calibrated screen simply vanishes on a cheap laptop in a train carriage. A poster sits still and a website refuses to Then there's the small matter that a poster sits still and a website refuses to. Websites need hover states, focus indicators, error messages, forms, navigation that collapses gracefully, and buttons that look like buttons, and most traditional guidelines have nothing at all to say about any of it, because it never occurred to anyone in the room that it might come up. Add a few thousand combinations of real content, German translations that run half as long again, and pages that grow arms and legs over 3 years, and the handful of polished examples in the brand book stops being much use to anybody. The familiar symptoms The same problems keep turning up, and once you've noticed them you can't stop seeing them. There are the walls of capital letters that slow reading to a crawl, the brand color pairings with contrast so poor you can fail them by squinting, the decorative typefaces that turn to mush at small sizes, and the logos that only really work when they're the size of a bus, although responsive logos that simplify as they shrink are a lovely solution whenever anyone can be bothered to make one. The typographic hierarchy problem The one that genuinely baffles me, and I've run into it twice in recent months, is a brand book from an actual branding agency with no meaningful typographic hierarchy in it at all, as though headings were a passing fad we'd all agreed to ignore. Then come the layouts that can't cope with content nobody wrote in advance, the complete silence on containers and components, and the fact that there's frequently no contrasting color available for a call to action, because heaven forbid anybody should click on anything. Where the cost lands Any one of those on its own looks like a small compromise you can live with, but stack them up and the brand starts shoving people away from the website it was supposed to make them love. The cost lands on accessibility, on comprehension, on conversion, on trust, and eventually on the confidence of the designers themselves, who stop challenging the daft rules and start ignoring them when nobody's looking, which is how you end up with 9 versions of the brand in the wild and not one of them right. What the attention test showed I recently tested a brand-compliant homepage against a more flexible treatment using an AI attention prediction tool. The looser version came back with roughly 20% higher predicted clarity and focus, and predicted attention on the main call to action went up by more than 80%. These are modeled predictions rather than real conversion data, so please don't quote them as gospel, but they do give stakeholders something more interesting to argue about than personal taste. I was insufferable for the rest of the day. Where Marcus and I disagree I made the case for a digital-first approach, where you start with the personality and the principles rather than the letterhead, get UX, accessibility, content, and frontend people into the room while the decisions are still being made, design actual interface elements instead of stationery, and use the website as the place where the brand gets proven with real content on a small screen before it ever reaches print. Marcus' counterargument Marcus has tried that more than once and reckons it falls apart in practice, because running the digital interpretation alongside or ahead of the main branding project leaves nobody clearly in charge, and Headscape's opinion carries very little weight while the brand itself is still up in the air. He'd rather let the branding agency finish and get everything signed off, then arrive afterward with the client already warned that interpreting it for digital will mean changes, the typeface being the almost inevitable first casualty. He also pointed out that very few branding agencies build anything, which is a fair argument for keeping the two jobs apart. Where we landed I came round to that as the more practical position, with one condition attached, which is that whatever gets handed over has to be understood as a starting point for digital rather than a finished artifact that must never be questioned by anyone with a browser. Consistency, not uniformity The phrase I keep coming back to came from Neil Eastell at the National Trust years ago, and it's consistency, not uniformity. A brand system that works online is one that's clear about what has to stay fixed and honest about what's allowed to bend, so the logo keeps its essential form while picking up responsive versions, the core colors stay recognizable while gaining accessible digital variants, and the typographic personality survives a change of body typeface for the sake of download size or legibility, as long as nobody wanders off from a serif to a sans serif and hopes we won't spot it. Think of it as a jazz standard rather than a military march, where everyone is playing the same tune but there's room to improvise around it. Questions worth asking Which parts of this identity genuinely express the personality, and which are just habits nobody has questioned? Was this rule written for screens, or copied across from print because it was already sitting in the document? What user or business outcome does it protect, if any? Does it still hold up on a small phone, at 200% zoom, in bright sunlight? What does that logo honestly look like as a favicon? How are you getting feedback on the brand from users rather than only from the board? When brand compliance and readability disagree, which one wins, and who gets to decide? The practical takeaway Whatever you do, don't open with a demand for a rebrand, because that conversation is over before it starts and you'll be the difficult one for the rest of the project. Find the specific places where the brand rules and basic usability are openly at war, fix the ones doing the most damage, measure what changes, and use that evidence to earn permission for the bigger conversation later on. And when you hit a wall, offer to test the brand-safe version against a looser one and let the numbers do the arguing for you, because a brand guardian will happily fight your opinion all afternoon but they'll rarely take on their own users. A brand exists to build recognition and trust, so when protecting the rules starts eating away at both, we're guarding the wrong thing entirely. Or, to put it the way I put it on the show, be a bit more flexible about your bloody brand guidelines. And because Marcus got his plug in twice, here is Headscape. Marcus' Joke When a TV antenna married another TV antenna, the service wasn't great, but the reception was amazing! Marcus fluffed the delivery, which if anything improved it. Find The Latest Show Notes

