AI Roundup for Product People - June 29, 2025
Trust, taste, and the future of building with AI
It’s easy to focus on what AI takes away. But it’s also creating new responsibilities—and new jobs.
This week we look at how product designers are starting to ship, what new roles are forming around trust and AI integration, and how to sharpen your own AI skills and intuition with simple, hands-on habits.
🎧 Dive Club: Emmet Connolly - Transitioning into the next era of design
This is a must listen if you're designer, especially if you work at an Enterprise or Scale Up. Emmet Connolly, VP of Design at Intercom, talks about the transformation happening within his design organization to adapt and take advantage of AI.
A couple of highlights.
This quarter, designers are challenged to ship code to production. It doesn’t need to be fancy or a whole new feature… small text update counts.
It’s exciting to hear about companies pushing the lines of where design ends and engineering begins.
Imagine all the effort that we put into creating pixel perfect mockups and prototypes instead redirected to improving customer experience!
Later in the episode (around 46:30), he talks about designing an AI-infused reporting tool.
The process didn't start in the traditional way of discovery, JTBD mapping, wireframes, and mockups.
It started by working directly with customers, processing their raw data, iterating on different approaches to applied AI, and figuring out what’s even technically possible and delivers real value. Only after that do they work on an interface.
This inversion of the design process—starting with the data and capabilities, then layering on UX—might become the norm for AI-native products.
🎧 The Artificial Intelligence Podcast: The New Jobs AI Will Create
This episode digs into a recent NYT article, "A.I. Might Take Your Job. Here Are 22 New Ones It Could Give You."
The main idea: yes, jobs will disappear—but new roles will emerge around three core capabilities.
Trust. AI can write the report. But someone needs to stand behind it. In regulated industries or high-stakes decisions, someone has to review, sign off, and be accountable for the output. This includes new roles for auditing, interpreting, and validating what AI produces.
Integration. These are the people who know which models to use and how to apply them. They may not be building the models themselves—but they know how to route AI into the right parts of the business. These will be internal AI consultants, problem solvers, systems integrators, and automation specialists. Adam D’Angelo at Quora recently posted a role like this that reports directly to him.
Taste. In a world of infinite AI-generated everything, selecting what’s good becomes its own skill. This is where designers, marketers, and creatives come in—not just creating, but curating. Picking the best outputs, shaping them to fit a brand, and defining what “good” even looks like.
This is what it says specifically about product design:
There are some titles we already have, like product designer, that will simply grow to encompass a whole lot more. In the future, product designers will have a much greater ability to own products, from top to bottom. The role will be not just about the big picture but also about all the choices that bring that big picture to life.
I found myself nodding along because, frankly, this newsletter is a reflection of those three themes.
Curation and taste? I go through a ton of material each week, but only include a handful of links.
Integration? I built a workflow with SuperWhisper and ChatGPT that turns raw voice notes into structured summaries. It lets me do this work while still juggling everything else.
Trust? My name's on it. AI helps, but I edit its output and own the content, style, and perspective. That accountability matters.
It’s not a utopian take—but it is a grounded look at how our roles might evolve. Definitely worth a listen and a read.
🎧 AI Daily Brief: 3 Ways To Get Better at AI Right Now
AI Daily Brief is one of my favorite podcasts. Episodes are short, sharp, and useful. The first half is news; the second half goes deep on a topic.
This one lays out three ways to immediately improve your AI game:
For yourself to use OpenAI’s o3 model like a colleague. Draft memos. Workshop slides. Play through scenarios. The idea is to treat it as a strategic collaborator—not just a task monkey. The more you use it the more intuition you’ll build on what it’s good and bad at.
Vibe code something. Use tools like Lovable, Bolt, or Replit to bring an idea to life. Could be a feature you’ve been noodling on, a side project, or a weird little clone of a game you loved as a kid. Doesn’t matter what—just build.
Set up an agentic workflow. Zapier and n8n are all good places to start. Even if you don’t need one now, the process of wiring together flows for research, outreach, or content generation will teach you how this new class of tools actually works.
If you’re already experimenting in this space, none of this will be new.
Personally, I’ve known about agentic tools like n8n for a while, but haven’t had a solid use case. Hearing it called out here made me want to revisit it.
This is a great episode to send to friends or coworkers who want to level up but don’t know where to start. Simple, actionable advice.
📺 Introducing Gemini CLI
Google just launched Gemini CLI, a retro-styled command-line tool that lets you access Gemini Pro 2.5 for free.
Free. As in: 60 requests per minute and 1000 per day. That’s a ton of compute for zero cost.
Gemini CLI supports agentic workflows. You can set it off on tasks that require reasoning, iteration, and code generation—without racking up a huge bill.
Unlike other tools like Cloud Code or OpenAI’s paid APIs, the pricing makes this viable for high-frequency usage. Perfect for developers (and maybe designers?) who want to experiment with more autonomy and less cost.
I gave it a go on an existing prototype I was working on. In light testing, the results weren’t better than what I get using Claude 4 Sonnet on max mode inside Cursor.
But that shouldn’t stop anyone from trying this out. It’s free, and could be especially useful for different kinds of projects or experimentation.
I'm still curious how long Google will keep this pricing strategy in place. Strategic move to build loyalty?