My performance review this year was the same as last year and the year before: “Joey, you’re doing okay, but you need to share more, be more vocal about your opinions. Put a stake in the ground!” That's Stephen by the way. You should bug him on Twitter.
Here’s the thing: I don’t know how to talk about the things people like to talk about in the way they like to talk about them. If I grew up in my son’s generation, I’m fairly sure someone would have slapped a learning-disorder label on me. I just want to get to the point, which means I often don’t bring a lot of colorful language or elaborate framing to the table.
On top of that, I seem to process things slowly. I’ve heard you, I understand you, I just need to mentally review what you’ve said. But because you want me to respond as fast as you did, I start talking anyway and fill the silence with fluff while my brain catches up. My system 2 runs faster than my system 1, which is unfortunate because system 2 is supposed to be the slow one. People close to me have learned to be patient. If you’re my sister and you’re stuck talking to me, I genuinely feel sorry for you.
“Okay Joey, those are all excuses. You could just write like everyone else.” Thanks for that. The problem is, I’m painfully shy, even when it’s just me and my computer. If I don’t know you and we’re in conversation, know that it’s taking everything in me to engage, and I’ll need at least a day to recover afterward. And there’s this Maurice Switzer quote I read in secondary school that has lived rent-free in my head and quietly sabotaged my adult life: “It is better to remain silent at the risk of being thought a fool, than to talk and remove all doubt of it.”
So yes: odds are you’ll get bored, annoyed, or exhausted.
But I want my next performance review to be fantastic. So prepare yourself: you’re about to be hit with a lot of sparse, incomplete, foolish thinking. You’re welcome.
On today’s episode, I finally built software successfully with a coding agent.
I’ve been trying to do this since I first touched ChatGPT in 2022. I’ve watched these tools go from truly terrible at writing simple code snippets to genuinely useful, at least if you’re not a code purist. And honestly, most of us who need these tools have no business being purists about anything.
In 2021, a lawyer came to me with a product idea. I told him to go read a product management book if he didn’t want to waste his money, and more importantly his life, building an MVP. Today, I’d just tell him to pay $20 for Claude and he’d have an MVP in a few days.
I’ve also spoken with founders who want to test ideas quickly without adding engineering costs or distracting their teams. Now they actually can do that without blowing up the budget or the team’s cadence.
These are the people who really need this. I don’t buy the narrative that engineers are going to lose their jobs. They’ll actually become more valuable and better at what they do.
If you’re worried about losing your job to AI agents, don’t be, unless you’re unwilling to become a better manager or architect. Then you should be worried, because that’s what the role will demand over the next few years.
Anyway, here’s what I built along with Claude. If you work in VC and you’re curious, fork it, improve it, and use it for your portfolio. If you are a DFS Lab portfolio founder or want to just send me updates for fun because you know I like them, come help me test.
See you soon.