Design of AI: The AI podcast for product teams

When reports like Adecco’s Global Workforce of the Future survey find that the average saving for workers using AI is 1 hour a day, we should question this. * What did those workers do with their time savings? * Should that time savings benefit the employer or the employee?* Can we trust such a hard-to-measure stat?Our latest episode tackles this and other disruptions happening to the creative and production processes. Matthew Krissel is the Co-Founder of the Built Environment Futures Council and a Principal at Perkins&Will. For over two decades, he has led transformative architectural projects across North America and internationally. We discussed how AI is disrupting architecture and lessons for digital product teams. He really struck powerful points many times during our conversation about questioning the role of time and permanence in a world when we want more, faster.Other points covered in the conversation:* Commoditizing design makes production easier, enabling societies to tackle challenges like housing shortfalls* Commoditizing design devalues other vital processes, like community engagement, respectful place-making, and longevity of projects* Over-indexing AI’s potential as a workflow optimizer, while under-indexing the potential to reimagine how complex projects are planned and operationalizedListen on Spotify | Listen on Apple PodcastsIn this newsletter, I’d like to tackle the concept of time saving and what it means from the perspective of crafting an AI strategy. Here was the most important quote from the episode: So just because something took half the time it did before, what happened is we just did more. So we just filled the time. Is there something higher and better use? I suspect that somewhere along the line the designs got better. Also I suspect that somewhere along there was diminishing returns. We were just doing more because we could not that it was actually yielding anything better.  Are you gonna focus on fewer, but better increase your quality? Are you going to spend more time on business development or some entrepreneurial side hustle? Just go home early? What you decide to do as we start to gain productivity time is going to shape a lot of where this is all happening.Newsletter recommendation: Scott BelskyEssential insights and lessons from Scott Belsky that anyone building with AI must read. His newsletter is fantastic and a must-subscribe because of his unique cross-section of expertise across creativity, product, and innovation. His books have also always been pivotal reads to advance your craft. Hopefully, we can do some of the same with our Design of AI podcast and newsletter. Who should benefit most from your ability to learn AI: You or your employer?The challenge to creatives and builders is to decide who should benefit from these transformative technologies if you’re self-taught:* Should you gift your employer the benefits if you’ve taught yourself ways of getting 25% more work accomplished in a day?* Should you gift yourself the benefits of your increased productivity and work on side projects, or spend more time with your family?Historically speaking, employers were responsible for the means and training of production. They paid for novel technologies —desktops, SaaS, big data— and were responsible for training you on how to use them. AI is different because employers are often lagging behind employees in embracing and educating on how to use the technology effectively. It is very easy to argue that the 200 hours you’ve spent learning AI outside of work hours should exclusively benefit you.AI Time Savings: Benefits & RisksTechnologies have consistently saved us time, but the resulting effects have been questionable. The internet and mobile phones connected the world, while also leading to increased poor health outcomes due to more time sitting. We also spend more time alone than ever.Further back, the Industrial Revolution raised the quality of life for everyone. Still, the commoditization of work led to industrialists exploiting child labour and putting everyone into deplorable working conditions that polluted communities. The time the workforce saved most benefited employers, with employees giving up their ways of life in favour of steady incomes. Most relocated to cities, got cut off from their families, and learned the pain of commuting for the first time.When it comes to AI, the benefits we hope for centre on automation and augmentation. The hope is that we will benefit from less shitty work (automated away) and that we can our new capabilities (augmented by AI) will enable us all become wealthy entrepreneurs. Sure, this may be true for the top 0.01% of AI users who learn how to run a typically 10-person business by themselves. For the rest of us our work may in fact get a lot shittier. At least that’s what the authors of the upcoming book, The AI Con, believe. The authors (and upcoming Design of AI guests), Alex Hanna and Emily M. Bender tell a tale of how AI’s risks have been severely hidden under the rug. In their book, they document many examples of the technology performing so poorly at tasks that products were shut down within weeks.Maybe the future of businesses will look a lot like Amazon: A business offering endless products of questionable quality and provenance with no humans in sight except those working the worst possible jobs in sorting information, like something out of Severance. In this scenario, the majority of humans will be employed as mall cops of the technology, swooping in when a problem happens that slips between the programming and policies. At this point, AI hypers would argue that even if the enshittification of work is inevitable, AI will open up new and better types of jobs. Only time will tell. How does AI change our relationship with time?When buying productivity-boosting hardware and software, the expectation has always been that the results are undeniable. Going from handwriting to using a typewriter was immensely faster. The same is true when buying a new Saas platform that makes managing projects infinitely easier. Now, with GenAI-powered products, the ROI is unpredictable. The vast majority of capabilities deliver the illusion of rapid progress. Think of image and video generation —the immediate results are shockingly impressive. But getting results to be production-ready requires mastery of probabilistic software and/or resetting your expectations. It all means that the operator —you— ultimately plays a bigger role in the ROI of using this technology than with previous ones.So-called Vibe coding is a major testament to the time savings that AI can create. Anyone can now build a website and app without writing a line of code.Vibe coding platforms —like Cursor, Lovable, Replit, and many more— are fantastically easy to use… until they’re absolutely painful to use. The stunning early rewards turn into confusingly broken components all over.Again, results depend on the operator’s ability to debug using an entirely new interface paradigm (conversational). This continues the technology’s remarkable inversion of the value paradigm, where workers define the quality of outputs.Looking ahead, mastery of data will triumph over mastery of interfaces. This favours employers who unlock the power of their first-party data and build solutions that augment and automate the expertise of their employees.Always worth reading, strategist and tech critic Tom Goodwin posted an intriguing analysis on LinkedIn this week. At the core of the guiding philosophy regarding AI-assistance is that the more complex the task, the less qualified AI is to work on the task unassisted.Check our previous podcast episode and newsletter for more details on how to unlock the power of your data. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