AI is changing far more than the speed at which designers produce work. In this episode, we talk with Josh Clark and Veronika Kindred about their book Sentient Design, how intelligent interfaces can respond to people in the moment, and why designers need to understand the character of AI before they can use it well. --- Use the code SENTIENT-BOAG to get 20% off the book through Aug 31 at rosenfeldmedia.com. --- Designing With AI as a Material Josh and Veronika describe AI as a design material, much as paint, paper, code, or the web itself can be materials. Every material has a grain. It has qualities that make some things easy and other things awkward, unreliable, or downright foolish. Designers get better results when they understand those qualities rather than forcing the material to behave like something familiar. Large language models are probabilistic. They can interpret intent, adapt tone, change formats, and produce many plausible variations, but they may also give different answers to the same question and present shaky information with alarming confidence. That makes them poor choices for some deterministic tasks, especially when a single correct answer matters. Asking one to count letters or provide an exact food-safety temperature without verification rather misses the point of what the material does well. Designers need enough experience with AI to make an informed choice about when to use it and when to leave it alone. Refusing to engage with it leaves that decision to ignorance, which has rarely been a dependable design system, despite Paul's suspiciously successful career testing the theory! The comparison with the early web runs throughout the conversation. Print designers initially approached websites with expectations shaped by paper, while the people who learned HTML and understood the new medium found different possibilities. AI creates a similar shift. Its rough edges can feel threatening, particularly when companies use it to cut costs or flatten skilled work into production, but those edges also point toward forms of interaction that were difficult or impossible before. Moving Beyond the Chatbot Chat has become the default AI interface, partly because our culture has spent decades imagining intelligent machines as talking machines. It can be useful because both the input and output remain open, but a blank text box also makes the user do a great deal of work. People must know what to ask, how to phrase it, and how to judge the resulting wall of text. Sentient Design describes 4 broader postures for intelligent experiences: Tools accept an input and return a controlled, precise output. Shazam is a familiar example. Chat uses a turn-based exchange, although those turns can involve images, interface components, or shared artifacts rather than paragraphs of text. Agents receive a goal, plan and perform the work, then return with a result. They still need direction, oversight, and review. Copilots remain quietly present, notice context, and offer assistance when useful, much like spellcheck waiting behind the scenes. These postures allow teams to choose an interaction that fits the task. A conversational box might suit exploration, while a focused tool is better for a clear transaction. An agent can handle delegated work, while a copilot can notice opportunities without demanding constant management. Josh and Veronika also share examples of AI taking part inside an existing interface. Salesforce's Generative Canvas assembles relevant components using trusted customer and calendar data. Miro's sidekicks can enter a canvas with their own cursors, while Pointer participates in a Google Doc as an editor, using the collaboration patterns people already understand. The interesting design question is how intelligence participates in an experience, not where to bolt on another chat window. Defensive Design for Uncertain Systems Trust becomes a design problem when systems are probabilistic. A generic disclaimer saying that AI can make mistakes does very little for someone deciding whether to believe a specific answer. Confidence scores often fare no better because most people have no useful way to interpret a claim such as “73% likely.” Defensive design communicates uncertainty through language, interface, and context. A system can present an answer as a signal rather than an unquestionable fact, show nearby possibilities, reveal where information came from, and give the user sensible points for review or intervention. Veronika describes “spaghetti scenarios,” borrowed from weather forecasting, where several possible paths are shown together. A search interface can do something similar by presenting adjacent questions and contrasting answers. This helps people see how wording, assumptions, and context affect the result. Tone matters too. Human beings constantly signal confidence through phrasing, body language, and shared cultural habits. AI systems tend to speak with the polished certainty of Silicon Valley, even when the evidence underneath is wobbling like a pub table with one short leg. Designers