This summer I had plans to build a research operating system for other consultants and small businesses. I wanted to rebuild my group travel planning platform. I wanted to find better ways to connect with all of you, beyond this hollow one-way conversation. None of it happened, and honestly, I’m happy about it. I spent the summer living outdoors, being a member of the real world instead of the digital one.
It’s easy for any of us to feel the guilt and shame of not being more productive, not reading more, not protecting our careers. That guilt isn’t human, and it isn’t normal. We’re not supposed to feel bad about declining productivity when it buys us time that’s better spent. I worry that ChatGPT, and especially Claude Code, has put a pressure on all of us that isn’t healthy. So what if you didn’t vibe code a new tool that you’ll trash in a couple of weeks? That was never real productivity, and it’s genuinely debatable now whether it helps your career at all. ETH Zurich tested 100 developers in a commercial-grade vibe coding environment and found the tool amplifies what you already know rather than handing you what you don’t. The people who reported the most daily AI use performed worse, not better.
The latest wave of models is what makes this urgent. With OpenAI’s GPT-6 Astra, which Greg Brockman introduced with “welcome to the AGI era,” we’re seeing the outline of AGI take shape: autonomous agents with recursive learning, capable of shipping products that adapt with no human in the loop. It’s easy to read that as a clean win for businesses and a loss for every worker who spent a career mastering their craft.
Whether you’re an AI bear or a bull, it lands in the same place: vanilla products, all screaming for intrepid thinking and unique appeal. If your ecommerce platform runs on the same agents as your competitor’s, the bar rises uniformly for both of you. The real gains will happen at the frontier, where new capabilities roll out in waves, or at the interaction layer, where small decisions make an enormous difference in the mind of a customer. Neither one comes from spending your evening vibe coding an app that reminds you to ask your kids what they did today, and neither comes from stretching your expertise past the work you actually love.
Truly transformative thinking happens at the margins of your real life. The more you experience the communities you want to serve, and feel the highs and lows of the moments that matter to them, the more you carry those fascinations back into your craft.
For example, my experience of losing a friend to treatable cancer because he looked too healthy forces me to imagine a world where biometrics are more than a vanity metric you brag about with friends.
S02E19: Upgrade from Vibe Coding to AI-Native Product Design
Metalab has spent 20 years designing products for Windsurf, Midjourney, Suno, The Atlantic, and Uber. AI made feature parity fast and cheap, so Myles Palmer, their Design Director, gets asked constantly what separates good work now. His answer isn’t “taste.” He thinks that word is a hollow way to dodge a much harder question.
➜ Why Metalab’s 20-year discipline, not any single tool, is what actually differentiates its AI-native design process.
➜ Myles Palmer on why “taste” is a hollow answer, and what actually differentiates once AI makes feature parity cheap.
➜ Brittany’s own vibe-coded travel app cost thousands in API calls and still lost to Google — the false confidence trap.
➜ What founders actually pay for later when they skip design early, and why it defends against a “sea of average.”
➜ What AI design looks like past the chat box — ambient transit, try-on shopping, and rethinking university sites.
➜ Digital twins, the 23andMe cautionary tale, and why Apple’s health-data model is where consent design is headed.
“I hate the idea of taste. The idea that taste is this differentiator drives me crazy because it’s like, well, everybody’s taste is different anyway.” — Myles Palmer
“Investing early in design defends against that sea of average.” — Myles Palmer
Listen now: Spotify | Apple Podcasts
S02E20: AI Shouldn’t Be Like Bolting a Spoiler on a Crummy Honda
Two years ago Josh Clark and Veronika Kindred previewed a book about a coming era of sentient digital experiences on this show. Sentient Design is out now, and the future it predicted arrived faster than most companies’ data, governance, or restraint could keep up with.
➜ Don’t wait for perfect data — legacy data, not ambition, is blocking three live Fortune 50 sentient design projects.
➜ Personalization is quietly becoming individualization — reading the moment, not building a permanent profile.
➜ Defensive design: teaching AI systems humility and users healthy skepticism, so neither is taken at face value.
➜ Agents need their own front door — if you don’t design one, somebody finds the back door instead.
