Each week, I'll be diving into some exciting AI tools I'm checking out, sharing a few experiments I'm working on, and offering some interesting insights about what I'm noticing in the world.

🔧 Three Tools I’m Testing

👨🏼‍🎨 Paint with Ember - An interesting take on creating AI images, you paint concepts onto a canvas, and the AI fills in an image based on your painting. It’s a weird and quirky tool that I find both fascinating and useless. Ha.

🖼️ Black Forest Labs Flux Playground - Another image creation and editing suite. They’re all starting to run together but are continuously improving, so there’s that.

📹HeyGen Avatar IV Video - Upload a single image of a person, provide a script, select a voice, and voilà. My end result was a bit creepy and not quite something I’d use. I’m going to try other images to see if they’re any better.

🧪 AI Experiment of The Week

This week’s experiment was a ton of fun and a small project I’ve wanted to accomplish for a while. I’m a massive Spotify user. If I’m not listening to a podcast (I use Overcast for podcasts), then I’m listening to Spotify. Lately, I’ve been yearning to create my playlists based on my preferences; however, I want to be introduced to artists and songs I don’t normally listen to. To make this seamless, I integrated a Spotify MCP server with my local Claude application, allowing it to connect and create playlists based on my prompts. 

The Spotify MCP server I decided to use was the one provided by Marcel Marais. It took a few minutes to follow the setup directions. The setup included creating a Spotify developer application. There was one important adjustment to the directions. Instead of using localhost in the callback URL, use 127.0.0.1. This means the callback URL is:

http://127.0.0.1:8888/callback

Once the MCP server was set up and connected to Claude, I created a couple of new playlists. Recently, I really enjoyed a few songs from an Apple Fitness+ workout and wanted to create an extended playlist based on these songs. So, I took a screenshot of the playlist from the workout (5 or so songs). I uploaded this to Claude and told it to create a Spotify playlist from these songs and similar ones.

Create a spotify playlist named "Apple Workout Country" with the songs in the image attached and find 30 similar songs.

A few minutes later, I had a new playlist on Spotify to enjoy! Not to mention, the music it picked matched the vibe of the original songs very well while also introducing new artists I don’t usually listen to.

As a bonus, I had ChatGPT create cover art for my new playlist, which is crushed. I noticed that Spotify has added AI cover generation to their mobile app as well. I’ll play with this later.

📰 Article of The Week

Robin Sloan's "brittle intelligence" explores our growing dependence on cloud-based AI systems through the lens of a devastating ice storm in Northern Michigan that left his parents without power and internet for weeks. He argues that the appealing sci-fi vision of local AI "backup generators" that keep devices working at reduced intelligence during outages isn't realistic given current technological trends, where computing and capability consistently move away from users toward distant data centers. The fundamental concern is about building increasingly fragile systems that depend entirely on maintaining a connection to remote supercomputers.

We're already living in a world of brittle intelligence, just as smartphones become nearly useless without cellular data. But the AI version will be more insidious than a simple on/off switch. Instead of binary failure, tiered intelligence systems are being created that degrade, similar to video quality, based on available bandwidth. Your smartwatch handles basic tasks, your phone manages moderate complexity, edge devices process real-time decisions, and the cloud provides deep reasoning. When connections fail, you don't lose everything at once. We experience gradual erosion of processing power. 

The real brittleness isn't technological; it's cognitive. We're rewiring human thinking around the assumption of AI assistance, creating what amounts to external brains distributed across multiple systems. Unlike previous tools that augmented our capabilities, AI is becoming integral to how we process information, make decisions, and solve problems. Children growing up today think differently about memory and creativity because they assume this assistance will always be there. When these systems fail or degrade, it's not just an inconvenience; it's cognitive disruption.

The question isn't whether we'll have backup intelligence but whether we'll maintain the mental muscles to function when our distributed intelligence network starts failing piece by piece. We're building a future where intelligence isn't just connected. We are dependent on the intelligence. Unlike Robin's parents, who adapted to life without modern conveniences during the ice storm, we may find ourselves genuinely unable to think effectively when our AI scaffolding crumbles. The brittleness lies not in the technology failing but in our growing inability to function without it.

🌎 Where the World is Going

Reflecting on my daily CEO routine, which includes endless email triage, fire drills, and constant context-switching between tasks. I can't help but wonder how fundamentally different this job will be in five to ten years. We're not just talking about better calendar apps or smarter email filters. We're approaching a reality where AI becomes less of a tool and more of a cognitive extension, transforming not just how we work but what work even means for leaders.

Imagine having an AI assistant who not only schedules meetings but also understands the nuanced context of your business well enough to draft board updates, negotiate vendor contracts, and flag strategic opportunities before competitors do. The tedious parts of my day include email triage that eats my mornings, note-taking during all meetings, and the constant task list juggling. This will all be handled by AI systems that learn my decision patterns and values. However, what excites me most is that this isn't about automation replacing human judgment. It's about amplification. When AI handles the operational noise, I can focus on what actually matters: vision, strategy, and the deeply human work of inspiring teams and building relationships.

The implications go beyond individual productivity. We're heading toward a world where the best CEOs won't be those who can personally juggle the most balls but those who can most effectively orchestrate AI systems while maintaining the human touch that no algorithm can replicate. The competitive advantage will shift from operational excellence to creative vision and emotional intelligence. The fire drills won't disappear. They'll level up. Instead of server outages and contract hiccups, we'll be navigating AI ethics decisions and managing the human side of AI-augmented teams.

What strikes me most is how this transformation will democratize leadership capabilities. Startups with AI-savvy founders will be able to operate with the sophistication of much larger companies. The playing field won't exactly level; resources still matter, but it will tilt in fascinating new ways. As I think about my kids heading off to school each morning, I wonder what skills they'll need in this AI-amplified future. The CEOs of 2035 won't just be managing companies; they'll be conducting symphonies of human and artificial intelligence, creating value in ways we're only beginning to imagine.

👨‍💻 About Me

🔗 Extra Links to Check Out

• Claude 4 System Prompt Information 📢

• DeepSeek Updates R1 And Drops it on Hugging Face 🤗

• NLWeb is Bringing Conversational AI to the Web 🌐