As much as image generation is fun, the power of GenAI is prediction. The technology operates very similarly to people you might meet: * Some people have studied and are experts in a single topic for a decade. They’re experts in that topic and can easily infer, correct, and complete tasks. They’re unreliable for everything else.* Some people are generally knowledgeable and have a good understanding of many topics. They aren’t experts but can reliably assist you in many ways. But they’ll also be wrong sometimes.OpenAI, Anthropic, etc.— are highly knowledgeable in almost every topic. That’s the result of being trained on all accessible information online, data they’ve licensed, plus data they’ve allegedly stolen. AI products built on these frontier models are immediately powerful for completing any task. But if you build a point solution on proprietary data explicitly trained on a narrow topic, it can achieve an expert level. That was the focus of our conversation with Tyler Hochman, the Founder and CEO of FORE Enterprise. We discussed unlocking AI’s predictive power by focusing on expensive and repeating problems. How any business or founder can leverage and/or specialized data sets to train AI models to deliver powerful prediction capabilities.Listen on Spotify | Listen on Apple Podcasts | Watch on YouTubeHe’s built AI-powered software to predict when employees may leave their jobs, offer fashion advice, and help professional sports teams improve performance. This video explains how to train your model using Figma files.This conversation highlights how important your first party will become. This data includes more than just your customer data; it should include documenting workflows, quantifying initiatives, and developing a matrix of your offerings/capabilities. Anything repeatable must be quantified as a learning tool.Example of a data collection strategy for AI trainingWhen OpenAI launched a new image generation feature in ChatGPT, everyone jumped on it. AI-generated images infested our feeds in the Studio Ghibli style. These images sparked a lot of worthy debate about copyright infringement, which added to the ethical concerns about how OpenAI trains its model. A recent study highlighted evidence that ChatGPT is trained on copyrighted works.Given that AI models are running out of data to consume, they need to find clever ways to access a new data set. Enter ChatGPT’s image generation tool and Ghibli craze. Millions of people have been feeding their photos into the model, giving it access to an entire universe of new training data to improve the quality of its image generation capabilities. Lesson: Collecting user-generated content can provide your custom model with access to training data that was never possible before. This holds true whether your product is a document scanner, video generator, accounting software, run tracking app, or anything else. As we move into the next phase of AI model evolution, the data you have access to might become your best competitive moat. Thus, businesses with access to ethically sourced content from their communities and customers have an advantage.Thanks for reading Design of AI: Strategies for Product Teams & Agencies! Future of AI-powered workforcesYesterday, LinkedIn exploded with screenshots of an internal memo sent by Shopify CEO Tobi Lutke to teams. It marks the most public evidence that AI is moving from a toy we experiment with to a critical skill that you’ll be scored in your next performance review.The data backs up that AI adoption is surging within workplaces. A study by the Wharton School at the University of Pennsylvania collected data on which use cases AI is most used for. The report highlighted use cases that every business and employees rely on daily or weekly. Not so long ago, employees secretly used AI at work. The year-over-year data indicate that AI products are becoming adopted at an organizational level.AI’s impact on our lives will be dramatic & potentially dystopianStanford’s 2025 AI Index Report offers metrics demonstrating the significant leaps forward AI has made across performance and usage metrics. The technology has already surpassed human baseline performance on many measures. And the technology’s predictive capabilities are showcased in how effective LLM’s performance in clinical diagnosis. It points to a future where every one of us —physicians, educators, factory workers, and beyond— will rely on AI to make more informed decisions. MUST READ: Futures essay about future of superintelligenceThe AI 2027 essay, written by researchers and journalists, examines the question of what happens on a global level as we approach AI superintelligence. A long and worthy read, it illustrates that we are much closer to superintelligence than the public may believe and that the snowball effects of achieving it are massive. They predict dystopian outcomes unless the world unifies around regulations and safety guidelines.If their predictions are true, we’re being distracted by the table stakes of Ghibli image generation and coding tasks. This technology will utterly transform our personal and professional lives. It will give governments immense power over one another. And it will open Pandora’s box of dreams and nightmares.If you need to chat through the implications of these predictions, email us info@designof.ai. We’ll definitely discuss this in detail in our upcoming episode with the authors of the AI Con book and hosts of the Mystery AI Hype Theater 3000 podcast.Podcast recommendation: The Most Interesting Thing in A.I.The Atlantic’s Nicholos Thompson started an amazing podcast showcasing strategic topics about AI.Listen to the Andrew Ng episode. It dives into important topics about the future of frontier models and the implications of running out of training data (if it happens). Thanks for reading Design of AI: Strategies for Product Teams & Agencies! This post is public so feel free to share it. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

“The future is already here, it's just not evenly distributed” Science fiction is inspiring, frightening, and often the best lens into the future. Many ideas about the future are b******t —just like this quote being misattributed to the ever-amazing William Gibson— but even the wildest idea shares truths worth discussing.This week’s newsletter is an exercise in imagining how AI will transform the way that we work. The future will impact us differently because some already live with a future-centred mindset, while others prefer to shift their thinking daily. One such future-centred thinker is John Whalen, the author of Design for How People Think and the Founder of Brilliant Experience. He shifted from being an AI skeptic to an advocate because he sees a tidal wave of change coming to how product teams operate.Listen on Spotify | Listen on Apple Podcasts | Watch on YouTubeIn the episode, we discuss how he’s implemented AI into his workflows and how he can now accomplish projects in one week that used to take seven weeks to complete. He makes a compelling case for why every team should use AI-moderation and synthetic users to enhance product outcomes. But most importantly, he’s become an AI advocate because, over his three-decade career, introducing new tools has always been met with doubts and resistance. Ultimately, businesses force the adoption of tools that deliver a clear ROI. There’s still much to debate about AI. Reports like this one from Microsoft continue to show that AI isn’t ready to replace humans at key tasks. Another 2024 study found that ChatGPT delivered inconsistent results on a key qualitative research task, compared to humans. The most important thing about this study wasn’t that humans outperformed LLMs; it was the significant performance improvement from GPT-3.5 to GPT-4.0. AI is getting much better at tasks that seemed unimaginable to automate. We’re hearing the same shocking stories across design, development, research, marketing, and sales. Undoubtedly, AI will be able to automate most of our work within a few years.Will that mean we’ll be replaced? Yes and no. Just like the industrial age and globalization destroyed artisans, AI will significantly reduce the headcount of “artisanal” product people and the rest of the work will be an assembly line of tool operators.Automation will significantly change many people’s lives in ways that may be painful and enduring. But for the economy as a whole, more jobs will be created, and those jobs will look different from those today.Thanks for reading Design of AI. Subscribe to receive new posts.Should we be worried about our jobs?These same conversations are happening across all fields:* Will AI Replace Therapists?