How Far Can One Developer Go?

Can I build enterprise-grade software—to the level of expectation I developed at Mega-Corp—by myself with Claude Code?

That's the question I'm trying to answer. Siren is my attempt to find out.

---

The Surplus Value

When I go to work, they keep the surplus value. I write code, they capture most of what it's worth. That's the deal.

I want to keep it. So the question becomes: how do I produce as much value as possible on my own? And can I actually hit enterprise-grade quality doing it?

---

I Know What Good Looks Like

I've been doing this for 15 years. I started in support and learned to debug—not just fix things, but understand what's happening versus what should be happening. I went on-site for go-lives and heard users frustrated with what we built. I found ways to make their jobs not suck.

I moved to dev at a product-focused company and learned what good looks like across the whole stack—not just writing code, but testing it, deploying it, observing it, keeping it running. Unit tests, integration tests, performance tests. CI/CD pipelines. SLOs, traces, metrics, logs. Security scanning. All of it.

I know how enterprise software is built, tested, secured, observed, and maintained. Claude Code makes me faster at all of it. Not because it knows what I don't—but because it's fast at the things I already know how to do. I know I need tests, so Claude writes tests. I know I need security scanning, so Claude helps me configure OWASP. I know I need observability, so Claude helps me instrument.

The hard-won knowledge isn't optional. It's the whole point.

---

Stay in My Lane

AI is a multiplier. 10x of what I know is 10x. 10x of zero is still zero.

So I'm using MVC because I know MVC. Vanilla JS because I can debug it. No mobile app because that would kill my velocity. Every decision runs through the same filter: does this keep me where the multiplier works?

I lean on my network for the rest. I can set up Cloudflare, configure the WAF, get the certs routed. But at some point I'm going to call my friend Cory and ask "am I going to get more availability if I switch to Route 53?" Knowing when to ask is part of staying in my lane.

---

The BS Detector

A lot of this is just being perceptive.

I don't care about implementation. I care about outcomes. When I push back on a coworker's solution, it's not personal—it's because there's a part left unexplained. And when people accept my ideas with no pushback, I get frustrated. My first pass at this problem can't be the best idea we have.

I used to sit next to a complete genius at work. He was right almost all the time. But I still pushed him to explain why. I didn't just accept "well, you're a genius and I failed out of college, so let's do your idea." If you don't understand, ask.

The value is in having your ideas challenged. When working with Claude Code, I need to detect Claude's BS just as much as my own.

Here's a real example. I was trying to figure out a database solution for Siren. I said to Claude, "I'm on Azure, what's the best managed database option?" and naturally got Azure solutions. Cosmos DB came up. Made sense.

A few days later I asked differently—less "what's the Azure answer" and more "what's the right answer." That's when CockroachDB came up.

Same skill you need with humans: be clear about what you want, don't steer too hard too early, and if you're not getting what you need, ask differently.

---

The Enterprise Is in the Way

I don't understand why enterprises are so excited about AI coding. None of what slows us down is the code.

Great, Copilot gave me a PowerShell script and I tested it in five minutes. I still can't deploy without three approvals. I still can't get the firewall change until Thursday when they batch those requests. The security office still needs two weeks to review. The architecture board meets monthly.

AI automated the fun part and left you with the bureaucracy. That's a tough trade.

Half the reason I'm doing Siren is because enterprises don't capture the speed. All the friction is still there—AI just means you hit it faster. I'll accept not writing the code for the joy of seeing it shipped.

With Siren, I don't wait. I don't have to know it's secure because someone signed off—I have to actually know it's secure. I don't file a ticket for a firewall change—I manage the Azure network myself. I gather evidence, implement, test, and decide. It goes to production when I say it's ready.

I've spent a lot of time at work wishing some senior engineer would come tell me what to do. But at some point you realize people are looking to you. You have the context. You need to make the call.

With Siren, that's perfectly clear: I own it.

---

Conclusion

Stay where your strengths are. Don't marvel at your own ideas. Lean on the people who know more than you. Verify everything. Make the call. Do it fast. And do all of that without breaking production.

Ok. I can do that.