need to shape that presentation for the domain, audience, and seriousness of the decision. The pair argue that language models work well as the master of ceremonies for an experience. They can understand intent, coordinate with specialist systems, and present results in a useful format. Facts can come from trusted sources and deterministic tools, while AI handles interpretation and presentation. AI should communicate uncertainty in a way people can act on, rather than hiding it behind confidence or a disclaimer. From Doers to Directors As AI becomes part of the product, designers move toward directing systems that make design decisions in the moment. The book describes them as creative directors for systems that design themselves in real time. They define the available components, language, behavior, constraints, and rules, then guide how the system adapts them to each person's context. That shift can compress the handoffs between design, development, and product. Josh describes teams working in a common space, often a Git repository, where requirements, design guidance, and code sit closer together. AI can help each discipline contribute beyond its old boundaries, although the resulting overlap creates fresh questions about roles, careers, and management. The conversation also acknowledges the uncomfortable part. Production work is already being absorbed, and some companies assume that one senior person can manage a fleet of agents instead of developing junior designers. That approach may remove the path by which the next generation gains judgment and becomes senior. It also mistakes design for assembly. The stronger opportunity lies in behavior design, problem definition, research, creative direction, and the design of intelligent systems themselves. Designers can document why components exist, when to use them, how content should behave, and which outcomes matter. AI can then apply those rules while the designer improves the system through research and observation. Paul shares an example from his conversion optimization work. Rather than creating a single landing-page template for many audiences and campaigns, he built a component system with detailed guidance covering placement, content, accessibility, funnel stage, and use. AI could assemble a suitable page within those constraints, while the human designer had more time for research and strategy. If you are facing similar questions about moving design beyond production, the free UX Strategy email course offers a useful next step. The future of design depends on widening our view of the job, from producing screens to shaping behavior, systems, and decisions. That may be less comforting for anyone whose greatest joy is pushing pixels around in Figma, but it also brings design back to understanding problems and exploring possibilities. Sentient Design Sentient Design brings together the vocabulary, patterns, and examples emerging around intelligent interfaces. It covers 4 experience postures, 14 experience patterns, defensive design, adaptive interfaces, and the changing relationship between designers, developers, product teams, and AI. You can find Josh Clark and Veronika Kindred at Big Medium. Use the code SENTIENT-BOAG to get 20% off the book through Aug 31 at rosenfeldmedia.com. Marcus' Jokes Velcro. What a rip-off! Exit signs. They're on the way out. Find The Latest Show Notes

This month we dig into whether Claude Design is any good, why so many people feel like AI is costing them time rather than saving it, and what it really means to stop being a doer and start being a director. Along the way we wander into the loss of craft, the ethics of AI, and a joke so niche it needs its own history lesson. App of the Month Claude Design is the tool that grabbed our attention this month. It builds out designs for you, and it is genuinely impressive. We used it to rebuild the website for a small UK charity that funds children's education in India, going from nothing to a finished static HTML site in around eight hours, with Claude Design handling the design and Claude Code doing the build. Beyond the standard twenty pounds a month subscription, it cost roughly fifty quid in extra credits, which for a small organization is a no-brainer. Claude design and code together allowed Paul to create a fully working website in less than 8 hours. It turns out it does more than websites. It builds presentations too, and exports them to PowerPoint or PDF for offline editing. We put together a fifty three slide deck for a client in about two hours, work that would normally have eaten the best part of four days. Here is what we liked. It works with design systems, you can import one from Figma, you can make manual edits without burning tokens, and you can select elements visually to tweak them. The things that hold it back are that you can't export back to Figma, there's no easy publish button, and the usage allowance vanishes in what feels like five minutes flat. When you hit the wall it cheerfully suggests you try again on Sunday, which is no use when you're mid project and have already forgotten what you were doing. One word of warning. If you don't guide it heavily, Claude Design has tells, like a recurring decorative bar under the hero section that serves no real purpose. Then again, every designer has a style you can spot, so we're not convinced that's the criticism people think it is. From Doer to Director A lot of people tell us AI isn't saving them time, it's costing them more of it. That confused us at first. How can a tool that turns four days of slide work into two hours possibly slow anyone down? The more we coached people through it, the clearer the answer became, and it has very little to do with the tools. It comes down to how organized you already are. If you're not fundamentally efficient in how you work, and especially if you've never had to delegate to other people, AI exposes that