* As Technology Progresses, Certain Accounting Jobs May Fade Away* The Risk of Dependence on Artificial Intelligence in Surgery* AI could terminate graphic designers before 2030You’re probably reading this with a sense of confidence that you’re shielded from the impacts of AI because you’re working on the bleeding edge of technology. It’s true. You should be better equipped to navigate the changes as they happen and adapt to the future better than others. Conversely, your roles face additional pressure to change faster than in other industries. The business realities of being backed by venture capital and private equity mean you’re always chasing the future. Tech and agencies have to unlock benefits from AI or risk losing market share and funding.The problem is that nobody can agree on AI's expected impact because it’s still just science fiction.According to the OECD report, the level of impact will largely depend on the level of adoption. High adopters might expect a 3x gain compared to those who adopt AI minimally. A McKinsey report highlights the pressure being placed on employees. Their data shows that C-suite executives blame employee readiness as a barrier to gaining benefits from AI. Only 1% of them believe their AI investments have reached maturity.Combined with last week’s conversation with Jan Emmanuele, AI investments in creative augmentation and automation will surge in 2026 and beyond. This suggests that employees will be under a lot of pressure to become more productive or else be replaced. Listen to that episode for more details on how AI is being adopted:Listen on Spotify | Listen on Apple PodcastsHow will jobs change as a result of AI?There’s no doubt that our jobs will change. They’ve had to change every time a transformative new technology becomes widely adopted. The only difference now is the speed at which change is happening.Let’s analyze how roles are changing from the perspective of product teams.* Our jobs used to be distinct. Each of us had specialties and expertise in areas that protected us.* Our jobs are increasingly commoditized, meaning people from other jobs can do many of our tasks.For example, a designer can now do tasks that previously were out of their sphere:* Use ChatGPT and Cove to explore a strategy and build a business case.* Use Wondering and Vurvey to launch and analyze a research campaign.* Use Lovable and Cursor to prototype and build out a product.Our roles are blending into one another, and employers no longer need as many people to deliver the same amount of work.How we work is also changing. AI is simplifying core tasks along our workflows and automating cumbersome steps. Here’s an example of how AI will transform UX Research:If you map your workflow, you’ll find a similar transformation happening to your role. Humans will drive decision-making, but AI will increasingly inform those decisions.Maybe John Whalen’s vision of product teams as AI-conductors is most appropriate: Maybe there will really be fewer UX researchers. Maybe they're more focused on this I'm calling sort of storytelling or conducting. I picture someone orchestrating these things.What you can do to enhance your futureJohn Whalen’s story shows that you can be an industry expert who has written a respected book and led a successful practice, yet still need to adapt to the coming change. He’s shifted from being a researcher to being a research technologist, one who delivers projects that used to take much more time and many distinct roles. This is similar to what Phillip Maggs said on episode 20 about becoming a design technolgist (Listen on Spotify | Apple).Recommendations to help you:1. Get closer to the decision-making processWe’re all anxious about the economy. The viability of many businesses is at risk, and job security is no longer guaranteed. Our goal should be to bring confidence and certainty to our work. That means pinpointing what our internal and external stakeholders are most worried about and delivering solutions that address those.In the case of John Whalen and UX researchers, stakeholders had questioned the certainty of insights. With AI, John and others can deliver a 10x larger sample size in more markets.Similarly, designers, writers, PMs, and developers should use AI to deliver work more confidently. You’re able to get more user feedback at every stage of the process. You can scale your work to be localized to more markets. You can automate tasks that are cumbersome and error-prone.None of this is to minimize being human-centred. But the industry has been questioning whether orgs have been perpetuating the illusion of user-centred design. Managing stakeholders’ expectations puts you closer to the decision-making process and gives you the ability to dictate how good work happens.2. Challenge the assumptions that limit expectationsNew apps are released every month that bend our perception of what’s possible. If you had collected a list of capabilities that you wished were possible, they probably exist now. Your job must be to push the work beyond the assumed limitations. To do this, you must test new apps and see if they can confidently overcome the limitations to your work. Explore new capabilities in the apps you already rely on. Experiment with combining applications that excel at key parts of your work.Being tied to a single legacy app is the worst thing you can do. You’re hitching your future to that product’s ability to be better than the dozens of other teams simultaneously trying to disrupt each other.3. Walk into every situation with clarity about your value drivers and superpowersWe can obsess over clients and our work, but understanding what you're exceptional at is more important than everything you deliver. We’re much more than our performance reports and more capable than the best project we’ve ever worked on.It requires us to be self-critical about what drives us, what limits us, and where we can excel. For example, you might identify that:* You’re envigorated by structuring and organizing * You’re envigorated by hacking solutions and testing capabilities* You’re exceptional at building alignment and support for initiatives* You’re exceptional at taking on complexity and uncertaintyThese fundamental truths enable you to dictate your path to success better:* Who you should be working for* What types of projects and roles you should be working on* What unique capabilities you should be highlighting* Which principles you should use as a north star for leveraging AIIf this is a topic you’d like to me dive deeper into, please leave a comment or send a message.4. Remember that the future is not evenly distributedThe closer you get to the centre of tech, the pace of change will increase. The gravity of the situation is exciting for some and utterly exhausting for others. Find the orbit that best suits you.If you’re reading this newsletter, you’re clearly a future-centred thinker. You can leverage that in the centre of tech to push projects and productivity to new heights. You could also work in a traditionally slower industry —healthcare, government, legal, education— and affect more change by challenging long-held assumptions.All change is relative but what brings you joy and meaning is deeply personal. Embrace that.One last and important consideration…Erika Hall speaks the uncomfortable truths that we need to hear. Follow her.Some jobs simply aren’t worth keeping. Some uses of AI are appalling. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

