Building in the Age of Agents — Issue #1
I've been posting about AI on LinkedIn for a few weeks now. Articles about multi-agent frameworks, self-healing documentation, semantic memory layers. Real work. Real projects. And I was proud of them — still am, actually.
But last week I scrolled back through everything I'd written, and something hit me.
I'd missed the mark.
Not because the content was wrong. It wasn't. I've built these things, they work, and if you're deep enough into the AI/LLM world to know what a "semantic memory layer" is, those posts might genuinely help you. But that's the problem — the people who could already understand what I was writing didn't really need my help. And the people who did need it couldn't get past the second paragraph.
I'd been writing for the wrong audience. Or maybe the right audience in the wrong language. Either way, I needed to stop and think about what I was actually trying to do here.
The Rabbit Hole
Let me back up.
I'm a product manager. Have been for years. I fell into AI the way most people do — curiosity, then obsession, then looking up from my screen at 2 AM wondering where the evening went. I watched hundreds of videos. Downloaded and evaluated more GitHub repositories than I can count. Started building tools to make my own work easier, then tools to help my job search, then tools that connected to other tools, and suddenly I had this ecosystem of AI-powered systems doing things I couldn't have imagined a year ago.
Multi-agent frameworks. Autonomous workflows. Memory systems with safety gates. The whole deal.
And I loved it. Still do. This stuff is genuinely fun for me — not in a "look how smart I am" way, but in a "I can't believe this is possible and I get to play with it" way. The rabbit hole is deep, and I went all the way down.
The problem is, when I came back up for air and started telling people about what I'd found down there, most of them looked at me like I was speaking a different language.
Because I was.
Selling Fridges to Eskimos
Here's the moment I knew I had it wrong.
I was talking to a friend — someone smart, capable, the kind of person who'd benefit enormously from even the simplest AI tools — about my latest project. I was excited. I was explaining the architecture, the integrations, the possibilities. And I watched their face go from polite interest to glazed confusion to something I wasn't expecting: anxiety.
Not excitement. Not curiosity. Anxiety.
"I could never do that," they said. And they meant it.
That conversation kept happening. With different friends, different people, different versions of the same wall going up. And it wasn't just confusion — it was real fear. The kind fed by every headline about AI replacing jobs, about environmental costs, about society unraveling. I was showing up with enthusiasm and a demo, and they were hearing "the world is changing and you're already behind."
I was selling fridges to Eskimos. And I was doing it loudly, to people who didn't want a fridge, in a language they didn't speak, while they were worried about the weather.
I had some heated conversations with close friends about this. The kind where you realize you're not just failing to communicate — you're actually doing harm. Driving people further away from something that could help them. Straining relationships with your own enthusiasm. Being so excited about what you'd built that you forgot to ask whether anyone wanted it.
That's a rough realization for a product manager, by the way. We're supposed to be the ones who ask "what does the user need?" first. I'd skipped my own step one.
The Mystic and the Schizophrenic
There's a saying I keep coming back to: the difference between a schizophrenic and a mystic is all in who you tell.
I'd been telling the wrong people the wrong things in the wrong way. Not because I was wrong about what AI can do — I wasn't. But because the gap between "this is amazing" and "this is useful to you" is enormous, and I'd been leaping over it like it wasn't there.
So I changed my approach. Stopped leading with the tools. Stopped showing off the architecture. Stopped trying to convince anyone that they should care about what I'd built.
Instead, I started asking one question:
What's a problem you're trying to solve?
That's it. Not "have you tried AI?" Not "let me show you this agent framework." Just — what's bugging you? What takes too long? What's tedious? What do you wish someone else would handle?
And then — only then — showing how a tool might help with that specific thing.
Don't Look at the Top of the Mountain
Here's what I tell people now, and it's the core of everything this newsletter is going to be about:
Forget about the tools. Forget about the possibilities. Forget about the headlines and the hype and the thousand things AI can theoretically do. All of that is noise when you're standing at the trailhead.
Focus on the problem in front of you. Then use a tool to solve it. Start there.
Don't look at the top of the mountain at the start of the hike and think about how hard it'll be to reach the summit — because in AI, that damn mountain keeps growing. Every week there's a new peak. Every day someone posts about a new trail. If you spend your time staring up, you'll never take the first step.
Instead, assess the step right in front of you. Take it. Then take the next one. And enjoy the beauty of the journey — because there is beauty in it. In that first moment when a tool saves you 20 minutes. In the realization that you just automated something that used to eat your afternoon. In the quiet confidence of knowing you're not behind — you're just at a different point on the trail.
The summit isn't the point. The walking is.
What This Newsletter Is Really About
So here's what I'm doing with this space.
I'm not going to talk about whether my APIs and cron jobs are configured correctly. I'm not going to post architecture diagrams or drop jargon to signal that I belong in the room. I've done that. It served its purpose. But it's not what matters most to me.
What matters to me — what I keep coming back to when I think about why I got into this — is helping people. That's what I loved about being a PM. Not the roadmaps and the sprint planning (OK, maybe the sprint planning a little). The part where you figure out what someone needs and then build the thing that gives it to them. The part where their life gets a little easier and you had something to do with it.
That's what I want this to be.
I want to take everything I've learned — the rabbit hole, the tools, the mistakes, the heated conversations, the projects that worked and the ones that didn't — and give it away. Not wrapped in complexity. Not guarded by jargon. Just: here's what I tried, here's what happened, here's what you might want to try if you're dealing with something similar.
And underneath all of that, I want to talk about something bigger: how we use these tools for good. How we, as the people actually using AI every day, demand that companies build safe and meaningful tools — not because regulators tell them to, but because we understand what these tools can do and we insist they do it responsibly. The race to superiority and money doesn't have to trample what makes us good and human. Helping each other. Making existence a little bit easier. That's not naive — it's the whole point.
What's Coming
Over the next few posts, I'm going to walk you through my journey. The real one. A product manager who fell into AI, built a bunch of things to survive a job search, broke some stuff spectacularly, learned hard lessons, and came out the other side with tools that actually work and opinions about what matters.
No jargon gatekeeping. No "you should already know this." Just the messy, honest, occasionally funny story of figuring it out one step at a time.
If you're standing at the trailhead wondering where to start — welcome. You're in the right place. We're walking this one together.
— Corey
See what I've been up to: coreyscherrer.com
See what fun I've been up to: coreyiscorey.com
This article was written with AI assistance (Claude) as part of an ongoing experiment in using AI tools for content creation. The ideas, experiences, and opinions are mine — the AI helped me get them out of my head and onto the page without staring at a blinking cursor for three hours.