straight away. The people struggling most are the ones who still want to be doers. They want to be in the code, pushing pixels in Figma, or typing every word themselves. To get real value from AI you have to shift from that doer mindset to a director one. Be the conductor, not the violinist It reminded us of the moment in the Steve Jobs biopic where Wozniak asks Jobs what he actually does, given that Woz writes the code and builds the hardware. Jobs answers that he conducts the orchestra. Woz is the finest violinist in the room, but someone has to bring all the players together. That conductor role is exactly the shift most of us need to make. Running agents in parallel A real example from this month. Working on a client presentation, we had three things running at once. Notion AI was drafting the outline in one window. Claude Design was studying the client's website to build a matching design system in another. A third agent was drafting video transcripts for a separate project entirely. Three workstreams all moving at the same time, where you would once have plodded through them one after another. That is a genuinely hard skill to build. The people best placed for it are those with management experience, because they're used to handing work off and holding several threads in their head at once. If you've never worked that way, it can feel distressing, and there's even a name for where it leads, which our reader of the month gets into. The micromanaging trap There's a design leadership parallel too. Talented designers get promoted, then can't resist sneaking back into Figma to do the work themselves. The same thing happens with AI. The agent produces something perfectly good, but it isn't quite what was in your head, so you fiddle and fiddle and fiddle, burning the very time you were meant to save. The upside is that you can't hurt an AI's feelings, so just say "no, that's not it" and move on. Get organized first The fix is unglamorous. Get organized before the agents fully take over. Build the digital playbooks, SOPs and policies we keep banging on about, so the AI already knows how you work and gets it right first time. Keep your knowledge in one place it can reference, so you're not repeating yourself endlessly. Run a task system it can see, and learn markdown while you're at it. It takes ten minutes and AI loves it. Tool or output, where's the joy? We didn't agree on all of this. Marcus prefers using AI linearly and still enjoys the doing, the writing itself, rather than conducting an orchestra. That led us into deeper water about craft. Is the joy in the tool or in the output? Paul has found real satisfaction teaching AI to write in his voice while keeping the part he actually loves, communicating ideas with passion. Marcus worries that stripping away the craft, the genuine ability to play the instrument, costs us something real, and asks how the next generation of designers will learn without the junior grind. We tested it against Monet, photography and the industrial revolution. Someone recently posted a supposedly fake Monet online and asked people to explain why it fell short of the real thing. They wrote whole essays about flow and composition, until it turned out to be a genuine Monet. So how much of our resistance is legitimate and how much is simply discomfort with change? We don't pretend to know. There are real problems with AI, the blatant disregard for intellectual property and the environmental cost chief among them, and they deserve people shouting about them. But the genie is out of the bottle, and as a species we've never once managed to put one back. The one takeaway The single takeaway. Start building your management habits now. Get organized, practice delegating, and learn to hold several threads at once, before that choice gets made for you. Read of the Month Marcus brought the counterweight to all that optimism. The article is Life with AI causing human brain 'fry', which introduces a term coined by the Boston Consulting Group. "AI brain fry" is the mental exhaustion that comes from using or supervising AI tools past your cognitive limits, the sort caused by reviewing endless AI generated code, juggling multiple assistants, and rewriting lengthy prompts over and over. It hits software developers hardest, since agents now churn out code faster than humans can review it for security flaws and overall coherence. Fittingly, the piece reached us via one of the developers at Headscape, who waved it about as proof that AI is a problem. We agree the problem is real. The article's advice is for company leaders to set clear limits on AI use to prevent burnout, though quite how they're meant to spot the issue without us telling them is another matter. The catch is that a cynical manager will just point out that their whole day already feels like relentless context switching, so good luck getting much sympathy. For a related read, Marcus also flagged AI didn't delete your database, you did, a sharp piece arguing that you should take responsibility for what you ship to production rather than blaming the AI when it all goes wrong. Marcus' Joke We end, as ever, with Marcus’ joke. This one needs a UK history lesson, so our apologies to anyone under forty. I guessed orange, but it was chocolate. I guessed toffee, but it was peanut. I guessed strawberry, but it was coffee. I was wrong on so many Revels. Find The Latest Show Notes