There are many reasons to debate the ethics and implications of AI. But while we do that, hundreds of the world’s biggest brands are rushing to implement the technology into creative and coding workflows. At a time when shareholders are being unforgiving and policy making is volatile, business leaders are looking to AI to gain any advantage possible.Jan Emmanuele is one of the experts that these Fortune 500 corporations rely on to identify and build GenAI creative workflow augmentations and automations. He works for Superside —whom you might remember from our episode with Philip Maggs (Listen here)— because they’re on the leading edge of creating an LLM that interprets your briefing process, design system, brand guidelines, marketing campaigns, and data to automate high-volume creative tasks. In this episode, we focus on how and where AI is applied within organizations and workflows. It details how organizations can prepare themselves for implementing AI and how to address the core barriers and risks of the technology.Listen on Spotify | Listen on Apple PodcastsWhat was most interesting about this conversation was his prediction that the adoption of AI will explode in enterprise orgs starting in 2026 and that it could continue into the 2030s. He believes that the value of AI in enterprise has already been proven and that more use cases exist than anyone can believe. That adoption thus far has only been limited because of legal and procurement policies.If this is true, organizations that aren’t already at least planning for this workflow-automated future will soon be at a huge competitive disadvantage. Finding 10x augmentations of creative output is routinely achieved, and more will be possible for organizations with highly-structured and easily-repeatable workflows. The gains will be largest in orgs that leverage the uniquely-LLM capability of contextualizing outputs based on data. Examples include localizing campaigns to micro-niche segments or regions of the world. Thanks for reading Design of AI: Strategies & insights for product teams! This post is public so feel free to share it.Headwinds will reduce the number of creatives earning a living wageAs we barrel towards the increasingly inevitable reliance on LLMs, it puts creatives in the uncomfortable position of fighting for their survival and protesting for what’s ethically correct. The music industry is the canary in the coal mine in this battle. Many artists earn the majority of their income from their back catalogues and LLMS are effectively using those albums as mulch to improve generative capabilities. On one side, you have an entire way of life being threatened; on the other, you have artists that will quickly need to learn how to master generative capabilities to become an indispensable musician regardless of the headwinds that will reduce the amount of music earning a living wage. As platforms get better, we’ll just generate the music and images we need instead of hiring professionals.Overcoming the uncanny valley: Not being able to determine what was generated by AIWhat has made all of us feel more comfortable has been that AI still sucks at a lot of creative tasks. Blooper reels and countless articles of AI creative generative fails give us hope that the technology isn’t ready to replace anyone yet. But we’ve learned from our latest episode and many previous ones that the technology is much more ready for primetime than we might believe. Many of the failures we see today result from the false sense of confidence the platforms offer novices. While the simplicity of these tools has exploded the amount of experimentation happening, we’re flooded with more fails than fantastic examples.Another factor is that the simplicity of the GenAI interfaces obscures the complexity happening in the background. We believe we can generate a campaign-ready 20-second video by typing in a prompt. But the complexity comes from knowing what models, protocols, data sets, and projects to connect for the best outcomes. This is an era dominated by creative technologists who can see these possibilities and stay up-to-date with the latest capabilities.In the hands of someone who understands how to overcome the rawness of the technology, the possibilities are limitless. And for every project we see published, there are at least another dozen working to push those capabilities further in the near future. Sesame is another example of technology overcoming the uncanny valley by delivering conversational voice capabilities indistinguishable from humans. These developments are happening at such a pace that it’s impossible to keep up. For example, researchers have created an agentic, autonomous framework that iteratively structures and refines knowledge in situ.The point is that whether you agree with the hype of an AI-powered future or not, businesses everywhere will implement it because the impact is increasingly undeniable. Action items: What can we do to prepare ourselves and our workI hate that the ethics of AI seem like an afterthought to the beating drum of business automation. It’s deeply uncomfortable that many professions and industries must adapt or face extinction. The only way to stare into this abyss and feel hopeful is to believe that the rising tide of resentment against big tech will fuel a renaissance of altruistic misfits building the models and layers that do less harm. But that won’t calm the nerves of the musicians and artists who see an end to their way of life today.We can mourn the tidal wave of change while also preparing for the new world order that comes next.If you’re a creative:* Stop undervaluing yourself and your work. Listen back to yourself explain the work you do. Recognize all the steps, decisions, and life lessons you neglect to mention. You need to document who you are to such a granular level that you spot where your genius is most pronounced and where you’re on autopilot. Then consider how to leverage AI to amplify/automate each of those.* Tap into your most significant creative strengths. You are more than your outputs. You fell into this career for a reason and persist because of at least one exceptional creative strength. Document it and the conditions under which it enhances your work more than others. Now find AI tools that can make that happen more often and for longer periods. * Lead the change you want to see. Don’t wait for inspiration and innovative products to land in your inbox. Go find them, test them, implement them, and prove if they can or can’t help you achieve your goals.If you’re a business leader: * Accept that change is coming fast. You can feel unsure about the technology, worried about the risks, and apprehensive about the costs. But you cannot wait to start imagining what the future of your business and industry might look like. Go through future casting exercises and monitor the countless startups slowly eating away at your competitive moat.* Empower your team to succeed. Even if people tell you they aren’t worried about the coming change, they probably are. You need to lead them through this and create a shared vision of what the future version of your business and workforce can look like. Include teams in co-creation processes to determine the best ways to empower them to succeed by eliminating barriers and inefficiencies.* Structure your data and production workflows. AI is most effective in highly repetitive situations where success can be easily evaluated. Businesses will succeed that have standardized their key workflows and have structured data that adds critical context about situations and success. Do the work now before an expensive consultant charges you millions once there’s a veritable gun to your head due to competitive concerns. Contact me if you need helpThank you for following the Design of AI podcast and this newsletter. This year, we’ll spend more time discussing this seemingly insurmountable challenge of implementing AI effectively. Please comment if there are specific questions or topics you need us to discuss. And feel free to vent about topics that you’re most frustrated or concerned about so we know what our community needs.We’re also hoping to launch some events in major markets this year to bring together early adopters and experimenters with those eager to leverage this technology effectively.And if you need help with any consulting related work related to envisioning your AI-powered future, email me at arpy@ph1.ca Product of the month: RaycastRaycast is a perfect example of the disruptive potential of AI. While everyone else is running to add bullshitty AI features to make using their products easier, others are rewriting the way we interact with digital experiences. Raycast basically looked at MacOS and said, “Let’s rebuild the entire finder and launcher experience.”It’s ironic for me because one year ago, I worked on a project where the outcome was the real potential value of AI in a mobile phone experience would be as an assistive launcher experience that eliminates all the inefficiencies of Android. Well, here it is! Thanks for reading Design of AI: Strategies & insights for product teams! This post is public so feel free to share it. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

