AI in the Organization

Design leaders have a narrow window to shape AI's impact on their teams

Andy Budd argues that design leaders have roughly six months to proactively define how AI transforms their organizations before executives do it for them. He urges design leaders to present a clear vision for AI-enabled design teams, invest in design systems, and shift their hiring focus toward designers who can work effectively across the design-engineering boundary. A key perspective: AI will reduce the need for production-level designers but increase demand for systems-thinking designers who can orchestrate hybrid human-AI workflows.

Design leadership in the age of AI: Seize the narrative before it's too late
⚡ Quick Read (4 minutes)

AI and Product Development

When to choose vibe coding vs. low-code platforms

Sebastian Mertens contrasts the two main approaches that make software development accessible to non-programmers:

• Vibe coding: prompt-driven development where you describe features in natural language and AI generates the code.

• Low-code development: platforms with drag-and-drop interfaces and pre-built components. Examples include Bubble and WeWeb, as well as Salesforce and ServiceNow in the enterprise.

His key takeaways:

• Vibe coding is the fastest start you can get.

• With vibe coding, every extra slice of complexity costs exponentially more prompts (and patience/time).

• Low-code charges a “learning tax” up-front, but repays it every time you scale.

His post includes a visualization that captures the current trade-offs, while acknowledging that vibe coding tools will undoubtedly support higher complexity over time.

Vibe-coding vs low-code
⚡ Quick Read (2 minutes)

Using AI

Knowing which model to use can make all the difference

Ethan Mollick on choosing the right AI model:

"When you watch people use AI, you see how absolutely confusing chatbots are, not just the obvious problems (like o3 being much better than 4o), but also how ill-explained so many features are. For most people, just realizing that you need to switch models to get more serious work done is a major revelation."

To that end, he's put together an excellent guide for selecting the right model for the job at hand, which is especially important for professional-grade outputs. His approach includes a three-part high-level framework:

• Good for chat (fast but not as smart) Ex: Claude 4 Sonnet

• Good for work (smarter, lower errors, but slower) Ex: Claude 4 Opus

• Good for hard problems (very slow) Ex: Claude 4 Opus Extended Thinking

A key takeaway: If you're using AI for a work-related purpose, avoid the mistake of relying solely on the default model (which is often the "Good for Chat" model that's best for casual exploration).

His entire post is a must-read for understanding the current models and how to utilize them effectively.

Using AI right now: A quick guide
☕ Medium Read (10 minutes)

AI and Product Development

In the future, designers will own the entire frontend

Emmet Connolly, VP of Product Design at Intercom, argues that designers will soon be responsible for both designing and building the entire frontend. His prediction is based on two converging trends he's seeing at companies like Intercom: designers are already "vibe coding" throwaway prototypes using tools like Lovable, and they're making small code changes directly in production using tools like Cursor.

"These two trends—starting to generate the first build of the UI from one side, and nibbling away at quality-of-life changes from the other—are converging at the same place: designers will be responsible for the end-to-end design and build of the user interface."

A key element in enabling this shift is the ability of vibe coding tools to ingest design systems and build with actual product components.

LinkedIn Post
⚡ Quick Read (2 minutes)

Bonus content: Hear Emmet Connolly on this week's Dive Club podcast talk about transitioning into the next era of design.

AI and Product Development

The CTO + Engineer method for designer-friendly vibe coding

There is no doubt that designers will need to become more comfortable solving problems in code. The best way to gain experience with coding is to tackle a real-world need: specifically, solving a problem that matters to you. This advice serves as a preface to a recent video by designer and content creator Ridd, who shares his two-part approach to vibe coding —a method focused on both helping you learn and generate working code.

In Ridd's approach, he utilizes two tools: one as a CTO (he's using ChatGPT) and another as Engineer (for example, Cursor). Key to his approach is providing initial context to your "CTO" and then having it ask you clarifying questions about your project until it fully understands the problem you're solving. With this full understanding and context, your CTO can create a plan to build the code in phases. Then, you use your Engineer (Cursor or similar tool) to actually implement the code according to your CTO's plan.

He demonstrates the approach with a real-world project he's working on for his dad, who is an insurance adjuster.

Ridd's playbook for vibe coding
Watch time: 9 minutes

That’s it for this week.

Thanks for reading, and see you next Wednesday with more curated AI/UX news and insights. 👋

All the best, Heidi