This week, Paul and Marcus dig into why traditional user research repositories fail almost everyone in an organization, and how AI is quietly changing the game. There's also an App of the Month pick that's a little too on-the-nose, some pointed Google bashing, and a sheep-based punchline. AI-Powered User Research Repositories The pattern in most organizations is depressingly familiar: user research gets done, a PowerPoint gets presented to stakeholders, everyone nods along or ignores it entirely, and then the research disappears. It might prompt some short-term action, but the knowledge evaporates. Nobody references it again six months later. The traditional solution has been to build a research repository: a central place to store everything from interviews and surveys to usability tests and diary studies. The problem is that these repositories almost always become what Paul generously describes as "dumping grounds." Dense folder structures, difficult navigation, and search tools that require you to already know what you're looking for make them practically unusable for anyone outside the UX team. And who ends up using them? Other UX professionals, the people who already understand the research anyway. Everyone else ignores them. AI changes this in three meaningful ways. First, it makes the initial build far less painful. You can throw everything at it, PDFs, old PowerPoints, interview transcripts, survey exports, and AI will structure and organize that material into something coherent. What used to be a daunting, months-long project becomes manageable. Second, it makes the repository accessible to people who aren't UX specialists. Instead of requiring a precise search query, a conversational interface lets anyone ask vague, natural questions. A product manager can ask "what do our users think about the checkout process?" and get a synthesized answer drawn from five different studies they never knew existed. That's a genuinely different kind of value. Third, and this is the part Paul finds most compelling, it can identify gaps in your research. When someone asks the repository a question and there's no relevant research to draw on, a well-configured AI won't fabricate an answer. It flags the gap and notifies the UX team that this is an area worth investigating. Over time, the questions people ask become a demand-driven research roadmap, shaped by what people in the organization actually need to know rather than what the UX team assumes they need. Marcus pushed back on the reliability question, which is fair given AI's well-documented habit of confidently inventing things. Paul's response: proper setup matters enormously. You instruct the AI explicitly not to fabricate, you add a quality gate that checks answers before they're returned, and you can even have it verify claims against source material. Even with pessimistic assumptions, say one in ten answers being wrong, that's still more useful than having nothing at all. And the failure mode is reassuring: if the AI can't find relevant research, it defaults to generic best practice rather than making something specific up about your users. Paul then connected this to something he's discussed before: AI-powered virtual personas. The repository feeds the persona generation. AI analyzes the accumulated research and builds queryable personas from it. Unlike static persona documents that go stale almost immediately, these update as new research is added. And here's the detail Paul is clearly delighted by: put a QR code on your printed persona posters. Scan it, and you're now having a conversation with a virtual version of that persona. Marcus had recently written about the value of physical personas on walls as simple reminders of who you're designing for, and this neatly bridges the physical and digital. The upshot: organizations that invest in an AI-powered research repository end up with something that prevents duplicate research, makes user insights accessible to everyone, identifies gaps in what's known, and gives the whole organization a quick way to gut-check decisions against actual user data. The reason more organizations aren't doing this, Paul notes with characteristic subtlety, is that UX teams are too small and too busy. "Hire me to do it" being the conclusion he arrived at, live on air. App of the Month Notion Paul's pick this month is Notion, which he acknowledges he's almost certainly recommended before, given that he runs his entire business on it and describes its potential failure as roughly equivalent to his own. The recommendation here is specific though: Notion as the platform for building AI-powered user research repositories. Two things make it well-suited for this. First, structural flexibility: you can organize a repository however your organization needs, and bring in almost any format of research artifact. Second, Notion has a powerful built-in AI agent that can reference, search, and synthesize across everything stored in it. That said, Paul mentioned conversations with the RNLI, who use SharePoint and Copilot to achieve essentially the same thing. The principle works across platforms. Notion is Paul's preference, but he'd be the first to acknowledge the bias. Interesting Reads "Google is quietly rewriting headlines with AI in search results" Dan at Headscape surfaced this one. Google has been quietly rewriting the titles of content in its search results, not a new practice, but one that has apparently accelerated significantly with the arrival of Gemini. The example from the article: a piece originally titled "I used the cheat on everything AI tool, and it didn't help me cheat on anything" was shortened to "cheat on everything AI tool." The meaning flips completely. Paul's view: this isn't really an AI problem so much as a "no human in the loop" problem. Remove human judgment from the process and you get outcomes like this. "Testing suggests Google's AI overviews tell millions of lies per hour" This one prompted a longer and more genuinely interesting conversation. The article references New York Times analysis suggesting Google's AI overviews are incorrect around 10% of the time. The illustrative example: AI Overview cited three sources to answer a question about when Bob Marley's home became a museum. Two of the sources didn't address the date at all. The third, Wikipedia, listed two contradictory years, and AI confidently picked the wrong one. Paul and Marcus ended up in partial agreement. Paul's argument: we don't hold websites to a higher standard of accuracy than we hold AI, and the expectation of AI infallibility is inconsistent. The real issue is the word "confidently." AI states things with a certainty it hasn't earned, and the interface doesn't adequately signal uncertainty. Marcus's counter: AI summaries have effectively removed the click-through step, so an error now goes unchecked in a way a traditional search result didn't. They concluded it's largely a user interface problem, acknowledged that Google isn't going to remove the feature, and briefly proposed a BBC-funded public search engine before moving on. Marcus' Joke I'm entering the annual Give Helium to a Sheep contest again, and I'm a bit nervous. Last year the bar was very high. Find The Latest Show Notes