Whether we admit it, like it, or believe it, we’re in a relationship with AI.That’s the first of many powerful reflections made by Sara Vienna, Metalab’s Chief Design Officer, in her must-read manifesto about how design and product must evolve. Unlike the design leaders who speculate about AI's impact, Sara and her world-class team are years ahead. They are designing disruptive AI product experiences and leveraging AI to elevate their workflows. Sara’s episode is one of the most important conversations we’ve had about the future of design and products.Listen on Spotify | Listen on Apple PodcastsShe believes that AI will change how we work and what we build. Those who embrace the potential of AI will succeed in the oncoming disruption. But most importantly, the future of product+AI will be in making five mindset shifts:They’re fundamentally principles for humanizing experiences. The hope is that AI will finally bridge the divide so products can deliver the value we’ve always wished was possible in the most humanized way possible. But there will be challenges in accomplishing this:* Most product orgs are built around the concept of delivery, not design excellence* Unlocking user data: Getting access to valuable data and knowing how to use it in a meaningful way are still more fantasy than reality* In every direction we turn, trust is being diluted* Design as we know it will need to be reborn to adapt to move from creating pixel-perfect interfaces to ones that adapt and spawn based on user interactionsAgain, I highly recommend listening to the entire episode.Thanks for reading Design of AI: Strategies & insights for product teams! This post is public so feel free to share it.Envisioning the future of design & productIf we extrapolate on Sara Vienna’s vision of how design should change, a couple of core reality checks come to mind:* Today, we can’t even conceptualize what products will be able to do tomorrow. Just like new AI tools are being released faster than we can read about them, more teams than ever are competing to deliver the use case & interaction model that will redefine a category. It’s a race to an undefined & moving finish line.* The underlying models may be the heartbeat of future products, but design will always be the brain. Products plug into whichever model suits them best at a particular moment, usually based on cost and accuracy. But just like each of our minds brings a different lived reality and way of using knowledge, the models are less important than the strategy that’s been designed into the product.* Fewer designers and product managers will yield immense power. AI automation platforms —like Make and Loveable— can effectively replicate more than half of products today. This percentage will grow until such a point that any product will soon be able to be cloned, undermining its competitive advantage. The designers and product managers working on the future of design will have the funding that enables them to compete in a global race that they’re likely to lose because they don’t know what competition they’re actually facing. The rest of us will be working to keep the lights on.Big question: How should we be using AI, today?Photoshop celebrated its 35th birthday today and is a perfect reminder of how disruptive platforms eventually become part of the boring vocabulary of the everyday.GenAI platforms, like ChatGPT, are in their infancy. Everything seems equal parts novel and confusing. We’re still unsure how to use this superintelligence, only that we should be using it. Photoshop’s rise was similar: a platform that opened up so many possibilities but whose ultimate impact wasn’t felt until it redefined the designer role many years later. What’s happening today is that employees are smuggling AI into work and this makes sense given the recent McKinsey report that finds that leaders are slow to adopt because of risks and a lack of vision.Our research finds the biggest barrier to scaling is not employees—who are ready—but leaders, who are not steering fast enough.Anthropic, the maker of Claude, published their Economic Index report and found that AI use is most prevalent in computer & mathematical occupations. Their AI model is mainly used for programming and administrative tasks.What the data also show is that design and creative tasks aren’t core use cases, yet. And rightfully so, large language models best serve requests about processing content and code, not pixels and ideas. A report about how generative AI is used in journalism showcases this by highlighting that even the creative tasks are largely operational ones, like resizing images and animating.This data highlights the divide in how leading organizations, like Metalab and Superside, leverage AI compared to the everyday user. While the average person uses Midjourney to generate stock art, leading designers automatically generate localized creative based on design systems and content guidelines.The reality is that product teams have three core workstreams:* Operations: Planning, organizing, editing, describing (e.g. Notion)* Creativity: Ideating, revising, collecting, analyzing (e.g. Cove)* Productivity: Deciding, planning, organizing, explaining (e.g. ChatGPT)Every designer, product manager, writer, producer, and researcher completes tasks in these three workstreams. And every one of you should be taking the time to break down your typical workflows into discrete activities so that you can explore what AI solutions can either augment or automate non-critical tasks.An example of this is how Kyle Soucy is using AI to streamline person and journey map creation. This type of knowledge work is considered sacred by traditionalists but as you can see in her article, she’s broken down her workflow to find effortful tasks to be augmented/automated. We can question AI’s accuracy all we want. We can challenge if models were trained ethically. And we can debate what percentage of your job may benefit from using AI. What will not change is the undeniable truth that the intelligence and capabilities of these models and tools will only improve. The sooner we embrace that truth, the better positioned we are to control our own fates. For example, a recent study evaluated AI vs. human-generated therapy responses from 13 expert therapists (clinical psychologists, counseling psychologists, marriage and family therapists, and a psychiatrist). The report found (questionable) data to indicate that users couldn’t tell if a human or an AI made the responses. The AI also outperformed human therapists on empathy, professionalism, and cultural competence.We’ll soon reach a point where generative AI can output designs that are indiscernibly human or automated. In this near-term reality, the role of designers must evolve or be replaced.A recommended action plan for how you should be using AI today:* Plan projects and workstreams using templates, resources, and added context* Communicate ideas and insights better by using AI to iterate and expand* Question the rationale, assumptions, and factors that impact the project goal* Compile inspiration, ideas, and information that will broaden your thinking* Analyze larger data sets and more sources than you could have before* Challenge your concepts by making variants and exploring new directions * Create more deliverables by automating localization, multiple formats, and generating content based on systemsIf you enjoy this content, please make sure to listen to the Design of AI podcast on Spotify and Apple. Make sure to follow us and rate the show if you like the show!Add me on LinkedIn if you want to ask any questions or discuss a project. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

