It's kinda quiet around here... a bit too quiet
I spend a big chunk of my day reading posts, scanning social media, and watching YouTube videos about the latest stuff happening in the world of AI. There's a lot to consume, and it regularly feels overwhelming. I think have a solution to that problem, which means I have a favor to ask of you (again), but I'll get to that in a second.
Until recently. Over the last month or so I have noticed the typical "talking heads" on YouTube have stopped talking so much. The OpenClaw hype has ebbed and it "feels" as if people have stopped sprinting, taken a deep breath, and collectively raised their heads and asked "what just happened?"
Maybe I got used to the ridiculous pace of announcements, stacked one on top of the other on an hourly basis. I really don't know, but it's making something very clear to me: I kind of miss talking to real humans about this stuff.
Whenever I push a video or write a note like this one, I get a tumble of comments that spark some fun discussions. People ask me "have you tried this" or "I've been using that and I love it", and it makes me happy because it felt like we're discovering something new together.
It kind of fired me up and reminded me why I love this industry! Something new happens and we get to explore what's possible. This includes the web itself, the Dot Com bubble, Web 2.0, NoSQL, Internet of Things, Blockchain, REST, etc. Some of these things lasted, some didn't.
Exploration is fun but the finding out isn't the best part, it's the f'ing around with your online friends.
Which gave me an idea.
Let's Do That Again
I wrapped up my first cohort in early April and I still talk to the group from time to time on Discord. I kept it up and running as a few of the members wanted to keep chatting, as did I. Just today someone posted an interesting video on LLM error rates and their growth over time.
It made me want to fire up another cohort, and I started planning the dates and outlining the themes. It got me fired up thinking through the topics! Things like:
- Using frontier models to plan, local models to execute.
- Scaling orchestration patterns, from local logging to GitHub project tracking.
- Pitting Claude Code vs. OpenCode vs. Pi.dev, seeing what we can do without a monster system prompt.
The fun part about these cohorts is that everyone's experience is so different. Their backgrounds are different, nationality and culture, and the results they get with their AI tools. Those differences are so much fun to dive into, and can really broaden your mind when it comes to your day to day. It also helps to have social interactions in what can feel like a very cold, harsh industry.
This is what one of the cohort members told me (shared here with permission):
There is no doubt that Rob's AI Cohort has changed the way I work in general and specifically the way I work with Claude Code
Now before you think I'm trying to pitch you on my cohort, I swear to you I'm not. I could use your advice, however.
Like That, But Bigger
The AI thing is moving quickly, and I'm doing what I do, which is to scout out what's happening, play with the tools, break them down, put them back together, break them again, and find the useful patterns and practices. This is what I've done my entire career.
I'm not some kind of guru or thought leader. I have a talent for exploring and sharing what I find, and to me, that works really well in a group setting.
So, to the point: I'm considering adjusting what I do, moving into more of a "community" format. It's not really a pivot or rebrand - nothing so dramatic - more of a shift away from traditional online learning structure to something a lot more loose and freeform. Live events, group hours, video posts, regular posts - all based on things I'm learning and readind.
And don't worry: if you have a sub to my current site I'll make sure you're taken care of.
The focus would be quite narrow: how to create high-quality code with AI.
There are some interesting platforms out there that let you do this, and I have a few choices:
- Circle, the obvious choice. Kind of complex to use, gets the job done.
- Skool, stripped down and easy to use, but smells of "hustle-bro" a bit.
- Build my own because I can.
I would rather not build my own experience, but ... in this day and age you can do precisely what you want in a weekend or two. Which is both a good and a bad thing.
The Favor
If you had just a minute, I would really love to read your opinion on this.
Would you be interested in a community (free and paid) that is devoted to "software craftsmanship" in the age of AI? The topic absolutely fascinates me, and there is so much I want to do here. I have loads of experience on this front, and I have some good friends who would love to contribute, I'm sure.
I'm finding that the current setup I have (Ghost) simply isn't geared for this kind of thing. Building my own would give me a lot of fodder to work with, which would be fun, but it would also be a bit of a distraction.
Anyway: would love to hear from you! And yes, as always, I wrote this 100% myself.
Rob
It's not hype, it's happening
Get a premium membership and enjoy timely, level-headed videos on the emerging AI industry.