There's a scene in the Steve Jobs biopic where Steve Wozniak asks Jobs what he actually does. Wozniak understood his own role clearly: he was an engineer. He wrote code. He built things. But Jobs? Jobs described himself as the conductor of an orchestra. I've been thinking about that exchange a lot lately, because I think it captures exactly where we're all heading. AI isn't turning us into supercharged doers. It's turning us into conductors, and that requires a completely different mindset. The problem nobody talks about I've been coaching a number of people on integrating AI into their workflows recently, and I keep running into the same pattern. The people who aren't getting time savings from AI aren't failing because they don't understand what it can do. They're not failing because they lack access to the right tools. They're failing because they're fundamentally disorganized. AI is only as useful as the foundation it's built on. If your work processes are messy, your context is scattered, and your task management is a loose collection of mental notes and sticky tabs, AI can't do much for you. It needs structure to work from. I hear this complaint constantly: "AI has been mis-sold to me. I'm not saving any time." But it hasn't been mis-sold. It's just that AI can only deliver on its promise if there's an organized workflow underneath it. Build that first, and the time savings follow. That's why I've written before about building AI playbooks and developing proper AI skills. These aren't nice-to-haves. They're the infrastructure that lets AI actually work. The conductor problem But here's the deeper shift, the one that's genuinely harder to adapt to. When you're doing tactical work, you're usually focused on one or two tasks at a time. You go deep, you finish a thing, you move on. It's cognitively manageable. A conductor doesn't work like that. A conductor holds the entire orchestra in mind simultaneously: what the strings are doing, where the brass comes in, what the percussion is building toward. They're not playing any of the instruments. They're managing the relationships between all of them. In a world of AI agents, we're going to be managing multiple projects running in parallel, all moving faster than any human team would. We're task-switching constantly. We're accountable for outputs we didn't directly produce. And we have to resist the urge to dive in and do the work ourselves, because that's precisely where we get bogged down. The design leader parallel This isn't a new challenge, as it happens. Design leaders face exactly this transition when they move from senior practitioner to managing a team. I've watched a lot of talented designers struggle with that shift. They get promoted because they're brilliant at the work, and then they spend the next year quietly sneaking back into Figma because they can't let go of doing. They micromanage their reports. They redesign things that were already fine. They can't operate at the level of abstraction that leadership requires. Working with AI agents is going to feel very similar. The temptation to wrestle with the AI until it produces exactly the output you had in your head, rather than accepting a good result and moving on, is going to be real. Learning to let go of that control is a skill in itself. The good news is that unlike a team of designers, you can't upset an AI agent by micromanaging it. But you can waste enormous amounts of time doing it, and that defeats the whole point. AI burnout is already real There's one more aspect of this I want to flag, because I don't think it gets talked about enough. When you're managing a team of agents all moving at AI speed, the cognitive load is significant. You're context-switching constantly across multiple workstreams. Things are completing faster than you can review them. It's relentless in a way that managing a human team simply isn't. This is what's increasingly being called AI burnout. Learning to pace yourself, to batch your reviews, to build in breathing room: these are the organizational skills that will separate people who thrive in an AI-augmented world from those who burn out in it. Where to start If I had to distill this to one practical thing: start building the habits of a manager now, before the agents fully take over. Get organized. Build the infrastructure that AI needs to work from. Practice delegating, even to imperfect tools, rather than doing everything yourself. Work on your ability to hold multiple projects in your head without losing the thread on any of them. If you want help working through that transition, I offer coaching specifically for this. It's something I'm increasingly focused on, because I think it's one of the most valuable things I can help people with right now. I'm also running a workshop with Smashing Magazine in July. Modern UX Practitioner covers a lot of this ground in a more structured way, if that's more your style. The shift from doer to conductor is coming whether we prepare for it or not. The people who handle it best will be the ones who start thinking like managers now. Find The Latest Show Notes