There’s little debate that AI will change the world. What we’re not so sure about is if AI’s expected disruptions to how we work will be outweighed by the benefits of accessing a super-intelligence.David Boyle thinks of LLMs as an electric bicycle for the mind, one that enables us to go farther than we ever imagined with much less effort. His opinion comes from being one of the first market researchers to experiment with LLMs and subsequently turn his learnings into the PROMPT series of books to help marketers, startups, researchers, musicians, and other creatives benefit from the emerging technology. He’s an audience research expert who has informed global strategies for many of the world’s biggest brands.In this episode we explore why David Boyle believes that AI can make strategy & research work faster, cheaper, AND better. Listen on Spotify | Listen on AppleThe conversation explains why any product manager, researcher, strategist, or creative should leverage AI. The greatest advantages are speed and quantity because GenAI overcomes research’s most time-intensive tasks: codifying and thematic analysis of large data sets.David admits that one of the biggest challenges is that AI are often confidently wrong and that experts must verify the results.This episode raises important questions:* If AI will make all tasks faster, what changes should we expect to our way of working? Consider how the internet is homogenizing the way we live globally.* If a human expert must verify results, how can we trust the results of AI tasks as soon as the velocity scales past the number of humans in-the-loop?* If executives are excited by AI reducing the cost of research, what will stop them from preferring synthetic or non-human verified data once the cost nears zero?Recommended articlesThe Future of Design: How AI Is Shifting Designers from Makers to Curators by Andy Budd“AI is transforming design, shifting designers from hands-on creators to curators focused on strategy” is the most common prediction about where design is headed. The author believes the design roles will evolve to where and how they can best deliver value and it will likely be in enhancing the quality of work delivered by AI. As optimistic as it sounds —hey everyone wants to be more strategic, yay!— the truth is that in this future scenario, the concept of being a design completely changes with most being dedicated to managing AI tasks and the best assigned to bespoke design tasks that must be perfect. The End of Programming as We Know It by Tim O’reillyMakes a case that each fear cycle about software developers getting replaced actually led to an evolution of the craft. He admits that “Eventually much of what programmers do today may be as obsolete” but that it will be more akin to how the old skill of debugging was replaced with roles tackling more complex tasks. As knowledge workers we have to be concerned because our work can’t be quantified and automated in the same way as the production-line model of development.AI agents will replace SaaS software by Ayan MajumdarIn this analysis of the CEO of Microsoft’s statements that "AI agents will replace all software" he breaks down common SaaS use cases and whether AI can replace those use cases. He concludes that “The shift towards intelligent agents signifies a move away from manual software interactions towards more intuitive, AI-driven processes.” Overall this is further evidence AI agents could replace the SaaS layer which often only existed to give custom lenses to your own data.AI-Generated Slop Is Already In Your Public Library by Emanuel MaibergThe enshitifaction of knowledge is now hitting libraries. Libraries, once keepers and curators of the world’s most important knowledge now can’t guarantee the accuracy, provenance, and value of many works being submitted. “My library, like most, does not have the resources to be checking Hoopla on a weekly basis to weed out what we wouldn’t want there.”What being replaced by AI in 2025 looks likeWhere does knowledge work go from here?Here’s an example of the disruptions possible today where OpenAI’s new Deep Research was used in combination with Gamma to do big consultancy-level research into a market and publish a stunning report. All in 2 minutes.Agencies & consultants: Any business that doesn’t learn to adopt AI to augment and automate workflows will be at risk of losing niche projects to competitors who are optimized for price, speed, and/or scale. Legacy and large orgs tend to be overloading team members so much to remain profitable that they will be slow to adapt to challengers who will turn AI into a major advantage in a price-sensitive market.Researchers & designers: Orgs are hungry to cut costs and will jump at the opportunity to automate rote tasks. Worse yet the entire value of design and research is becoming so commodified that at least one of your leaders will have the misguided belief that everything you do can be automated. Find a culture that values you and become an expert in leveraging the tools to augment your imagination, planning, iteration, and delivery.Analysts & marketers: AI is giving you an ever-expanding superpower to access more data and analyze it more effectively. Your value only goes up if you challenge your own assumptions of what is possible. Being flexible with the tools, platforms, and methods you use will only lead to better outcomes. Unlike other knowledge workers you’re experts in how to deploy copilots and agents effectively because you know how to structure data and requests.Recap from Autonomous AI SummitThis week thousands of industry leaders and strategies attended the Board of Innovation’s online summit. Content was largely focused on shifting perspectives about the technology, the future, and use cases.Day 1 recap by Chisoko Luala SimbuleDay 2 recap by Chisoko Luala SimbuleThanks for reading Design of AI: Strategies & insights for product teams! This post is public so feel free to share it. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

Up until recently Miro was the innovator’s defacto collaboration platform. In recent years a long list of apps added similar functionality to eat away at the online whiteboard segment. Our latest episode with Ioana Teleanu, Miro’s former Lead Product Designer for AI explores the challenges and opportunities of leveraging AI to enhance an existing product.Listen on Spotify | Listen on AppleKey takeaways:* When a product experience is already good, do we need to add AI?* AI makes it easier for more products to enter your category and add unexpected competition* Adding AI forces product teams to ship quickly to be able to learn, sometimes with uncertainty attached* You must consider if AI is the right solution to the problem you’re trying to solveIf you have any questions about these or other AI questions, reach out to us and we can help you upack what it means for your product.Thanks for reading Design of AI: Strategies & insights for product teams! This post is public so feel free to share it.Next week’s podcast episode features David Boyle who makes a case for why AI is transforming what we can learn about audiences and how those insights will improve our ability to strategize.Featured articlesAI Agents: How Businesses Must Adapt or Risk Obscurity (Arpy Dragffy)Ethan Mollick is right: #AIAgents are going to fundamentally change how websites, apps, and APIs are structured. But the implications go far deeper. We’re rapidly moving from a world where users seek out information to one where it is pushed to them by AI agents acting on their behalf. This shift has profound consequences for businesses of all sizes, and those that fail to adapt risk disappearing into the noise that these agents must sort through. (Read full article)The Wild Future of Commerce & The Rise of Conformative Software (Scott Belsky)This edition explores forecasts and implications around: (1) wild expectations for the future of commerce, (2) the era of “conformative software” that becomes more tailor-made as you use it, and (3) some surprises at the end, as always. (Read full article)25 Themes for 2025 (Bronwyn Williams)A cheat sheet of the 25 top things I'm watching unfold in 2025 (See presentation)Always remember… a good learner appreciates being proven wrongBiggest story of the week: Deepseek Ovetta Sampson is one of the must-follow voices in AI, bringing a rational perspective to an otherwise nonsensical chorus of voices. Read the full article here.Deepseek is the new Chinese model that is the biggest AI story of the year (so far). Yes the model is Chinese and may be compromised. But what’s most compelling about this story:* Despite the US places restrictions and pumping country-sized funding into AI, the model outperforms every model made in the USA.