Sara Zalowitz

You've likely seen the headlines of executives issuing AI mandates for internal teams, some of which hasn't landed well internally (or externally). Prior to these headlines making a splash we opted to take a slightly different approach. A set of expectations for the entire Design org: 1. You’ll experiment with AI and share what you’re learning. 2. You’ll become proficient at prompting. 3. You’ll stay alert to opportunities for which AI can improve the products, services, and workflows you’re responsible for. 4. You’ll maintain a very high bar for design quality, regardless of whether humans or bots do the work. 5. You’ll never lose touch with the ability to craft with your hands, mind, heart, and gut. How are we making these actionable so they're not just frothy maxims? One of the ways we're doing this is through a monthly series called "AI Playdates" where volunteers or delegates present on design-related AI topics to the entire org. Today was the second instance in our series, with four presenters sharing their experimentation with Lovable, Bolt, Replit, v0, and ChatGPT for generating UI concepts and complementing their design workflows. Through Playdates, we're directly chipping away at 1 and 2 above, and indirectly at 4 and 5. Expectations (or guiding principles) such as these aren't likely to make headlines. In fact this post isn't likely to garner much attention either. Yes, sometimes we need to act boldly as leaders, and yes even with mandates. But there is also wisdom, sometimes greater wisdom, in acting with more thoughtful judgment and more flexible guidelines. Slow and steady. Measured and principled. These don't make headlines as much, but gosh they sure can be an effective way to lead. #ai #design #leadership

311

31 Comments

Bruce Shields from Carhartt joined us on the most recent episode of The Frictionless Experience podcast, and he gave Nick Paladino and me a masterclass in how data and design actually come together. Here are some of the biggest takeaways: - UX maturity is a spectrum. Carhartt started with “UX = design.” Today, it’s about testing, personalization, and optimization at scale. - Data before design. Wireframes now get modeled against revenue goals before a single creative decision is made. - Confirmation bias is real. Early testing was used to prove assumptions. The shift came when Carhartt reframed “failed” tests as wins that brought them closer to the truth. - Benchmarks change the game. Two years of homepage interaction data turned into predictive models that can forecast component performance. - The metric to watch. Conversion rate between category pages and product detail pages is one of the most reliable predictors of revenue. - Clicks aren’t the enemy. An extra click that gives users clarity is better than a shorter, confusing journey. - Segmentation matters. New customers and loyal repeat buyers need completely different experiences. - Culture counts. A low-ego, high-curiosity environment made it possible to embed testing and data into everyday decisions. Every one of these points could have been its own episode. Big thanks to Bruce for bringing so much depth (and for making me jealous of Carhartt’s framed jackets).

8

1 Comment

I wish more digital product orgs realized that design is innately expansive. Design can happen in five minutes. Five weeks. Five years. It will fill all the time it is given. It begs a question that teams need to wrestle with. What is the right size effort? And what is the best use of that time?

9