➜ AI is a reckoning, not a feature: the real opportunity is removing interface entirely, not adding more of it.
➜ Google’s AI Overviews are the biggest live sentient design experiment, and businesses never agreed to it.
“Putting a chat sidebar on an existing thing, it’s like bolting a spoiler onto a crummy Honda.” — Josh Clark
“With LLMs, it’s a really siloing technology.” — Veronika Kindred
Listen now: Spotify | Apple Podcasts
Get the book: Sentient Design — Josh and Veronika set up a coupon for Product Impact listeners: SENTIENT-PI for 20% off, expires September 12.
Four Lessons on Actually Making Products Better
Almost none of the industry conversation right now is about how to make products better. We’re consumed with making teams faster and protecting our jobs, and the lessons that actually matter have stopped getting shared. Here are four I keep coming back to, and all four are about building the kind of unique appeal that no shared agent can replicate for your competitor.
1. You don’t sell AI. You don’t sell bolted-on sparkle-dust buttons, or promises you won’t deliver until next year. Your customers need to noticeably feel your latest capability. In practice that means solving the measurement problem that plagues most AI-powered products: if you can’t measure the outcome, not the output, you’re not proving anything valid.
2. What your competitors are doing matters more than ever. The cost and effort of product development is dropping every month, and AI is pushing businesses to take more moonshots than ever. You need to know about them and track how they land. Agents can help you stay current and follow the meta conversation around each release. More importantly, add competitive testing against at least two competitors: benchmark KPIs, attribute the lifts and drops, and run features head-to-head on a regular schedule.
3. There are failure patterns you don’t know about. Every business has shipped some AI-powered capability. As much as the models keep improving and output quality keeps rising, failure patterns are always happening that you have to accept you can’t see. Product teams get obsessed with golden queries and requirements, but there’s equal and often greater value to the brand in sleuthing the cause of poor and very poor experiences, well beyond basic telemetry and polls. In the extreme cases, the LLM does something wildly unexpected because training and guardrails are conflicting, opening blind spots that are dramatically bad.
4. The biggest blind spot is still conversion. For all of the above, the single biggest blind spot has always been understanding why prospective customers convert or don’t. Take a real leap forward by documenting the customer pre- and post-conversion, and map listening activities to each stage so you have robust telemetry on what actually moves conversion and LTV up or down.
Worth Your Time
Wearables Will Save Lives, But We Need a New Relationship With Biometrics First — the latest from Product Impact, and the hardest thing I’ve written. One of my best friends died of throat cancer at 44, told for months he was too young and too healthy for it to be serious. Nobody gets sick suddenly; they get diagnosed suddenly, and the suddenness is an artifact of measurement. The piece makes an argument that belongs in lesson one: step counts and resting heart rate are the health equivalent of pageviews and daily active users. Cheap to collect, visibly responsive, built to reassure — which is exactly what makes them useless as warnings. The industry optimized for the numbers that keep people wearing the device instead of the numbers that would tell them something they need to know.
The ResearchOps Review — a biweekly feature publication on how the research practice is actually evolving. ResearchOps went from an obscure Silicon Valley specialty to a global profession in about ten years, and AI has expanded the remit again while research democratization pushes operational complexity up. Read it as a proof of concept: what’s happening to research ops is what’s coming for every product team.
If you can’t measure the outcome, you can’t prove the capability. PH1 works with product teams on exactly that problem — competitive benchmarking, failure-pattern research, and conversion telemetry that survives contact with a real customer journey. Clients have included Spotify, Microsoft, Mozilla, and Search Atlas. Email me: arpy@ph1.ca. Contact me for:
Mastering the roadmap: Alignment, market intelligence, prioritization
Winning the last mile: Evals, testing, GTM iteration
AI Value Acceleration is looking for 5 startups to participate in a study on how to improve AI adoption. Brittany is CEO of AI Value Acceleration and is documenting the effectiveness of the latest methods for improving and scaling adoption. Contact her to discuss how she can help your business: brittany@aivalueacceleration.com.
Thank you for your support. As always feel free to send any questions or comments. They’re very helpful for us to know what topics and problems matter most.
Browse all two seasons of episodes at productimpactpod.com.