* Just like China has done in countless other industries (e.g. Shein and Huawei), they create copy-cat products that deliver 80% of the value for significantly lower cost.* We went into 2025 thinking OpenAI won the AI model wars that we’d all be subjected to whatever pricing they forced upon us. WRONG. Deepseek and many more will now come along and chip away at that expectation. More analysis about China’s disruptive Deepseek model:* Why DeepSeek Prompted a $1 Trillion Tech Sell-Off (Business Insider)* 🐳 I just finished a deep dive into DeepSeek’s latest R1 model & paper. I am genuinely impressed by their approach. A few thoughts. (Reuven Cohen)* Running DeepSeek r1 32B locally is kind of depressing. (Ethan Mollick)* DeepSeek is out, the Stock Market crashed and Silicon Valley is in tears. Here's what happened in the last 48 hours and what it means for your business (Tobias Zwingmann)* The labour market over the next few years is going to be a complete 💩show thanks to super smart, super cheap open source Ai. Here’s why. (Reuven Cohen)* What if I told you the $500B Nvidia selloff is missing the point entirely? I believe AI Revolution Just Had Its "iPhone Moment" (Simon Taylor)New AI products worth trying outStorm Stanford’s new platform enables you to build a paper about nearly any topic with summary and references. Helpful to get a deep dive into big topics, from your desired perspective.RileyNew market research platform which guides users through questions about their product, competitors, and data available. As a proof of concept of where things may head it is compelling.Thanks for reading Design of AI: Strategies & insights for product teams! Subscribe for free to receive new posts and support my work. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

Healthcare is constantly highlighted as the industry that will benefit the most from AI. The prospective opportunities are endless: Improve access to services, improve quality of service, patient outcomes, and medical research. An analysis predicts that the healthcare could save up to $360B a year by implementing AI.That’s we invited an expert to discuss what other industries can learn from healthcare’s massive AI opportunity. Spencer Dorn, the Vice Chair and Professor of Medicine at the University of North Carolina. He is a contributor to Forbes and one of LinkedIn’s Top Voices speaking on Healthcare + Innovation.Listen on Spotify | Listen on Apple PodcastsKey takeaways from the episode:* AI has been impacting healthcare for years, especially to create Electronic Health Records (EHS) as a way of centralizing information* AI is being explored today as assistants to medical professionals (e.g. Virtual/digital scribes) and across a variety of diagnosis scenarios (video)* But the rollouts have been plagued by consistent issues related to adoption and poor comprehension of the actual problems* To get EHS implemented EHS it needed an Obama-era law and incentive plan* Many of the initiatives aiming to speed up access to healthcare and diagnosis are undermining the relationships across the journey of being a patient * Technology is rarely the solution because the problem is typically bureaucracy, culture, lack of incentives, and externalitiesLessons for you:* Beware complexity: Most of AI products being sold by major corps and consultancies are ones solving micro-problems and not designed to tackle complex problems* Worry about adoption: It doesn’t matter how brilliant your solution is, getting buy-in and adoption within enterprises will be the most pressing challenge* Think of problems as systems: JTBD and user stories have a tendency of over-simplifying problems and underrepresenting the range of factors, dependancies, and implications of a problem on the system as a whole* Ethnography is key: If you want to make a positive change to a problem space you need to leverage deep qualitative research techniques, like ethnography, to document and assess what matters and why* Monitor for unintended consequences: Even after dedicating lots of time to research and planning, we must be monitoring for unintended consequences that may create more work or more anxiety for those stakeholders within the system.Thanks for reading Design of AI: News & resources for product teams! This post is public so feel free to share it.Challenges building truly human-centred AI products and solutionsAI thought leaders love to push this message of getting to the future quickly. It creates this narrative that we’re all falling behind.But let’s slow down and recognize that there are countless of questions to be addressed before throwing everything out in favour or the shiny new system. This paper from Microsoft explored the many questions that users are posing about using AI agents. And these are very important questions that every team should be able to answer clearly to their users before deploying any solution.This poll from Google’s former Chief Decision Scientist highlights that the technical part of implementing AI is no longer the biggest barrier, understanding humans is. If the organizations polled —ones who have successfully implemented AI— are struggling to identify good opportunities and to convince people to use it, then imagine what struggles an everyday org will have.And also worth considering that AI adoption is still much lower than we’d expect given all the hype. The implementation of aI —especially across large orgs— may takes a decade or more because we’re fundamentally asking teams to change the way they work. Moreso, those in regulated industries need the permission to change how they operate before they can even consider implementing AI products.In the background many workers are using AI without their employers’ knowledge, leading to an endless range of potential risks.Thanks for reading Design of AI: News & resources for product teams! This post is public so feel free to share it.Mindset shifts to help implement AI In the podcast Spencer kept highlighting that we need to go into problem spaces with humility and without the expectation that problems are easy to solve. Other guests have suggests other types of mindset shifts:* Jess Holbrook stated we need to be specific when talk about AI: Too many projects are built off of expectations, not specifications of what AI should do and how* Kristie J. Fisher believes we need to measure time well spent using AI: The best solution to adoption problems is making sure that the AI product delivers value AND time well spent* Josh Clark advocated for embracing the weirdness of AI: The imperfectness of AI outputs should be viewed as a creative and innovative feature to help you explore new directions* Phillip Maggs challenges us to imagine new possibilities with AI: This is your time to spread your capabilities into areas you always wished were possible * Alexandra Holness expects that designers need to be less emotional precious with AI: This is a time of uncertainty and what worked before may not work in the future, so especially designers will need to go into problem spaces with additional humblenessMetalab is probably the premier design shop in North America. They’ve designed many of the most popular AI products in market today. Sara Vienna, their VP Design published a great manifesto about mindset shifts that’s worth a read. And she’ll be a guest on the podcast soon!And for those wanting more of a blueprint: Tertiary Education Quality and Standards Agency of Australia put together a guide that has lots of helpful detail into some of these. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

In our latest episode, Lisa Weaver-Lambert dispels the belief that is incapable of delivering impact in her book "The AI Value Playbook." She also lays out principles for succeeding in your implementation of AI:1. Your tech stack determines winners: Orgs that already were built to process and leverage data as part of core decision making are at a huge advantage. Especially those that are focused on leveraging insights to learn and iterate.2. Leadership and strategy matter: The vision, guiding principles, and culture matter. They will dictate the strategy or lack of a cohesive strategy.3. AI shouldn’t be added on top: AI should be viewed as the pathway ro removing layers, friction, and complexity.4. Getting from proof of concept to value is harder: AI reduces the barrier to creating proof of concepts while also layering in a lot more uncertainty about how to make it production-ready.5. Centralize AI strategy & decentralize implementation: Orgs should have a cohesive strategy owned by a centralized team. But the workflows and use cases defined by the teams that are seeking to gain specific value.Listen on Spotify | Listen on Apple | Watch on YoutubePlease rate the podcastIf you’ve listened to the podcast, please help us by giving us a rating. It helps us get in front of more people and know that what we’re publishing is delivering value.Rate us on Spotify | Rate us on Apple PodcastsAnd if you have comments, questions, or suggestions: info@designof.ai New report showing use of Anthropic (Claude) doubled, while OpenAI lost 1/3Menlo Ventures published their 2024 report: The State of Generative AI in the Enterprise. It shows the continued maturation of the AI market and clear use cases where the tech is being leveraged. Not surprising, task-level use cases that can be directly evaluated/audited are coming out on top. Also, the layers of AI stack are becoming more distinct with some products starting to create their own moats. As we move into 2025 expect the Data layer to split as more orgs realize that they need a semantic layer to structure and make sense of first-party data.Thanks for reading Design of AI: News & resources for product teams! This post is public so feel free to share it.The LLM market share data makes OpenAI look like the big loser. But I suggest throwing out the 2022 and 2023 data since adoption was so low and leveraging the tech for experimentation rather than impact. 2024 is the year when AI became the workhorse for the first time powering countless products. Nonetheless, it is compelling to see Anthropic and Claude shoot up. Their focus on UX seems to be paying dividends, that or OpenAI’s dilution of trust is.Of no surprise, prompt engineering is falling off a cliff. It was a bandaid approach for a tech that had no standards yet. For reference a business that built their product through prompts often had to rebuild all those prompts whenever a model was updated. Thanks for reading Design of AI: News & resources for product teams! This post is public so feel free to share it.AI use & impact assessment surveyPlease share your experiences and point of view in our year-end AI research study.Your lessons and opinions will shape a critically important assessment of how & if AI is positively impacting individuals and teams.Less than 5-minutes of your time will help us a lot.Perplexity is one-upping Google by introducing AI-powered shopping journeysPerplexity, the upstart GenAI search form is firing shots at Google by taking a refreshing look at shopping. Rather than focusing on someone searching for a product (e.g. Patio furniture), they are taking a very human-centred approach by focusing on what a user is trying to accomplish (e.g. renovate my outdoor living space). The platform then provides ideas, support, and instructions. Plus, recommends products to buy.While this is immensely helpful, it brings up the ever-present concern that AI will pick winners and losers for us. Where Google served up dozens or hundreds of results and encouraged us to make our own decisions, AI only shows a handful of options. This is the beginning of the platform as expert and it could change how we interact with the world in a huge way. It could lead to small merchants being shut out or even grow distrust of options that aren’t recommended by a platform.Alarming data showing that achieving AGI could destroy market wagesEconomics at the International Monetary Fund have modeled data that shows that if Sam Altman & crew succeed at bringing AGI to the world faster than expected, it could set into motion a total destruction of market wages (aka devalue everything).Their model also showed that on the expected timeline of AGI, wages will continue to rise as humans continue to do the thinking for the machines.Read the reportThanks for reading Design of AI: News & resources for product teams! Subscribe for free to receive new posts and support my work. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com

The last two years have been extremely stressful for anyone working in tech. There’s been a consistent sense that we all need to do more with less. That our jobs are on the line. And now AI is being touted as the cheat code that will unlock productivity and profit gains.In our latest podcast, Peter Merholz (add him on LinkedIn) doesn’t see AI helping much in the short-term because teams are too over-tasked to believe they have the time to try new models of working. He also believes that most organizations don’t have cultures and leadership that promote experimentation and reward learning. Listen on Spotify | Listen on Apple | Watch on YoutubeWhat makes matters worse is that simply “using AI” won’t get you the results you need. Simply using ChatGPT or Claude will not give you and your business a significant boost because data is at the heart of AI. The more of your first-party data that you train models on and the more that you craft agents around specific workflows, the closer you’ll get to what AI acolytes are selling. Accenture calls this AI maturity: Advancing from practice to performance. And this is where Peter Merholz believes that most orgs will be blocked. His experience working in mega-corps has found that most aren’t learning cultures. Introducing new tools, mental models, and ways of working aren’t well-received. AI use & impact assessment surveyPlease share your experiences and point of view in our year-end AI research study. Your lessons and opinions will shape a critically important assessment of how & if AI is positively impacting individuals and teams. Less than 5-minutes of your time will help us a lot.Valuable lessons 💡 Nearly half of workers are uncomfortable admitting to their manager that they used AI for common workplace tasks💡 Evaluations —or “Evals”— are the backbone for creating production-ready GenAI applications. 💡 Ten lessons that separate impactful training from mere AI showcases💡 Even teams actively working with AI are wrestling with fundamental knowledge structuring challenges. The tools are advancing faster than our practicesThanks for reading Design of AI: News & resources for product teams! This post is public so feel free to share it.Exciting AI jobs👉 USA | Anthropic | Strategic Product Management👉 USA | World Economic Forum | Head of Data and AI Innovation👉 USA | Google DeepMind | Group Product Manager, Generative AI Tools for Music Creators 👉 USA | Amazon Web Services | Generative AI Strategist, Generative AI Innovation Center👉 Australia | Canva | Creative Technologist (Gen AI)👉 Canada | Autodesk | AI Research 3D Dataset Creation & Annotation Manager 👉 Canada | Robinhood | Staff Product Designer, AI Investing👉 Canada | McAfee | Sr Product Manager, GenAI This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit designofai.substack.com